When I sat down to edit a slideshow for this summer’s Roseville Thunder team, I decided to give iMovie another try. To make a long story short, I’m thoroughly impressed, and I think that in almost all situations iMovie is finally better suited for this task than it’s more professionally-skewed brethren! Overall, the result is on par with or exceeding efforts which in the past had taken me several days (of working time) to complete in Final Cut Pro; this took me less than a single day of work (about 5 hours, all told).

Now, though, the blow-by-blow.

iMovie Settings

First, of course, I’m using iMovie 2009, a component of iLife 2009. IMovie took a major break from its previous incarnations in the ’08 version, seeing a rewrite from the bottom-up to make it easier for a novice to use. One of the drawbacks for those of us who’ve used video editing software before is that it heavily obfuscates the standard “timeline” approach we’d been used to. No longer does a project consist of a timeline stretching from left to right, with multiple audio tracks and multiple video tracks. Now, it’s laid out like a word processor (the timeline wrapping from line to line) and transitions between tracks are separate entities rather than overlaps on the timeline.

iMovie '09 interface

It can take some getting used to, but the layout makes sense to me. Perhaps I’m an easy sell, given that I’m not in an editing bay 40 hours a week, but the wrapping layout seems like a minor change and a major usability improvement. The most glaring ommission in iMovie ’08 had been the ability to fine-tune those transitions, performing rolling edits (so both tracks stay in the same spots, but one lengthens at the end and the other shortens at the start to move the transition) and audio drop-ins (to allow the audio of one scene to precede the video by a second or so, making a much more natural cut than a straight cut of audio and video at once).

IMovie ’09 brought those two features back in its precision editing mode. It also added the ability to time pictures in the timeline down to the frame level, which is critical for our purposes.

So, the first thing I needed to do in iMovie was to turn on frame count in durations. In the iMovie “General Preferences”, this is a checkbox labeled “Display time as HH:MM:SS:Frames”. I also clicked the “Show Advanced Tools” checkbox, which among other things exposes chapter markers in the video (which I find useful in slideshows).

iMovie '09 Settings

Importing Photos

For this slideshow I knew up front that my overall organization of pictures was going to be around type of activity. So, before I did anything, I took the pictures I’d set aside for the slideshow and tagged each with one activity type: batting, fielding, catching an out, running bases, pitching, coaching, and “other”. In Aperture, I filtered to each of these in turn and set my “Order” custom field to be based on the general order I wanted in the slideshow. Since I had a lot of both batting and general fielding, I split each of those up. I sorted the pictures in Aperture by Order (the secondary sort was by image timestamp, so that kept things organized enough).

I dropped all the photos from Aperture directly into iMovie’s Project panel. I didn’t use the built-in “Media Library” tool because I find it way too hard to deal with my media library in that little quadrant of the screen, can’t sort on anything, and generally have no problem dragging from one application to another (it still mystifies me why Apple has put so much effort and emphasis on this rather useless widget!)

I selected the first picture, clicked the “gear”, then selected “Clip Adjustments” (shortcut: select the picture then hit the ‘i’ key). I typed in “1:00″ for the duration (remember: this is Seconds:Frames; if you just put in “1″ then you’d just have one-frame images!), and checked “Applies to All Stills”. This is only temporary, to get a good feel for the timing of the music.

Next, I dropped the music in from iTunes. The particular song I’d chosen I’d found on Jamendo and is CC licensed to allow for derivative works. It demands share-alike, and since I have no problem releasing this video under the same license, that’s fine with me. My main criteria for the song (besides a derivatives-friendly CC license) were (1) a fast pace, (2) no offensive lyrics, and (3) listenable for the several minutes the slideshow would take. MzW’s “Blind” fit the bill rather nicely, in my opinion, so that was that.

Once the song was overlaid, I played the video, counting beats. I found that there were four measures every six pictures. I also found that this was completely consistent from start to end (well, it’s techno-ish, so the fact that the drummer is a machine helps keep the beat consistent). 6/4 meant that a single measure was 1 1/3 seconds long, which conveniently enough turns out to be 40 frames (NTSC has 30 frames per second, roughly). Obviously, this is not always the case. Having to deal with inconsistent beats, or inter-frame beats (ex, two measures in 39 frames, meaning the pictures need to be alternating 19 then 20 frames each to fill a measure), would add time in the polishing phase, covered below in “Timing to Music”.

At 40 frames for a measure, I initially thought that all my photos would stay onscreen for 1:10 (one and a third seconds). I set them all to 40 frames, then played it to see if the pacing looked right. It didn’t; it positively dragged on. So, I knew that the majority of photos would end up at the next multiple down (2 beats, or 20 frames), and that held true through to the end.

Transitions

Now, honestly, I jumped the gun here and did a bit of precision timing before I decided what to do with transitions. Fortunately, though, the timing I did preceded all transitions except for the intro, and I decided that the music wouldn’t start until after the silent intro, so I didn’t need to change anything after applying transitions.

The “theme” being used here which seemed to fit best was the “Bulletin Board” theme. So, from that, I got the intro overlay (in the “Text” area of the effects library) as well as various “chapter” transitions (between each of the major types of action shots). I knew that right up front I wanted to have the Roseville Thunder logo, which I pulled down from our website. By chance, the logo is wider than a TV screen, so filling the frame vertically chopped off a bit of the left and right of the logo; I decided that I liked that look and so kept it as-is (otherwise I’d have to import it into a graphics program and add white above and below, as iMovie doesn’t appear to be able to use any color background except black when an image is “fit” to the screen instead of “cropped”). I applied the “vignette” style to darken the corners of the logo. After dropping it in the “Bulletin Board” intro transition, I decided I wanted it to stay on screen a little bit before transitioning away; I found that I could change the timing of the “clip” (ie, the logo) separately from the “text” (the zoom in from the bulletin board to the logo), and so set the logo to stay up a little after the text was done. I then applied the first BB transition between the logo and the first picture.

By default, the music was “unpinned”, which meant it would start on the very first frame of the video (which would be the title/logo screen). I wanted it to start when the actual pictures came onscreen, so I dragged it slightly to the left until it’s start lined up with the start of the first picture. This “pins” the audio and gives it a different background onscreen. The color indication seems odd to me; I’m not sure what other significant difference there is between “pinned” and “unpinned” music tracks besides that one starts as early as possible and the other starts at a specific time. Presumably there’s more to that, but right now it seems like a meaningless and confusing difference in the visual representation.

I dropped transitions between all the major areas of the project. At the same time, I placed a chapter marker immediately following the transition, titled according to the action type. This doesn’t show up on the web, but it does on the DVD (or, I believe, if you download the video). It also shows up on our AppleTV, of course.

Here is a bit of an annoyance: iMovie tries to make the transition “no-impact”, yet can not allow the transition to span multiple clips. So, for instance, if you drop a transition (with a set 2-second transition interval) between a 40-frame clip and a 20-frame clip, you will end up with something like 31 frames of the first, 18 frames for the transition, and 11 frames of the third.

I think what iMovie is saying with the transition is that there are four “sections” to consider:

1. Pure first clip
2. Transition running with first clip dominant
3. Transition running with second clip dominant
4. Pure second clip

iMovie Transition in the Precision Editor

Sections “2″ and “3″ need to be the same timing (and add up to the timing of the transition); 1 must be kept larger than 2, and 4 larger than 3 (which is why we end up with 9 and 11 instead of 10 and 10 for the last two in the above case). To keep the transition “timing neutral”, the time for “2″ and “3″ get taken from the original time of the first and second clips.

Now, say you wanted to make the transition longer, above. Well, the second clip’s timing is the constraint there. So, go to edit the second clip’s timing. Note that every two frames you add here, one will actually end up in bucket “3″ (and shift one from bucket “1″ to bucket “2″), and the other will end up in bucket “4″. So, to make the transition last 20 frames, you need to add 2 to the bucket 4 (from 0:11 to 0:13). When you click “Done” you’ll see that your “0:13″ changes to “0:12″ (the other one went into the transition), and the “1:01″ from the first clip changed to a “1:00″, and that the transition is now 0:20.

In my opinion, iMovie is trying far too hard here. Personally, I can see buckets 1 and 4 never being allowed to be “0″, but requiring them to be longer than the transition just makes for some goofy fast transitions.

If you don’t realize the shell game going on underneath the covers, fixing these timings is a maddening game. Change the timing on the first clip to be what you want, then on the transition, then on the second clip, but then you find that the second clip’s change didn’t really all “take” and only some of the increase to the transition happened, so do it again. It took me way too long playing around here before figuring out the rules iMovie appears to be operating under. IMHO, this is a case of an algorithm being too smart for its interface.

I don’t think I really figured out what to do here yet. I think next time I’ll try just leaving the transitions fast instead of trying to “fix” their timing.

Ken Burns

Usually, I eschew Ken Burns effects altogether, as they get too “cluttered”. However, I decided to give them a go here.

The default applications, though, left a lot to be desired. I wanted to keep the general “breathing” pattern (first picture zooms in slightly, second zooms out slightly, third zooms in slightly, etc), and did with a few exceptions, and tried to avoid too many radical movements.

One approach which worked rather well was to use KB to yield “motion”. By this, I mean that the main player in the picture should move “forward” in the frame from start to end. This can be done by zooming out and exposing action behind the player, or by zooming in towards a spot behind the player.

Another technique used here was a “scan” across the photograph, starting at one interesting spot and ending at another. For instance, the start might be tightly framing the player’s face, and the end has the face at the edge of the screen while showing the bat hitting the ball.

Overall, I think there’s definitely an “art” to using KB without letting the gimmick overwhelm the slideshow. As you can tell from the video, I’m still working on ironing out the details there.

Timing to Music

Once all the transitions were in place, I went into the groups and found “sequences” of shots. Here, I used a few techniques to tie them together. Generally, I half-beated all but the last photo in the series (sometimes even less). Of course, to keep the general timing together I knew that the series would have to add up to a multiple of 20 frames, so half-beating wouldn’t always work. For instance, if there are two shots in a series, putting one at 10 and the other at 20 frames would leave 10 frames in the measure, which would syncopate the rest of the clips.. So, I tried a few approaches here, redistributing the leftover frames amongst the clips.

A few times I put the emphasis on the first photo instead of the second, such as when the first photo is the bat smacking the ball and the second is the follow-through. I’m not altogether comfortable with how those sequences flowed in the video, though.

There were a few “longer” sequences I put in, which I treated as pseudo-video. For one of our pitchers I had a full sequence of her pitch, front-on, and put that in as 5 frames each; for another, I had a similar sequence but of a faster pitch, which I put in at 10 frames each.

In sequences, I find that Ken Burns effect ruins the effect almost always. So, between “Fit” and “Crop” I’d generally choose “crop” unless there was no way to get all the important bits of the frame on screen with a crop.

Once I had the sequences timed out, I repeatedly viewed the slideshow looking for musical hooks on which to hang sequences, and moved non-sequence images around to get the sequences timing right.

Once all of this was done, on a more “naturally produced” audio track, I’d need to then go through and listen for off-beat transitions throughout the video. Since the beat was completely unvarying, this was a quick pass for this video, and only found a few transitions which had ended up being a frame or two shy of what I’d wanted them to be.

Update to clarify: Note that iMovie ’09 also has “beat markers”, which are more akin to what I’d used previously in Final Cut. Since the music here was very regular in its beat, I didn’t use beat markers. Perhaps on the next project I’ll try those out! From a quick play, it works exactly like Final Cut’s markers: you play the audio in the Audio Clip Trimmer interface with your index finger on the ‘m’ key and tap out the beat as the song plays. For irregular-rhythm music this looks to be the way to go.

“Sharing”

I’d call this “Publishing”, but Apple seems to insist on calling it “Sharing”. Whatever.

Once I had the movie completed, it was a piece of cake to put it just about everywhere under the sun.

First, I wanted to get it up on our MobileMe gallery so I could point the parents of the team at it (especially to make sure the end credits didn’t have any misspellings prior to burning a bunch of DVDs). I told iMovie to do that, and checked off small medium and large formats (no sense in “mobile” for this, in my opinion). The export process took about half an hour to complete in all three sizes, and then spent quite a while uploading to the galleries (for some reason uploading to MobileMe from here at home is always incredibly slow; one of these days I’ll have to try uploading from work to see if the fault is more our DSL upload bandwidth or MobileMe’s server capacity).

Next, I also wanted it locally to play on our AppleTV. Since this was on my wife’s laptop, and my desktop is synched to the AppleTV, I did a normal “Export” in the high-quality setting, copied it to my desktop, then imported it into iTunes and synced the AppleTV.

Finally, this video is the core of a DVD I promised the families on the team (along with all the 1-star or better “picks” of photos I took, in original resolution). So, I clicked on “Share | iDVD” and up came iDVD with just this video as the sole attraction. From there, I fixed a few titles, picked a nondescript theme (none of iDVD’s themes seems to fit with iMovie’s Bulletin Board theme!), and added the folders of photos as extra DVD-ROM material. The DVD was a cinch, as one would expect.

Nits and Conclusions

Okay, so here’s my list of things I don’t like about iMovie ’09 as a photo slideshow editor:

1. The transition-timing algorithm is goofy, unintuitive, and makes getting the “right” transition hard instead of easy.
2. Editing text inside iMovie ’09 is painful. For instance, the end credits were incredibly hard to type because the cursor just would not show up in the editor half the time. I also wanted to change the positioning of the “center alley” to better center the player list, and could find no way to do this in iMovie. Fortunately, though, I could just copy the text out, paste it into TextEdit, and edit it there (the white-text-on-white-background was somewhat alleviated by the use of an outlined font style, so this my not be a solid working solution in all cases).
3. Before figuring out that text could be copied/pasted out to TextEdit and back in, it took me a long time trying to get the center-alley text style in a non-scrolling text pane. What eventually worked was using the scrolling title from scratch and then dragging a non-scrolling title in to replace it; since the actual text was kept there so was the center alley. Seems like a static credits with center alley title style would have been included. In the future, though, I’ll just copy my scrolling text out to TextEdit, edit the ruler however is needed, then paste it into a static text screen. Should work for multi-column credits too, I imagine.
4. There is no happy medium between “fit” (all of the picture on the screen, black bars top and bottom or left and right) and “crop” (no black bars, can only be as wide or as tall as the photo is). A partial-crop, allowing the “crop” to extend outside the boundaries of the photo frame itself, would be nice to have, both for static photos and for Ken Burns animations. The workaround would be to add black borders on the sides of my images manually before importing them into iMovie.
5. There is no choice of background color for “fit” photos and videos. Black is stylish and all, but hardly the only choice one might want!
6. It makes my first-gen Dual G5 system crawl (Final Cut works like a speedy champ on the same system!) Essentially this project is not possible on an old G5-era or before machine; it really requires recent Intel-based hardware to get it done. By way of contrast, the app was smooth as silk on my wife’s new MacBook Pro.
7. Seems like we’ll need a lot more themes to keep slideshows interesting. Presumably Apple will be coming out with new themes for iMovie each year, as they have with iDVD.
8. Speaking of which, I expected to see matching themes in iDVD as in iMovie. I couldn’t find anything similar, so to avoid too much dissonance, I chose a very plain template for iDVD’s menus. This was an unexpected bit of tarnish on Apple’s usual hyper-polished interface design.

Overall, a much shorter list than I’d had for iMovie ’06, and none of them absolute deal-breakers!

In exchange, relative to Final Cut, the workflow here is significantly easier for a non-experts, and the overall process from start to end significantly shorter.

Overall, I’m looking forward to my next projects in iMovie ’09.

In case you missed the link, the finished product is on my MobileMe Gallery.

Any comments or advice? Add them below!

It’s been two and a half weeks since we pulled the trigger on dropping satellite television, and about three weeks since we started the grand experiment of moving all our “passive video” intake to the web.

How have we fared through this transition? The riots were quelled without too much bloodshed. The voices in our heads rushed forward to fill the silence of the room, but were beaten back before they got overtly destructive. The kids are even talking to us again.

Really, though, it’s been more of a non-event than we had dared imagine. We’ve learned a thing or two, though, that somewhat surprised me.

What’s Broken?

Starting off with the bad: what’s broken?

Frame Rates

Hardware-wise, I’ve concluded that the old PowerBook laptop isn’t all that great of a media center. Watching Hulu, the video sometimes stutters rather significantly, pausing every minute or so for a half second or so at a time (it’d be nice if Hulu offered something which stated the actual throughput or frame rate we are achieving!). I think it’s primarily a bandwidth (either at the network or CPU levels) issue, although the couple of times we’ve seen it happening with mostly-static shots has me wondering. We’ve tested our network throughput and get pretty good numbers for that through the site Hulu recommends (SpeedTest.net), and Hulu states they should run smoothly on 1Mbps, so I don’t think we are hitting a WAN network issue. The PowerBook has 802.11g networking on it, and is getting good interference-free throughput to the modem, so I don’t think the LAN is an issue either. The problem I have is that Hulu won’t give any CPU minimum requirements, so the G4 not being able to keep up with Flash video is definitely a possibility. In fact, it’s my main suspect right now (Flash is notoriously under-tuned on OS X and PPC processors especially, and the 1.67GHz G4 was last made about three years ago and was under-performing in general for non-optimized software even then).

This is quite separate from the general 15-fps problem with Hulu and some videos. “Chuck”, for instance, comes across at about 15fps on Hulu, even on a fast Intel computer and the 30Mbps connection at work (tested after-hours as an experiment only, honest!) The same is true of Heroes. This makes the video seem to stutter and jerk, especially when there is a lot of action on-screen. I suspect this is an artifact of the process: simply doing a pull-down or even line-doubling from interlaced broadcast SD (30 fields per second, 2 fields per frame) to progressive digital sub-SD (15 frames per second). However, the better digital feeds for these programs exist (the SD download of Heroes on AppleTV, for instance, is completely free of these artifacts). At first blush, the choice between Hulu and AppleTV appears to be ad-supported versus bought; in reality, though, Hulu also throws in resolution and frame rate issues which AppleTV avoids. This pushes “anything with action” significantly towards the AppleTV front, in turn pushing everything else back towards Hulu or the like.

HD Downloads

On the AppleTV front, I’ve found that downloading HD shows is a mixed bag.

Begin camera geekery for a moment:

This is my first time watching HD video, and I was surprised at how much more obvious focus issues are in HD. Obviously, I knew that the “circle of confusion” would tighten with more resolution. I just didn’t realize how tight the DOF is being held in, for instance, Heroes. Watching a Heroes broadcast last week, I was distracted time and again by character’s faces going in and out of focus as they naturally shifted closer and further away from the camera. There was an entire scene where the main character’s face filled the screen, and their hairline was perfectly in focus, but their eyes out of focus! The previous week’s episode had me enthralled from beginning to end, counting the makeup imperfections and pock marks on previously blemish-free actors’ faces, and I didn’t notice the focal plane issues at all. I suspect it’s a director’s mindset at play here: if the episode’s director is used to the SD resolution and is tightening DOF for style as much as possible for that resolution, the HD DOF will be too tight.

Okay, camera geekery done. Safe for everyone to come out again.

Back to the AppleTV in general: HD episodes take a really long time to download. The AppleTV is also rather finicky about how it downloads them: if you purchase multiple episodes at once, it doesn’t download in the order you bought them, but instead starting with the most recently aired. Tip to Apple: if I’m downloading multiple episodes in a series, I probably want to watch them in order! The least-recent should be given priority, not the most-recent! I’m seeing times of about 2 hours before the episode is “watchable”, and in one case starting it right then we ended up about five minutes from the end having to pause so it could get the last bit downloaded before we watched it.

Overall, I think the remainder of our purchases will be in SD. It’ll save us $1 each, and will be a much more seamless operation.

Finicky Sites

“NickJr” is a frustrating experience. When it works, it works passably well. When it doesn’t, it fails in the most unpredictable ways. For instance, the other day it would not load at all on the kids’ iMac upstairs. Looking into Safari’s activity window, it would get hung up trying to download stuff from “overture” (which is a rather questionable bit to be tied into a kids’ website to begin with, but that’s another topic). Going downstairs, though, it loaded (albeit slowly as always) just fine on the G4 PowerBook. Other sites were slow, but not inoperative. Turns out we’d restarted the WiFi access point and it had shifted from Channel 11 (which the iMac receives very well) to Channel 1 (which, inexplicably, doesn’t work well at all on the iMac); changing the 802.11 channel to “11” from “Automatic” fixed the issue.

Luckily NBC.com is largely redundant with the existence of Hulu. However, the sites odd definition of “fullscreen” which seems to mean “about 2/3 of the screen with a thick border and distracting edge ad” is annoying. CBS.com also suffers from this dictionary deficiency.

Show Availability

Show availabilty is also frustrating.

• “The Biggest Loser” is available on Hulu, but, unlike everything else, is a week and a day behind broadcast instead of just a day behind. We watched last week’s episode on AppleTV, but I don’t think we’re willing to spend $2 per week on this show, so we’ll just be a week and some behind instead (or skip it altogether).
• “Project Runway” and “The Amazing Race” are not available anywhere aside from YouTube. YouTube’s interface is really crappy and the quality is substandard and everything is chopped up into 10-minute pieces requiring a whole lot of mousework to watch an episode from start to finish. Hitting it on the AppleTV might help, except that typing anything into the search interface there is a horrible experience as well. I strongly question CBS and Bravo pushing their users to get these shows ad-free and low-quality instead of providing them through their already-established avenues.
• “Primeval” was available on BBC-America’s website, I swear. We watched two episodes of it there even! But, no longer. Now they just offer useless “clips”. Another show which might die on the vine because it doesn’t necessarily make the $2 cut. Note to execs: it’s a lot easier for us to emotionally invest in a show and pay the $2 per episode to get it in high quality with the knowledge that if we need to cut back we can watch the remainder of the season for ‘free’ ad-supported. Before Primeval was cut from online availability it was on the cusp of AppleTV watching. Now that it’s no longer available ad-supported, it’s fallen off that cusp, and into ‘not watched’.

What Works?

So, what works?

• To get the laptop working really well with our TV, we used DisplayConfigX. Highly recommended to get rid of the overscan and deal with the flaky VGA output of PowerBook G4s (which tend to inexplicably start shifting output to the left and up, losing the entire Apple Menu and half the menu bar; apparently this is termed “losing the back porches” and if you google for “horizontal back porch” you’ll find more information on this than you could ever want).
• A Wireless Mighty Mouse helps a lot in controlling a laptop across the room, but for search fields one needs a keyboard. You can either get up and key it all in at the laptop, or bring up the “Keyboard Viewer” (go into International settings and enable it as an input method) to bring up a mouse keyboard on the screen.
• Even better than a wireless mouse and screen keyboard, if we have another laptop open in our laps: Leopard’s built-in Screen Sharing works really well for controlling the laptop from afar.
• On the AppleTV front, my wife’s iPod Touch has a really nice “Remote” application (free download from Apple in the iTunes Store). This gives a keyboard for keying in search fields, which is only about 100,000 times better than using the remote and the onscreen keyboard to type anything more than three characters.
• We’ve been watching quite a few shows online. My sister-in-law has even traded hours of Judge Judy watching for hours of Barney Miller watching on Hulu.
• HD quality is really nice as a novelty, but not something I feel I need for every single show. As I said above, I have found it distracting as often as I’ve found it beautiful.
• Boxee is an absolutely wonderful addition to our Apple TV.

A Few Words About Boxee

Boxee is an open-source streaming media application, available for OS X, Linux, and Apple TV. It’s based on the “XBMC” project which provides some (but certainly not all) of the same functionality for Windows boxes. Most importantly, it supports seamless streaming of YouTube, CBS, CNN, Comedy Central, Hulu, and the major podcast networks (Revision 3, Next New Networks, etc) in a consistent and big-screen-ready interface.

Boxee is in “invite-only alpha” right now. Anyone with an account on Boxee can invite anyone else to join up, or you can put your email address in to the site’s front page and hope they invite you next Monday. Hint: I have an account, and comments are open on this post.

I just put it on yesterday (coincidentally, the day that Hulu broke on it!), so haven’t been able to give it a really solid run-through yet, but it looks like most of our watching on the main screen from here on out will be via it and Apple TV instead of the G4 sitting next to the screen. Hulu changed something about how it delivered ads in the last two days, which broke Boxee, but Boxee had an update out this morning which appears to have fixed it.

It offers a significantly-improved interface for YouTube (full screen without the playhead bar at the bottom!) as well as Comedy Central and CNN (Comedy Central, however, is significantly more “skippy” than Hulu, so we’ll stick to getting The Daily Show from Hulu). YouTube searches are remembered, effectively giving us a “channel” for The Amazing Race and Project Runway (major caveat: YouTube video quality is still two steps below VHS in a strong magnetic field, and Boxee can’t do a thing about that).

The installation was rather straightforward, although it required creating a “patch stick” USB drive and restarting the Apple TV twice. The only hitch was that we needed to “Update” Boxee twice after the install because the first Update didn’t get the absolute latest alpha. Final bit about Boxee is that it is open source alpha software. It’s worked quite well for us so far, but I won’t be overly surprised if it crashes and burns some night. We’ll keep the G4 next to the TV just in case.

Overall, I’m very hopeful that Boxee will make our setup even nicer than we need it to be, saving us energy (no need to have another laptop going while watching Hulu) and annoyance (trying to operate the wireless mouse across the room is annoying). The verdict is still out, but it’s looking good both from a general quality standpoint and a responsiveness-to-crisis standpoint.

What’s Missing?

Surprisingly, not much!

• I really wish that Boxee’s functionality (streaming of ad-supported content) was built into Apple TV instead of a “hack” on top of it.
• The unavailable shows (Project Runway, Amazing Race) which drive us to revenue-free sources (YouTube) instead of letting us watch ads or pay $2 an episode are examples of network short-sightedness. Hopefully over the next several months these last non-streaming holdouts shift course.
• I really wish AppleTV supported a wireless keyboard and mouse in addition to the remote control and iPod Touch Remote app. The latter is a good replacement for a keyboard if you happen to have a Touch or iPhone sitting next to you on the couch, but I think a well designed “touchpad” type of stationary mouse and mini keyboard would be an awesome addition to the AppleTV experience!

That’s it!

Overall, this “experiment” is going quite well. Here in the middle of the fall television season, we’re purchasing about 2 shows per week on the AppleTV ($5/week, $20/month or so), which is a significant savings over DirecTV (on the order of $60/month). Everyone seems happy with the setup so far, and we haven’t felt the need to watch anything “live”, yet.

For the past decade and a half, there have been two major options to receive passive video entertainment (aka “television”) in the home.

One could choose cable service, which in our case would mean Comcast. Comcast has “basic” cable (meaning, none of the channels we watch, except the local stations) for about $30 a month, or “standard” starting at a smidge over $50.

Alternatively, one could choose Dish or DirecTV, which start at $63 for the shows we watch, plus $5 to use the DVR we bought from them. This is what we’d been subscribing to, more or less continuously, since 2000 when we first moved into our own house. Our actual cost for DirecTV is $85/month, after all the add-ons and taxes.

A few “third party” equivalents (digital TV from your phone company, for instance) have come up in recent years, but they are mostly in the same mold as the other two: a large monthly fee for a wide choice of “channels” to choose from.

But, times have changed since our household last consciously chose our source of passive entertainment. Costs have gone up; DirecTV when we first subscribed was $40 for the “Total Choice Plus” collection of everything-and-the-kitchen-sink channels. At the same time, other avenues for home video have opened up.

We sat down a week and a half ago to look into our options and ended up with a stunning conclusion: we’re dropping satellite TV completely and moving to “fully on-demand” television watching.

New Sources for Video

There are two main types of new video sources we are planning on using: pay-per-episode and ad-supported. The for-pay service of choice (simply because it works with the computers in our house) is iTunes, and by extension our Apple TV device. The ad-supported services are exemplified by hulu.com.

“Premium” shows will be bought for viewing via iTunes and our Apple TV. The quality of shows on the Apple TV is absolutely stunning, far superior (in SD at least, which is my only frame of reference) to DirecTV’s offering. A few weekends ago we noticed we’d missed the premier episode of Primeval, which we had been wanting to watch. So, we bought the premier on the Apple TV and watched it, then immediately watched the second and third episodes directly from DirecTV; the difference was highly noticeable, especially in the “matrixing” of the darks. At $1.99 – $2.99 per episode (the higher price for HD episodes, where those are available), we come out far ahead using the Apple TV’s buy-what-you-watch model for our most “discriminating” viewing; I’ll get down to dollars and cents on this later though.

“Regular” shows will be watched via one of the ad-supported online video outlets. Currently, these include:

• Hulu: Aims to be a “hub” for multiple networks’ content. Includes a smattering of movies and a lot of TV shows. Point of aggravation: it is eager to mix clips with full-length episodes, and so you never know if it really has a particular show available until you click down to the show. Major conveniences: queuing and subscriptions to shows allow us to turn it on and see what we haven’t watched yet. Episodes are up the day after they air, and remain up for “a while” but not indefinitely. Also experimenting with live broadcasts, although we’ll have to see how that goes!
• NBC: Most of this content is also on Hulu, but sometimes things will “expire” from Hulu and live longer on NBC.com (or vice-versa).
• CBS: Only outlet for CBS shows (Hulu will link over here). Major aggravations: too many clicks to find available full-length shows.
• ABC: Only outlet for ABC shows (which we don’t usually watch anyway, but this is the Lost outlet!).
• Nick Jr: Kid shows, including Dora, Diego, Backyardigans, etc.
• Adobe Media Player: Largely an “also-ran”. Nothing really interesting here, but it might fill out in the future. Downloaded AIR-powered client.

All of these are “free” ad-supported sources, and so all of them display ads at the start, end, and often in the middle of the shows. You generally can’t skip the ads, so there’s no avoiding them, although the length of the ads is much shorter than those on broadcast TV (15-30 seconds every act or two rather than 2-3 minutes every act). The ads are definitely tolerable, although the lack of variety on some shows has already gotten tiresome.

Connecting

We have three televisions which need to be “fed” content: our living room, the kids’ tv upstairs, and our bedroom tv. The main focus is of course in the living room, but the other two can’t be left out in the cold indefinitely.

In the living room, the Apple TV quite nicely handles displaying iTunes “premium” content. We can order new shows directly from the television set, and everything gets stored both locally and up on our home server. The television set has a DVI input, but that is already taken up by the DVD player’s HDMI output (which I’d prefer to keep “pristine”). That leaves a few component, composite, and analog VGA inputs as options for connecting a new source.

Obviously, to watch browser-based content we’ll need a computer with a browser next to the TV. For the downstairs set we’ve conscripted either my wife’s MacBook laptop or an old G4 PowerBook for that purpose; we generally use the PowerBook, but it also tends to “stutter” on Flash video as it’s a bit underpowered.

For both of these computers, we have a VGA adapter (the Apple MiniDVI-to-VGA adapter for the MacBook and an Apple DVI-to-VGA adapter for the PowerBook). We also obviously have to hook the headphones output up to the receiver.

On the upstairs kids’ tv, we have a computer sitting right next to it. It’s an older-model iMac G5, which sports “Mini-VGA” output (one of the few machines ever made with such a connection). Another $20 adapter from Apple, and we can output it’s screen to SVideo, which the TV up there can accept (no VGA or DVI inputs on that old CRT clunker!) The sound goes through a pair of RCA jacks, or if we forget to connect the audio outputs, through the computer’s speakers. Here, unless we move the AppleTV from downstairs (which isn’t a big deal), we can watch AppleTV premium content via iTunes’ “bonjour” networking; turn on iTunes and connect directly to our home server’s library to watch a bought show.

On our bedroom tv we are set up to move one of the laptops in and output DVI video direct to the TV. We have a little alcove up there where the DVD player sits, perfectly ready for a laptop to be set and the audio/video connected.

The next issue is the 10-foot interface. Specifically: how do we control that interface from the couch?

With the AppleTV this isn’t an issue at all. Our Harmony remote controls the little box with aplomb.

Hulu et al are not at all set up to make this easy. We have to use a mouse and keyboard to get anywhere (keyboard only to search). One option is that we have to physically get up, walk over to the computer next to the television, and click the mouse there. The other option is a little bluetooth mouse, like the Apple Mighty Mouse. For the downstairs set (and if we move the laptop upstairs), we will be using the latter option. For the kids’ set, the room is small enough that they can just get up and push the mouse button. With a bluetooth mouse, we are still lacking a keyboard; to keep us from heading across the room, we have enabled the Mac OS X “Input Menu” (under “International” in System Preferences), so a click on the flag in the menu bar and we get a floating “keyboard” on the screen. This way, the bluetooth mouse can act as a keyboard in the cases where we need one (for instance, when searching for something).

Implications

The implications of this shift are significant in our household.

1. TV is no longer “background noise”. This is a good thing – from sociological and energy perspectives – so you won’t see me bemoaning it. At the same time, though, it is a change from the status quo, and so will be difficult. We’ve already felt the sting of not being able to lay down in bed with the local news rambling in the background.
2. TV “events” are no longer available. This is the most worrying effect. Instead of local news, we can read up on the local stations’ web sites, and we can get opinion pieces from national sources. We won’t be able to participate in the “watercooler conversations” (although half the time we couldn’t in any case because the hot show was sitting on our DVR waiting for us to watch it the next night anyway). I’m excited to see if the live streaming model of Hulu’s Presidential Debates and NBC’s Olympics takes a stronger hold.

Sociologically Speaking …

The television can no longer act as background noise. That’s an alarmingly drastic change in our household. I suspect we’ll be watching significantly fewer shows overall, and be much more discriminating about which we spend our time in front of.

If we are watching an hour-long shows each night of the week, that’s 7 hours per week. Taking all 7 as “premium” shows would mean us paying $56 per month for SD quality. Obviously, that isn’t the best option here.

At the same time, the “TV Schedule” has been a major drag on us. We haven’t been able to get to bed early because shows we (Jodi and I) want to watch are aired later at night. We haven’t been able to go out for the night because some “event” show was going to be on and we’d “miss” it. DVR recordings help somewhat, but we never know where to record things because it’s hard to say where we’ll want to watch them later on.

This move forces us “off” the TV Schedule. There is no schedule any more. Everything appears at about 4:00 in the morning on Hulu, and can be watched whenever we find the time to watch it. It also forces us away from “background noise”; each show must be consciously selected from the list of available shows, and nothing just “comes on” because the other show ended (although Hulu’s queue does act that way).

At the same time, it also moves the living room centralized viewing area to a more decentralized system. It’s not much worse watching the show you want to watch upstairs or in our bedroom; we no longer have the “it was recorded on the living room DVR” excuse to force us all to watch something in the same room. I don’t see this as overly significant now; we’ll have to see if it does become a factor.

Energy-Wise

The TV consumes a lot of energy. For “background noise” it’s about the least efficient of the options available (“silence” and “radio” coming at the most efficient end).

I fully expect our energy bills to go down as a result of this shift. At the same time, there are competing factors. The following calculations are based mostly on guesses, not actual power readings, so may be slightly off.

1. The living room TV uses about as much energy as 4 100Watt bulbs when on (it has a 300W rear-projection bulb; I’m estimating the circuitry as wasting another 100W but may be over in that estimate). Having this off for three hours in the day (where it was merely background noise) is a significant energy savings (1.2kWh)
2. The computers we are using are all fairly efficient energy sippers. Adding these to the circuitry mix when watching shows adds about 35W to the overall system (0.035kWh per hour; this is a very high estimate; Apple puts a PowerBook G4 battery at 58Wh and lasting for 4.5 hours, for a run rate of about 13 Wh or 0.013kWh per hour; I’m assuming the demands of Flash use much more energy, and the battery life of our G4 when watching Flash is more like 1.5 hours, which would be around 38Wh).

So, if we watch TV under the new model for 1 hour, we will be using 0.035kWh extra, the extra energy equivalent to about 5.25 minutes of “background” TV. If we eliminate only 1 hour of “background” noise watching in the day, we are energy-positive so long as we watch less than 11.5 hours of “real” television.

Imagine two scenarios: we have the TV “on” for 5 hours in the day, and then we have the TV and laptop “on” for 2 hours the next day. Let’s compare the costs (Assuming Tier 1 rates for SMUD, rounded up to 10 cents per kWh):

• TV on 5 hours
  • Energy Used: 2.0kWh
  • Energy Cost: $0.2/day : $6/month
• TV and laptop on 2 hours
  • Energy Used: 0.870kWh
  • Energy Cost: $0.09/day : $2.70/month

This doesn’t take into account the fact that the electricity rate might go down due to our conservation (higher usage customers get higher rates), or that “2 hours” of television is almost 3 full 42-minute shows on Hulu instead of 2 42-minute shows plus 36 minutes of ads on “regular” TV. It also should be noted that we’re not taking the electricity usage of the DVR into account on the old-model side.

Overall, the monetary effect is small, but the narrow energy usage effect can be huge. We’d never dream of having four 100 Watt bulbs burning all day long just because we were in the same room! Why should we have been so complacent about the TV having been on all day?

Premium Content Cost Breakdowns

We have several “premium” shows that we want to watch. Assuming for the moment a “premium” month where all are showing every week (four episodes), this is the max monthly costs:

• Lost. $2.99 * 16 ($48)
• Heroes. $2.99 * 22 ($66)
• ER. $1.99 * 22 ($44)
• The Closer. $1.99 * 15 ($30)

The “season” costs of these four shows (bought on a show-by-show basis, not as season passes) comes to $188 over the course of the year (with a max of $40 in any particular month were all four to show four episodes that month).

DirecTV costs $85/month for us, all told, which comes out to $1,020 over the course of the year.

If these are the only “premium” shows we watch, we’ll save $832 in a year.

The other benefits are budget flexibility (money tight? Move a show from “premium” to “online” viewing), storage flexibility (these can all, unlike the DVR’d shows from DirecTV, be backed up), and a better overall viewing experience (no skipping through commercials because they’ve already been pulled out!).

But, I hear you saying, you haven’t factored in the cost of the internet connection! Well, that is true. Obviously, we’d have an internet connection anyway (we are paying $50/month for internet and phone combined), but we might want to upgrade our connection to allow for the increased bandwidth needs. Moving from 3Mbps to 6Mbps internet will cost us $10 per month, eating another $120 in the year. However, the need for this is not foregone: we haven’t had any latency or bandwidth issues yet in the week we’ve been trying this out. We’ll see, though; the phone connection is switching over to digital as well, so the overall bandwidth needs may be going up soon. We’ll see, and adapt. No matter what, though, it’s not going to bust the budget.

Assuming we move to a higher-bandwidth internet service, and add two additional shows in the “premium” bucket ($2 per episode and 22 episode seasons), the cost comparison is:

• DirecTV: $1,020
• New Model: $396
• Savings Total: $624
• Monthly Savings, Total: $52

This is the “all bells and whistles” option, and it’s still over 60% less than DirecTV!

Future Options

As I just mentioned, it’s definitely possible we’ll be expanding on this new model in a number of ways:

1. Increase internet bandwidth
2. In-house, beef up networking (saving $50/month here we can put aside funds to pay for these upgrades quickly)
3. In-house, beef up connected laptops (much less likely to happen, but a better laptop or Mac Mini in the living room without having to take my wife’s laptop away would decrease the need for “premium” shows some of the time and enhance the overall Hulu-class experience)
4. Additional “premium” shows without removing existing “premium” shows from the lineup

In addition, there are other avenues:

1. Subscription movie services (Blockbuster, Netflix). We’ve canceled Blockbuster as we just haven’t been getting $18 worth out of it each month. However, we might reconsider this once we’ve lived in this new model for a few months.
2. Upgrade to capture Over The Air HD (purchase an EyeTV USB dongle or home server for $100-200). This would move all “network” content off the “premium” cost table in exchange for a one-time cost and excepting scheduling mix-ups. It would also allow realtime viewing of local TV shows (ex, the nightly news and late night shows), should we be feeling too deprived without that.

The Future Now

This is a paradigm shift for our household. We’re excited to be moving to this new model, and definitely ready to realize the savings it will entail. Our trial week on this has gone well. I’ll keep you up to date as we move forward.

The original portable music devices were centered squarely around albums, be that delivered in the form of a cassette or in the form of a CD. They strongly favored putting the album in, hitting the play button, and listening to it straight through. Whle CDs are obviously “digital music”, I don’t include them as such, as they are a digital equivalent to the analog forms which were three decades old at the time (LP, cassette).

The first digital players (MP3 CDs and Digital Rios, etc) broke through the album-orientation of portable music, allowing us to carry as many as ten albums with us and listen to them all mixed together. For the first time we could shuffle the entirety of our Pink Floyd collection!

The next wave of digital players brought a huge chunk of our libraries along: the Creative Nomads of the world. The interface, though, was still centered on two major use cases: navigate to a specific playlist to play, or put your entire library on random shuffle.

The third wave of digital music players came when Apple unveiled the iPod. Now, we could easily navigate to any of the artists or albums on our device, shuffle based on artist or album or entire music collection. We also had, importantly, desktop integration with iTunes. The “scroll wheel” metaphor was intuitively obvious for people seeing it the first time. Gone were the banks of buttons with indecipherable icons. The basic form underwent some serious tweaks and streamlining through the early 2000’s, but the main mechanism and interactions remained the same throughout.

I contend that the iPod Touch and iPhone constitute the fourth wave of digital music players. With this new device model, Apple (or someone else) has the chance to reinvent the basic interaction model.

Why would they want to do this? Primarily, because our digital music collections have, in general, far outgrown the bounds of the original iPod control scheme. I would go so far as saying that the iPod control and navigation scheme on an 80GB iPod Classic borders on unusable if you actually have even half that filled with music.

So, smarty-pants, what should we do about that? I’m glad you asked.

Ratings Filtering

Sample Filter / Sort screen for album

One of the great things about Aperture is the filter drop-down. The same controls that can be used to create a “smart album” are also available in the filter drop-down. For instance, with two clicks on the filter control (or one keyboard shortcut), I can quickly filter the view to only include the photos I’ve given one or more stars.

I often miss this functionality when listening to my iPod. There are albums that I love, start to finish: those I always just want to listen straight through. But there are many more albums where I’d much rather just listen to “my picks” from that album. You could arbitrarily say that’s everything 3-stars and above, but it’d be much nicer if I could also at times look at just the 4-star songs, etc.

Sort Order

In iTunes, I can sort by just about everything, including secondary fields (like “Album by Year” which sorts by year, then within that by album name, then within that by track number).

What do I get on the iPod? Generally, just sorted by Album. When coming in by Artist, the albums are sorted alphabetically, and the tracks beneath them in album order; the “All” pseudo-album has each track sorted alphabetically by album then by track order, making it useless to scan the artist’s discography (unless they released their albums alphabetically) as well as to find a specific song (unless you know all songs’ albums and track numbers).

The sort order need not be “hard coded” on the iPod Touch. The same “Filter” screen above could include a drop-down selection of various “Sort” options. I’m sure Apple can figure out the five most absolutely necessary ways to sort each of the screens on the iPod.

Major Artists

The “Artists” list on my iPod is next to useless. Why? Because every artist who ever sang one song which I liked is listed there. There’s no ability for me to scan through the artist list to find an artist I like as an artist (rather than as a one-hit-wonder).

I’d love to be able to filter the “Artist” list by “Only with at least this many songs [of a specific rating or higher]”.

I’d also love to be able to turn off/on the “Compilation Only” artists in the Artists list. Most of the time, I don’t need that artist who only exists in my library because I ripped my “The Good Crap from 2006” CD a few years back. I want to keep the song around (a one-hit-wonder category song), but I don’t need the artist showing up in my Artists list. Except, of course, for when I do: I sometimes hear a song on the radio and know the artist, or someone asks me if I have that song by that guy, and it’s nice to be able to look the artist up, compilation-only or not.

Best Albums

Similar to “Major Artists”, there are albums then there are albums. It’d be nice to be able to filter or otherwise highlight (highlighting would be really cool) a handful of different “album” types:

1. “Albums” with only 1-2 songs in them. These are “one-hit-wonder” albums, regardless if the artist is a one-hit-wonder. If “of #” information is in the track numbers, then replace this with “less than 50% of the album”.
2. Albums with fewer songs than the full album. This requires the “of #” track information to work, and would include all albums not in the above category and not full.
3. Full/Overfull albums. Albums where all tracks are present and accounted for.

In addition, an indication of “average rating” would be nice. It shouldn’t be hard to pick up my iPod and find the ten most loved albums!

Navigation

I live and die by my playlists. I have a “smart playlist” which automatically picks a certain percentage of 3-star songs, a higher percentage of 4-star songs, and all 5-star songs. I put that on “Shuffle” and it’s almost always a very Good Thing.

Invariably, though, ten or so songs in it’ll hit a song and I’ll say “Oh! I haven’t listened to that album in a long time!” Or, I’ll say, “Oh, that’s a good song by that group, now I want to listen to (some other specific song by them).” I then have to stop the playlist shuffle, go out to the Artists list, find the artist, find the album, and start playing.

There should be a shortcut while any song is playing: go to the artist for the song, go to the album for the song, list all playlists containing the song.

I tend to crop as a rule, as it means I capture “outside the crop” routinely.

One scenario I’ve to fairly often is needing to fit a well-framed picture into a different aspect ratio (like, say, a TV screen rather than the native 4×6 ratio). The choices here are:

1. Live with black bars top and bottom
2. Remove (presumably vital) bits of the picture left and right.

If that 4×6 was cropped down by 10%, then I get a third option:

1. Replace extra picture context top-and-bottom.

The question then becomes: which preserves the integrity of the picture better, adding potential distractions top and/or bottom, or removing potential vital bits left and right.

Of course, when always cropping, you “lose megapixels” (or, more precisely, resolution, which might be lens resolution or sensor resolution). That is of course true, but I find I generally have more resolution than I need, especially considering three or four years ago I was happy enough with 3MP so long as I didn’t have to crop anything.

Keeping the master aspect ratio is my default action here. It’s rare that I change the directionality of the image (ie, from portrait to landscape), and only happens if I mess up in the field. I also keep my “reference images” consistent at 4×6 crops (even cropping down the 3x4s my wife’s camera makes), which makes flipping through them less jarring than otherwise and keeps them ready to proof out on 4×6 paper to share with anyone who wants to hold a print. “Special” crops are kept as separate versions, and might be any aspect ratio (the frame may dictate this, or I may choose an aspect ratio which does the best job of capturing the spirit of the picture).

I also find that the crop-as-default is a great tactic when working on image series. While I try to move with the action when, say, a player is sliding in to home plate, I find that when I’m looking at the series later on I will want to adjust the focal point from frame to frame. This may be keeping the focal point unmoving throughout, or it may be moving the focal point across the screen. Being able to try both (and the various middle grounds) in the comfort of my home allows me to make the right choice for a particular product (ex, the end-of-season DVD) and change my approach for other products (ex, a single-frame blow-up of the tag).

Example Series

Here’s an example from a recent tournament game for my daughter’s team. One of our players stole home on a wild pitch, sliding in flat under the tag from the pitcher. It was a close call. You can step through the series starting at the first image.

The first image is cropped in to bring focus to the primary players (the catcher and the runner). This is where the focus is kept throughout until literally the last image. I know these are going to be displayed as a series, though, so I will need to discreetly adjust the frame to allow more space to the left throughout the series.

By the fourth image, I’ve kept the crop pretty stationary on the right while pulling it out a little on the left, so that in the fifth image (to right) the umpire’s hand is well in the frame as well as the billowing dust cloud from the action.

(Note: I’m not completely agreeing with the Ump’s call here, as it looks like our player had her foot up off the plate when she was tagged, but as I tell our girls: as many of these go against you as go for you.)

If not for the valiant and selfless efforts of Queen and Survivor, where would season-end slideshows be?

Following Video Slideshows with Aperture and Final Cut, I got the following comment:

Tom I coach a 7 yr old girls softball team and am going to try to make a slideshow dvd for them can you suggest some music. This seem to be the hardest part for me. Thanks

Picking the right music is definitely one of the hardest parts, whether your audience is 7 or 17 or 37. If it is a mix of ages, picking music is even harder.

I tend to spend a lot of time worrying over the music, and in the end feeling a bit silly for having worried so much. It’s hard, and can become a game of infinite second-guessing, to pick something catchy enough to propel the slideshow yet not cheesy and obvious, inoffensive literally and in innuendo yet not Lawrence Welk bland.

What I tend to find later is that if you have some good pictures in the foreground, it really doesn’t matter as much. People subconsciously bob their heads or smile at the memory the song brings to their minds, but the high-level reactions of laughter and comments all come from the pictures themselves.

Still … There are a few different ways to go with music.

Commercial Music

Commercial music (ie, what you’ll hear on the radio and buy at the record store) has two advantages, one major disadvantage, and a minor disadvantage:

1. Advantage: it will give an instant “connection” with the audience. From the opening riff of a popular song, your audience might know the “feel” of the slideshow, and what to expect.

2. Advantage: it is easy to find out what your audience likes (observe what they listen to / talk about)

3. Disadvantage: It is, technically, illegal to reuse copyrighted music as a major part of a larger work. There are exceptions there, and many thousands of web pages all over the place debate what precisely they are, but in general, if you are including more than a few seconds, and are not parodying the work itself, then you need to get permission from the copyright holder. I am not a lawyer, but I think that is correct. How important this is to you depends on how much you’ll want to show off your work. If it’s just going out on a DVD to the families of the team, then there’s probably not a lot to worry about; if you’re going to be posting it to your web site and especially if you are going to be building a business around it, then

4. Disadvantage: there is a good chance that some portion of your audience has already formed an opinion about the specific music, and so will deem it “not my taste”, “annoying”, or “overplayed” before you’ve had a chance to wow them with the whole presentation.

For picking commercial music, my tips are:

1. Pay attention to what your children listen to. At 7 this can be hard, and maybe not overly productive (our 7-year-olds didn’t have a “group think” music taste and listened to just whatever we gave them to listen to … ) This is easier with older children.
2. Listen to music yourself and try to judge what they’d like. Almost certainly, they’ll like music you find mildly “too old” for them, so skew that way.

Finally, if you are going to use music bought online, be aware of the effects of DRM. While iMovie will allow you to put a protected audio file in as background music, Final Cut will not. You will need to “de-DRM” it first (for instance, by burning it to a CD and ripping it back in as MP3 or unprotected AAC).

Creative Commons

With Creative Commons music, I’d tend to trend non-lyrical. I’d start at Jamendo for a good selection of CC music. Obviously, no one is ever going to have heard of any of this music, so you’ll not get the immediate connection you’d get with your audience using commercial music. Still, it’s legal, it’s fresh, and it can blend in with the background or stand out based on your own preference.

Tips:

1. “Creative Commons” does not mean free use. There are several rights which may or may not be granted by a particular license. Please pay attention to what you can do with an album freely! Jamendo puts a “Your rights on this album” block off to the right. If you see a “=” sign in a circle, it means you are not allowed to alter or build upon the album. If you do a search, click on “Advanced Search” when viewing results, tick the “Find content I can modify, adapt or build upon” checkbox, and click “Ok”. The Creative Commons license very specifically terms what we are talking about (synching pictures to a CC’d soundtrack) as an Adaptation, and this is reflected in the “build upon” right. Also pay attention to the “Share Alike” requirement: this means that your slideshow will need to adopt the specific CC license granted by the music author.
2. Often a right is not generally given, but contacting the author directly will allow you to use it. Where this differs from the same situation in the commercial music case is that you are much more likely to get a 24-hour response from an email to a CC artist than you would from Queen or Survivor (or whatever their particular rights-holding organizations are).

Oh, and, generally, it is free. Finding “good” music is much harder when someone else isn’t telling you if its good or not, but with nothing more at risk than the time to listen to a song or two or ten, why not take a little time? You may well find a few artists to add to your collection! Even the not-free music (ex, Jonathan Coulton) usually is as easy to preview or stream as you’d find on iTunes or Amazon.

Do pay attention to the licenses. Make sure you give credit prominently on your materials (an MTV-style attribution block at the start/end will do, as will a specific “About the music” screen on the DVD).

Finally, “share alike” or not, I’d highly recommend that you put a Creative Commons license on your slideshow. It makes explicit how others are allowed to use your work (and generally speaking you’ll want a carte blanche type license) and keeps people from asking you later on. Creative Commons is a really good idea, which greatly simplifies using media for personal or semi-public projects like this, and should be supported wherever possible.

Make Your Own

The ultimate free music, of course, is what you create yourself. Fire up Garage Band or Soundtrack Pro, throw a few loops in there, and let it rip. Do make sure you “audition” your creation with a few other folks first, though. Nothing is worse than an 8-beat riff looped 25 times!

Final Notes

In the end, though, remember that the music should be second fiddle here. This shouldn’t be a music history lesson, nor should it aim to widen anyone’s musical horizons.

The star of the show is the video, the pictures, and the memories the team has formed over the course of the year. That is what will make your video a success. The music should be a framework to provide structure and tone, but should ultimately fade into the background.

As our kids go through season after season of sports, I’ve taken the opportunity to bring my photography into my spectatorship. Each season for the past several years, I’ve spent a significant amount of time capturing pictures of their games, and then even more time at the end of the season wrapping the last several months’ worth of pictures into a slideshow. I’ve decided that this is something I do often enough that I’ve found a pretty optimal workflow, but not often enough to remember the details the next time I go through it. So, as much for my benefit as yours, here is my general workflow for assembling the final slideshow.

Note that I am using Aperture 2.1, iDVD, and Final Cut Studio 2 (Final Cut Pro and Motion, specifically) to do this project. You could accomplish the same with iPhoto and Final Cut Express, although the iPhoto organization workflow is a bit different than Aperture’s, and FCE will require the use of its built-in titling instead of Motion for opening/closing titles. You could also use DVD Studio Pro instead of iDVD, although personally I prefer the interface of iDVD and generally have not hit any annoying limitations there (this DVD is fairly straightforward, little more than a delivery mechanism for the slideshow). Note that iMovie will not work for this purpose, primarily because it is impossible to set markers down at the audio beats. If you don’t want the slide transitions timed to the music then that might be adequate, but I find that seriously distracts from the overall slideshow.

Finally, the time commitment for this project varies. Not counting the up-front time taking and cataloguing the source pictures, compiling the slideshow can take anywhere from 2 hours to several days (wholly dependent upon how much fine-tuning I decide to put into the particular slideshow). If this time commitment is too much, you should consider using an automated tool for the job instead (see the last section on Alternatives for a few links pointing you in the right direction).

Setup

First, a little file management. I like to keep my projects as organized as possible. This allows me to both not get lost now, while I’m working on it, and to be able to look at what I’d done while planning out next year’s project.

1. Create a folder for the slideshow project. I put this in a “Projects” folder, which contains all the video or DVD projects I’ve ever done, categorized by year and topic (“2008 Softball”).
2. Inside that folder, create multiple subfolders, including “Pics”, “Titles”, “Music”, and “Output”.

Photo Management

Throughout the season, I have been taking pictures. My general workflow is documented already, and largely remains unchanged for this type of project. After each game, I:

1. First-pass the pictures of the game (separating any family pictures which mean something to me but which I wouldn’t want to share with the rest of the parents into their own project). The output of this step is a set of starred photos which should be significantly smaller than the overall set of photos.
2. Apply a jersey-vibrancy color correction across all pictures. The idea is to make our team’s jersey “pop” just a little more than natural.
3. Apply any cropping, highlight/shadows corrections, etc, as needed. For the most part, any softball pictures which include batters in their helmets will need some shadows adjustments; care needs to be taken to not overdo this or the result will look more like a cartoon than a photo.
4. Add any comments on the specific picture to the version name and/or comments
5. Collect all one-star-or-better pictures from the game, create a new Web Gallery of the selected versions, and upload to my .Mac server (hiding the album, allowing addition of pictures, allowing download of original size images).

On previous seasons I’d also tagged each image at this point with the key actors in the picture. This season I skipped that step, as I didn’t have the goal of compiling a scrapbook-per-player. That said, more than once I kicked myself when I couldn’t with any certainty say that I had a picture of a given girl. Next season, I won’t be taking that particular shortcut (which, truthfully, didn’t save much time in the first place).

At the end (although it obviously could well have been done beforehand), I selected all the pictures of the team (by clicking on each web gallery in turn and selecting all) and applied a single keyword to them all (“Sports > Softball / Baseball > 2008 – The Blue Angels”).

I then went through each gallery, sorting the various actions (“Sports > Softball / Baseball > Actions”) between “Offense > Base-Running”, “Offense > Batting”, “Defense > Pitching”, “Defense > Catching”, and “Defense > Fielding”. I also have top-level “Cheering” and “Goofing” actions. I set up a keyword button set to include each of these base actions, as well as the “Offense” and “Defense” general groups. Working my way through each gallery, I would select all contiguous “batting” pictures, and click on the “Batting” button, etc.

In my case, I wanted to sort the slideshows so that the “early season” (meaning, all the way up to the final game of the mid-season tournament) was the first group, then the climactic mid-season game was the second group, then everything “late-season” was the third group. To give better continuity, I chose to place all offensive plays before all defensive plays in the early season (helps the viewer track who is doing what in the pictures as they zip by), and the same in the late season, but to keep the climactic game properly chronologically ordered. To do this, I set up three smart albums (based on the “2008” level of my project folder hierarchy so that it would include anything taken Fed through June, yet not have to filter through the tons of older photos):

1. “All Blue Angels” – this includes all starred images with the “2008 – The Blue Angels” keyword.
2. “Blue Angels Offense” – this includes the above, but also requiring the “Offense” keyword (note that “Batting” is in the “Offense” group and so is included in this filter, as an example)
3. “Blue Angels Defense” – this is the same as above, but the filter is for “Defense” instead of “Offense”

One thing which annoys me with albums in general is that all smart albums lose their “custom” ordering. A few sets of pictures were unstacked continuous-shot groups, with as many as 6 containing the same timestamp. This befuddles the Aperture ordering algorithm, which tries next to order by version name. Nothing wrong with that, until you give one or two of them a descriptive title instead of the IMG_1234.CR2 name your camera came up with. What I’d like to do next is drag the first half of the “Offense” album into a new album, then the first half of the “Defense” album, then the climactic middle of the “All” album, then the last bits of “Offense” and then “Defense”. But, if I do that, then go back to the new album, we quickly find that everything is back to “timestamp ordered” (with version name as the secondary sort), exactly how we don’t want it.

So, here’s my trick:

1. Create a new custom metadata field called “Order”. You can do this by clicking on the “Metadata” tab, clicking “Other” at the bottom, then typing “Order” and “010” into the name and value fields then press enter. This puts the given field on the currently selected primary photo.
2. Select the group of photos we want sorted first (the first half of Offense).
3. Go to MetaData > Batch Change.
4. Click on “Replace”, then next to “Order” type “010”. Hit “OK”.
5. Repeat for each subsequent group
6. Finally, in the “All” album, make sure the “Order” value is showing, and sort by it.

Oddly, when sorting by Order, the secondary sort appears to be master file name instead of version name. This just happens to be perfect (although it might be a problem if you are pulling from multiple cameras).

The final step in Aperture is to export all these photos, in order. I select them all (be sure to click on the first in the album list then Command-A), go to File > Export > Versions …, then use a custom file naming scheme I call “Custom Counter Version”. That particular file naming scheme uses “Custom Name”, then “Counter”, a hyphen, then “Version Name”. The “Counter” is set to four digits for me, as that is never too little. I export these pictures all into the “Pics” subfolder of the project. Then, I’m done in Aperture.

Note that I export them all as full-resolution JPEGs. This is primarily so that when I put them on the DVD-ROM portion of the disk they look as good as they possibly can. However, a case could be made to instead export only NTSC-boxed (fit within 720×480 rectangle) images in one folder and separately at full resolution for the DVD-ROM, assuming I’m not planning on doing any zooming-in or such.

Choosing Music

What music to use? Well, generally speaking we should limit ourselves to Creative Commons music that no one’s ever heard, or self-created music, or public domain performances. Depending on the size and composition of your audience, you may feel comfortable using copywrited materials from your own library.

In either case, one thing I do is keep a playlist in iTunes of “Soundtrack Favs”. As you might guess, these are the “short list” of tunes I’d generally consider using as a backing soundtrack to a slideshow or other video. Obviously, they are family-friendly songs, usually with some bit of quirkiness to them.

I choose the specific song by looking at the number of photos and the intended mood (do I want hyper-fast slides, 2 per second, or contemplative, 4 seconds each, or something in between?). In the Blue Angels’ case I knew I wanted a medium-paced slideshow, about 1.5-2 seconds per picture, although I also knew that I’d want a few burst of faster (sub-second) timings in groups. This left me at a little over four minutes.

Looking at the four minute songs, plus or minus 30 seconds, I then listened to every one of them (at least the , looking for one which both fits the intended mood and which has some special relevance.

Once I’ve picked the background music, it’s time to start working in Final Cut. I place it into Final Cut as the A1/A2 tracks. I double-click the track from the timeline to bring it up in the Viewer window. I then make the Viewer window as large as possible (I’ve saved a Max Viewer layout), set the playhead at the start, and listen through it again. This time, while listening, I lay down an audio marker at every beat (press the ‘m’ key at each downbeat).

I’ve played with doing this in SoundTrack, which provides a slightly better interface for marking the beats. However, I haven’t been able to get these markers to show up in Final Cut no matter what I do.

Once this step is done, moving back to the standard layout in FCP, we should see that the markers we’d laid out in the Viewer show up inside the timeline block. This is separate from “timeline” markers, which show up in the top ruler of the timeline. However, like “timeline” markers, when snapping is on, edits will “snap” to these markers. That’s the real key.

Adding Slides

Now, let’s save ourselves a bunch of time. Look at the timeline. Find a rather consistent beat (some songs will speed up and slow down throughout; we’re looking for a decent stretch where the beat is staying consistent). Set the playhead at the first beat of this section, and note the timecode. Move over ten markers, and lay the playhead back down. Now, look at the timecode again. Subtract the first timecode from the second. Multiply the seconds by 30 and add it to the frames part. Divide this by 10. This will tell us how many frames lie between each beat in that area (averaged across the ten beats).

In Final Cut Pro, go to Final Cut Pro > User Preferences … dialog. Click on the Edit tab, and modify the Still Frame Duration to be just this number of frames (enter “;32” for 32 frames, for instance: note that that is the semi-colon preceding the frames count).

Drag the entire Pics folder from Finder into the “bin” of Final Cut Pro. This should create an FC folder of the same name, with all the pictures inside.

Inside FCP, select all the pictures and drag them from the bin onto the timeline (make sure that when you release, the cursor is a downward-facing arrow, not a right-facing arrow; this is “overlay” versus “insert”).

Now, you should have a pretty good approximation of the slideshow, timed to the music. A little clean-up, making sure your pictures end on the beat consistently, and you could be good to go.

Add Titles

I do most of my titling for slideshows in Motion. I try not to have too many timing-specific transitions in the titles, so that I don’t have to match them up to the underlying music beats. However, if I do, then it’s easiest just to look at the frame timecodes for the beats in Final Cut and replicate that in Motion.

Note that titles can go “inside” the audio track (meaning, the background music is playing during the title) or “outside” the audio track (meaning, there is silence during the title), or somewhere in between (meaning, the music can start/end in the middle of the title).

Titles can likewise go “inside” or “outside” the slideshow. I tend to avoid putting them inside (overlaying) the slideshow as this distracts from the pictures of the slideshow. That having been said, I’ll often pick a “representative” photo to back the titles, and have it grayed out (de-saturated) under the titles.

Fine-Tune Pictures to Show

Most likely, you either have too many pictures for the song (the video track extends well past the audio track), or too few (the video track ends before the video track). There are four major changes which can be made to fix this situation:

1. Change picture timings. If there are too many pictures, shorten the lead-in or trail-out pictures in a series. For instance, if you have five pictures of a player rounding bases then sliding home, maybe the first one is kept long, the next three half-sized, then the last one kept long (a 1.5 picture reduction). There is an art in shortening the right frames of a burst or series. Experiment with which you want to do and see what makes the right impact. If instead there are too few pictures, try expanding the timings on some shortened pictures, or hold one or two pictures to increase their effect.
2. Add or remove pictures. Adding is easier than it sounds. You probably already have “loser” pictures which could go in a series (flashing by at half-duration no one will notice the runner was out of focus or the coach had stepped into the frame); this is where you take off the still-photographer hat and instead put on the videographer hat. Each frame in a movie isn’t perfect; instead, it lives on due to the excellence of its neighbors. Removing, on the other hand, is just as hard as it sounds. You simply have to suck it up and take the old writer’s advice: kill your babies without mercy.
3. Titles can be moved inside (to pad) or outside (to reduce) the music track, and inside or outside the slideshow track. Generally, I have a good idea where I want the titles to be, more or less, and only give that up if I can’t live with any other choices. Still, there’s often wiggle-room there if push comes to shove.
4. Manipulate the audio track to make it longer (find a repeatable section and clone it) or shorter (cut-out or fade-out early). This will almost always require a round-trip to the likes of Soundtrack. Luckily, such a round-trip won’t remove your beat marks, although obviously they may well be incorrect afterwards.

I try to generally settle on the content of the slide show at this stage. However, I take good notes on any “compromises” I made above, in case the next steps make it so I can undo one or two. I also keep note of the “next compromise” to be made if the next steps move in the opposite direction.

Fine-Tune Timings

Now that we have all our pictures lined up and the rough timings (one beat, two beats, half beats, etc) figured out, the next step is to fine-tune those timings.

Starting at the beginning of the timeline, I use the “Ripple Edit” tool (press ‘r’ twice) to “fix” each transition point to its nearby audio marker (the beat location). This tool changes the duration of the left picture, and shifts all pictures to the right to make room (alternatively, you can think of it as not changing any other durations). This is in contrast to the “roll tool”, which changes the duration to the left and inversely the duration of the next picture to the right; no other picture timings get shifted.

Once I’ve put everything “right” in the timeline, I maximize the output window (again, this is a convenient layout so I have it saved as a custom layout) and play the sequence through, paying attention to how well the pictures work with the underlying music. A double-speed series of shots will often seem “out of place” with the location it has fallen in the tune. The options here are simple:

1. Move “standard” pictures from one side of the double-speed series to the other side, to move it “left” or “right” in the timeline.
2. Make the series standard speed again. Obviously, you’ll have to compensate for this action elsewhere to keep the overall timings matched.

Once this fine-tuning is done with, I’ll often find that the video track has either shrunk or expanded beyond the audio track. I then loop back around to the previous step and either make more compromises or undo previous compromises to get things to match up again. After any changes there, I repeat this step (assuming I didn’t just remove or add something to the end of the video track).

Add Transitions and Effects

Once all the timings are down and locked in, I look at the slideshow to see if there are any transitions which make sense. Generally speaking, I find the straight-cut transition is the best approach most of the time, especially on fast-paced slideshows. However, on more ponderous slideshows a slight “Ken Burns” type effect can break up the monotony and enhance instead of distract.

Burning to Plastic Discs

Once the slideshow is complete, I save the output as a .mov file in the “output” folder. I then open up a new iDVD project (save it in the same folder structure as above), pick a template, and drop the movie on.

I also go into the DVD-ROM contents and add the entire contents of the “Pics” folder. This allows every recipient of the disk to take any of the included pictures and print them out without having to bother me about it later on.

I always do a two-step burn: first create the disk image, the burn it to plastic. This makes burning multiple copies fast and easy (no need to recreate the disk image for each output disk, just use Disk Utility to burn as many discs as necessary).

Alternatives

Obviously, a lot of people would be happy with just a standard-timed 3-4-seconds-per-image slideshow backed by random music, perhaps with a random Ken Burns effect thrown in. This workflow is obviously not for them. Still, I often find I’d be about 90% happy with such a thing, if only I could tweak it to get the last 10% out. Very few tools allow for this.

One “starting point” tool I haven’t played around with much is the ConnectedFlow Aperture-to-Final Cut plugin. It is a free tool which takes an album of pictures in Aperture and creates a timeline in Final Cut which contains them. This holds some great potential for simplifying a few of the steps above, but I haven’t had a chance to try it out yet (came across a link to it after I’d already done those first setup steps for this year’s slideshows).

Also, see the O’Reilly series on slideshows from Aperture for more options.

This post is about a new camera. But, cameras form memories as solidly as they record them, and so I can’t help but get a little nostalgic over loves past, and amazed at how far we’ve come along the way. So please, allow me a stroll down memory lane. It’ll only take a page or two, but if you’re in a hurry, just click that scroll bar over there to the right a few times until you see the First Impressions heading.

I have enjoyed photography since I was a child. Starting with an old Vivitar, I eventually moved up to the heaven of SLR photography with a second-hand Pentax K1000 during high school (which had “Asahi” branded across its nose, betraying a better age, and a semi-broken light meter which had to be checked with common sense to avoid spoiling film; I believe my dad had bought it from one of his Air Force buddies to keep my grubby mitts off his well-maintained world-traveling-yet-mint-condition K1000). I loved that camera, as did legions of other photography buffs around the world. Even then, the K1000 brand had been relegated to all-plastic shadows of its former self, a product of an age needing a new take on an old camera every year or two instead of every decade, where “manual” was a cost-cutting measure all its own. I had just one lens to mount on the body – a standard 50mm lens which I believe had shipped in a kit when the original owner had bought it. I enjoyed the science of it, the ability to twiddle with the various parameters to get the right amount of light on the film with drastically different effects. On the downside, though, I never wanted to develop my own film. I had no desire to deal with the chemicals, the smells, the dark room. And so, my prints went out to the dime-store developer, which was both a time consuming (one week turn-around) and expensive proposition.

In college, in one seven-week period, my K1000 was stolen (from my luggage while traveling down to Washington DC), and then the Ricoh I’d bought there to replace it was also stolen (from my hotel dresser, damn you State Hotel of Washington DC!) A few years later, my wife (who had been my brand new girlfriend throughout that ordeal) bought me a Pentax P30t with a 35-80 zoom lens (my first zoom lens, ever!). This was her first act of enabling my addiction. Over the next several years, we kept the point-and-shoots close at hand (even traveling fairly far down the Advantix APS dead-end road before coming to our senses), but recorded the major milestones of our lives on that Pentax. It was a great film camera, as beloved to me as my K1000 had been. It’s congenital defect: being a film camera, it tended to shoot pictures on film.

Enter the digital age. Our first digital camera was a Kodak high-end point-and-shoot, the Kodak DC4800. The 3.1 megapixels of the camera seem quaint by today’s standards, but at the time it was a vast sea of information. Yet, for all that, it seemed the quality of pixels would never come close to the quality of decent film in my old Pentax. We’d print out the pictures, mostly at home on our printer, and were never completely satisfied with the results. Still, the convenience was a watershed. From then on, the Pentax saw less and less use, until finally it was set aside in its hip pouch carrying bag and not brought out again.

During the Kodak years, the final straw for my Pentax (and, coincidentally, my Windows PC) was the discovery of iPhoto. All my digital photos were available for instant review at my fingertips. I started taking the old film prints (the negatives unfortunately too far gone) and scanning them in. While the sheer pixel count wasn’t quite there yet, the permanence of digital images over that of my poorly-kept film images took me to the point of never looking back.

The Kodak saw use right up until late 2004 when, while I was away on a business trip, my wife went crazy and splurged on one of those digital SLRs I’d been drooling over for years. Canon’s EOS Digital Rebel (300D) had been out for about a year, and Nikon’s D70 had just been released. Turns out that at the same time my wife went and bought the Canon with a couple of Sigma lenses for me, my brother-in-law had gone and bought the Nikon (for about $500 more). Seemed like a big deal at the time, although I don’t think either of us has regretted our choice in sides of that particular debate.

That Rebel saw battlefield use for the past three and a half years. Several seasons of soccer and softball, school events, work events, camping trips and hikes and road trips: it gave us better-than-acceptable quality, astounding convenience, and best of all let me practice those aperture-shutter-speed-ISO muscles I’d long forgotten.

Alas, it also had its faults. The Sigma lenses were never as crystal-clear as they should have been. ISO above 800 was unusable. The 3-second startup time seemed like hours at times, and the 4-shot-max buffer led to frustrating missed shots. Reviewing RAW images on its little screen was so slow it was almost not worth the trouble.

Each of those faults, though, was still a vast improvement over the non-DSLR market. ISO 800? Try ISO 200 max. 3 second startup? The Kodak took 5. The 4-shot buffer was only 4 shots because I was shooting in RAW, allowing me unprecedented post-processing possibilities (which I sometimes used!) All said, I wouldn’t forgo the low-end DSLRs from Canon or Nikon because of any of these complaints.

This week, the EOS Digital Rebel has finally met retirement. Joyfully, “retirement” is really “repurposement” in the hands of our eldest children. May they enjoy it as much as I’d enjoyed my hand-me-down K1000!

In it’s place around my neck on every hike, you will find the Canon 40D, equipped with the kit lens (28-135 IS) or a new zoom (70-300 IS), both from Canon, both of which I’d been eyeing for quite some time.

Canon 40D First Impressions

It’s probably premature to pass much judgement here, but my first impressions are:

1. The 40D is fast. Blazing fast. Flip the on switch and take a picture in less time than it takes to get it up to my eye. Writing images to disk is super-fast, even with the larger files it is writing. Reviewing images is sub-second fast, where the Rebel took 3-4 seconds to load each RAW image into memory from the card before displaying it. It becomes a tool rather than an occasional hindrance, which is an astonishing change.
2. I swear I can pull a lot more detail out of its shadows. Perhaps this is the 14-bit circuitry at work, or perhaps it is something else. Perhaps it’s even all in my head. Who knows?
3. Awesome detail on the high-ISO (1600) images. I haven’t yet tried 3200, but 400 is a good all-around sensitivity now.
4. Auto ISO. I have had more spoiled outings because I forgot which ISO I was shooting in and ended up bringing back all blurred picks or battling with over-exposure. Three years later, it’s rare, but it’s nice to let a computer remember the ISO for me instead (except when I want a specific ISO for a specific effect).
5. The fit and finish on this camera is astonishing, coming up from the low-end range. Gone is the slick silver (in my case) plastic and barely-rubberized grip. This camera feels great in my hand.
6. All that stuff I said about CRW format not holding geocoding data so I needed to introduce DNG into my workflow? The 40D uses CR2, which does geocoding just fine. Did a test round-trip of our hike around Eagle Falls next to Tahoe and all worked swimingly well without the DNG kludgy step.
7. The eyepiece is large and bright, astoundingly so. When lining up to time a batter in a softball game, I’d get it all lined up in the viewfinder, tense my arms against the monopod to make sure it didn’t move, then pull back my face and watch the pitcher send the ball flying so I could hit the trigger at just the right moment (give or take). With the 40D, when I pull my face away and watch the pitcher, I can still see the view through the eyepiece, and so I don’t have to hope I didn’t jar the camera any more. In practice, this means I can zoom in even closer on the batter, because there is less variance between the lined-up shot and the captured frame; there is an astounding difference in detail and impact of a head-to-toe shot of a batter and a eyes-to-waist shot of the same at the moment of impact!
8. Frickin’ high-speed continuous “motor”, with lasers!
9. I can leave the “time out” setting on it really low, knowing that a tap on the shutter button will bring it back to life with no delay whatsoever. In practice, I see this increasing my battery life far more than any circuitry-based savings ever could: it’s a workflow-level performance improvement, orders of magnitude more significant than fine-tuning operational improvements could ever hope to be.
10. The kit lens is fantastic as a general-purpose walkabout lens. The zoom range is large enough to handle most hiking situations, from medium angles along the trail to wide angle family portraits trailside to mid-level zooms of distant features (although I’ll still carry the 300mm for bird-sniping). So far, it seems crystal-clear in all zoom ranges, which is far more than could be said of the Sigma zooms.
11. IS is nice, but the verdict is still out on if the aperture trade-off is worth it indoors. See, my 50mm 1.8 prime lens is about two stops faster, just in aperture, than the 28-135 zoom at its widest (f3.5). The IS purportedly gives two full stops “extra” without blurring (meaning, you can have a quarter of the shutter speed with just as much noticeable blur), which combined with its wider available view should make it a better lens for capturing dimly-lit indoors scenes with “natural” lighting. The problem is that stylistically taking that extra stops in aperture leads to a popped-in-focus effect for your subject, which is often necessary for indoor shots (the backgrounds are almost always too busy indoors), and the extra two stops in shutter speed arrest the movements of your hand but not those of your subjects. Given a straight choice, I’d choose one stop aperture over two stops of IS steadying. That having been said, the 28-135 IS is a great stand-in here for the 28mm f2 I don’t own.
12. Reviewing pictures on the 40D is finally something I can do. I ran out of space (on a 4GB card!) at a game, and had to quickly get some space for the last few plays. Zooming out to the four-picture view I was able to quickly scan through the stacks to weed out the obvious losers for in-camera deletion. Generally, I’d strongly advise just keeping every picture shot and sorting it all out back at home when we have our senses about us again, but I didn’t have a spare card and needed to get more shots. In an emergency situation, I was more confident that the filtering I did in the field was a lot closer to what I’d have done at home than what would have been necessary on the Rebel’s tiny screen.
13. Oh yeah. This sucker eats through memory cards. Combine the near-limitless high-speed machine-gun trigger with the larger file sizes (12MB for the RAW, plus 3-5MB for the JPEG preview if you want both) and just plain increased shooting bliss, and the “CF Card Full” message comes quite unexpectedly. You will want a couple extra cards stashed away in your cargo pants for his occasion. Fortunately, the card-change time can be about two seconds with a little practice, as shutting the camera down and powering it back up are so freaking fast.
14. Back on the down-side: there is no cheap and easy Canon IR remote control for the 40D? Apparently mid-range camera buyers don’t want to fiddle with remote triggers. It’s not a great loss, as the EOS’s built-in IR was limited-range and fussy, but it’d be nice to at elast have a Canon-supplied alternative for wireless triggering of the shutter in family portraits. That having been said, I went searching at Amazon and found an alternative for $85 (via Adorama), then went searching a little more deeply and found the same alternative devices abounding at eBay for $25 each. I took a chance on the eBay’d version (apparently one company makes this device which a half dozen other companies rebrand), and will know if it is a satisfactory replacement sometime next week. Should have significantly better range and no line-of-sight annoyances (the EOS IR sensor would often be blocked by a lens hood, requiring contortions to trigger it). I’d link to eBay for the specific model, but eBay auction links time out too quickly. Just search for “canon wireless remote 40d”; I bought one of the “Aputure” RF-based remotes if you feel the need to walk in my purchasing footsteps.
15. Final downside: Canon lenses don’t come with hoods. I’ve never wanted to use a lens without a hood since the first hood I’d gotten. It’s, what, about 50 cents of plastic? Why not just include it with the lens? Oh yeah: so you can sell it separately for-40 bucks each. Highway robbery. There are third-party alternatives out there, but will they fit as well as Canon’s? In any case, this also found its way onto my shopping list with all due haste.
16. Odd feature: two “On” positions on the power button. The far “on” setting enables everything; the middle “on” setting disables the rear control wheel. I can not for the life of me imagine why you’d want that middle ‘on’ setting, or, if you really wanted to work that way, why you wouldn’t rather it a separate setting or button.
17. Silly fad features: Print from the camera, picture styles. At least I can repurpose the Picture Styles button (which is useless to a RAW shooter!) Unfortunately I’ll be tripping over that silly print button well beyond when anyone knows why it was put there in the first place.

Moving On …

That’s it. I’ve moved firmly up in the world, taking advantage of both general state-of-the-art improvements (many of the features of the 40D above are also in the “low-end” Canon Digital Rebel XTi 400D) and prosumer quality bits in one purchase (well, two, including the 70-300 IS lens). To say I’m excited about digital photography implies that at one time I was not. Still, I’m looking for places to bring my new toy, and expecting that my photo library will grow significantly in the next couple of months.

Perspective. Remember: cameras are tools we use to record our memories. Not even that: they are tools we use to record images which can in turn jog our memories. Geeklust won’t accomplish this task better. There’s a good chance that experience with a camera will yield better pictures than the latest greatest toy. More important than the camera are the pictures it takes; more important than the pictures are the memories they trigger; more important than the memories are the events which yield them. Get out and take some pictures with your camera, be that a Pentax K1000 or a Kodak Advantix.

Live life, make memories, then record them with the tool you have. In that order.

I have just recently started down the path of adding geocoding to my Aperture workflow. This doesn’t apply to all the pictures I take yet; I only bother taking the GPS tracker out with me when going on a family hike, not to a sports field or such. Below is a summary of what my geocoding workflow looks like today. Keep in mind that it is still somewhat fluid and will likely evolve over the next 6-12 months.

Why Geocode

Why geocode?

Well, let’s take a step back. Why take pictures at all?

For me, the root answer is an obsessive compulsion to document what I’ve seen, where I’ve been, and how I look at life. This compulsion stems as much from innate narcissism as from the desire to provide a history for my children and grandchildren better and more complete than the one provided to me. My parents kept scrapbook after scrapbook of pictures, but they were never organized enough to look through and never complete enough to tell a cohesive story. This, however, was a huge improvement over the handful of photos with cryptic captions their parents had preserved.

A picture in a shoebox can be a treasure trove of information, but often is missing key details. Who is that person in the front? Maybe someone wrote something on the back. When was it taken? Maybe the clothes or age of some recognizable face will reveal this. Where was it taken? Maybe there’s a landmark visible in the background. If these questions can not be answered, then the photo generally falls into the “art” category: its only value is as something to look at, not as a historical document. Art, of course, is great. But connecting artistic photos to real people and places and events brings them to life and makes them relevant.

My father, in his last years, spent many nights poring over sepia-toned pictures of his parents and their families, trying to identify people and places and form a cohesive story of where we came from. Most of those pictures ended up remaining in the “unidentified” pile, with us unable to determine who was in them or where they were taken or even approximately when.

A digital picture combined with a well-heeded workflow fills in many of those unknowns, and, with a little luck, will keep my grandchildren from going through a similar ordeal (should they in fact care about where our family comes from). Who is the person? If a keyword was attached to it, that might say. When was it taken? Look at the timestamp. Where was it taken? Maybe a keyword or location tag was added to tell us that too. What other pictures include this person? A click of the mouse and a few keys can summon them all. What other pictures were taken in that town that year? Another few clicks, a little more typing, and all is apparent.

Who is in the picture, and where it was taken, rely on how well I as a photographer and post-processer stick to my workflow. As an engineer, I know that any process relying on human attention to detail is bound to fail in some measure some times. So, where possible, I want to remove those manual-workflow-reliant bits.

So, on a macro-level, I want to geocode my images so that I can look back a few years from now and say, “That image? Yeah, it was taken on the Hidden Falls trail.” That much, I could capture in my workflow. What is highly likely to not get captured is, “It was taken just below the falls in the pools of an adjoining creek”. Geocoding my pictures (with an automated workflow) gives reliable answers to both levels of the “where” question.

So, back to the original question: why geocode? Fundamentally, it is to automatically and reliably infuse an image with data to answer the “where” question. This is one crucial aspect to the larger goal of placing each picture in its historical context.

Geocoding Goals

• Geocoding should be attached to the master file, not kept in Yet Another Side Database
• Geocoding should be as automated as possible
• Accuracy of geocoding should be verified immediately in the workflow

Overall Geocoding Workflow

First, let me review my overall setup.

For trail equipment, I take along with me my trusty Canon Digital Rebel (300D, the original model), while my wife and/or children will often have their own digital cameras, and a Garmin eTrex Legend GPS tracker. The eTrex takes on two roles here: as a track logger (pertinent to this discussion) and as a trail map/guide (very useful out on the trail, but doesn’t affect this workflow). A more limited-display GPS track logger would work just as well as the eTrex where geocoding is concerned (and we’ve used my wife’s Garmin ForeRunner as a track logger just as effectively).

At home, I have my Mac set up with Aperture, where I do all my photo editing and library management. For this particular workflow, Aperture 2.1 (the latest version as of this writing) is required; earlier versions would require the GPS log mapping step to take place prior to importing into Aperture, which generally meant it would not get done.

On the trail, I start up the eTrex, let it acquire the satellites, then mount it on the shoulder strap of my pack. I’ve found that it needs to be high and unobstructed at all times to get a good track. I’ve also found that I needed to change the logging settings to log “more often” to avoid “jumps” in the track log. Occasionally, prior to clipping it onto my shoulder, I’ll take a picture of the eTrex screen to give me a “clap board” to sync up the camera’s clock and the GPS’s clock. Of course, this is only to-the-second accuracy, but that’s generally good enough for the GPS tracks (unless, say, you’re snapping pictures while tumbling down a hill or somesuch). Generally, though, I just set the clock in my camera to match that on the GPS if it’s more than a second or two off. This pretty much only happens when Daylight Savings Time starts/stops, as the camera (thankfully) doesn’t handle that changeover by itself.

From there on, I just take pictures whenever and wherever I want. Note of course that the GPS is capturing my location, not that of the subject, and not even the direction I am pointing the camera. It’s also not capturing the specific locations of my wife and kids as they take pictures, which isn’t a big deal when we’re all together.

At home, my first priority is always getting the pictures into Aperture, getting a first cut pass through them, getting any gross adjustments made, and syncing them up with the Apple TV so we can all revisit the day while sitting on the couch downstairs. I’ve talked about that particular aspect of my workflow previously, as well as the nuances that come into the mix thanks to the AppleTV, so won’t go into those topics here. There is, however, one minor change (or major change, depending on your perspective) introduced by the desire to geocode the source images.

Canon’s CRW format does not allow for geocoding. By this I mean that it physically will not allow storage of the “Latitude” and “Longitude” EXIF tags. Kind of puts a damper on the whole geocoding source images idea.

The two options available, then, are:

1. Put the geocoding data in an XMP sidecar. Aperture doesn’t support this today, so for today at least, it is out. Sorry. Will revisit in the future.
2. Transform Canon Raw into something which does allow geocoding tags, like Adobe DNG. Now that Aperture supports DNG, this is a viable option.

I chose the latter. For this, I needed to install the Adobe DNG Converter, then put together a little script which locates all *.CRW files on my flash disk, copies them to the hard drive, then runs the DNG converter on them to produce a separate directory full of *.DNG files; this folder is then, by the script, loaded into Aperture as a new project.

Of course, all of the above could be done by hand instead of by script, but being a highly repetitive bit of the overall workflow I see no reason to not automate it.

When I get around to it, I attach the eTrex to my computer using a Prolific USB Serial Port (drivers available for OS X at Prolific’s site – click on the md_pl2303… file), then run a second script which asks for a name for the hike (optimistically defaulting to today’s date; I typically correct the date then add a hyphen and the name of the trail we hiked) and runs LoadMyTracks to download all the logs to my track logs folder with that name.

Next, I load up GeoPhoto and open a new “scrapbook” for this trip’s photos. I tend to name this identically to the track name I gave the LoadMyTracks script, using copy/paste.

The next step is to load the Aperture-managed .DNG files into GeoPhoto. This isn’t so easy, though! We need to get the Aperture-managed Master files, not the Aperture-generated JPEG previews, so we can not just drag and drop from Aperture to GeoPhoto and expect it to work! To get the master files, I go into Finder (or PathFinder), use Show Contents on the Aperture Library, then open the blue folders to my project (“2008” then “05 May”), then Show Contents on the project package. From there, I do a spotlight search to get the list of all .DNG files below. Finally, I select all the found .DNG files and drag them over into GeoPhoto, into the prepared album.

An enterprising soul might notice that if one were to just drag the image from Aperture to an application or script, the file which gets loaded is “…/Previews/SomeVersionName.jpg” (as evidenced by dragging a photo from Aperture to Preview). The master file is the .DNG (or .CRW or .JPG) file at ../. Surely, this information could be used to select the master files to drag into GeoPhoto or the like!

Inside GeoPhoto, I select all the pictures in the album, then use the “Map Items to GPS Track” option off the Item menu. In the resulting dialog, I select the saved .gpx file from LoadMyTracks, then select Manual timestamp correction (none), then select the timezone of my camera (Americas/Los Angeles), confirm that all the images got mapped, check the “store location” checkbox, then click “okay”. Maddeningly, then I need to re-select all the images and use the “Save locations to original files” option.

Once geocoding information has been added to the .DNG files, I go back into Aperture and update it’s database by using the “Update EXIF from Master” menu option (under Metadata).

Voilà! Now, clicking on an image from Aperture, the “Show on Map” option is available, and loads the image up in Google Maps. From here, a host of tools is available, including Google Earth and Flickr synchronizations.

Geocoding Application Alternatives

I use LoadMyTracks to copy my GPS data from the eTrex to a .gpx file in a file hierarchy. Another popular tool is GPSBabel, although I could not get it to reliably download from the eTrex. I am sure the fault is user error there, but with a much more usable free alternative (LoadMyTracks) I don’t see the point in trying to find out what I was doing wrong! Note that other tools using GPSBabel as a library (HoudahGPS and GeoPhoto) likewise failed to get tracks from the eTrex, so perhaps the library just doesn’t really support eTrex and/or my USB-Serial adapter.

I have settled on Ovolab’s GeoPhoto as my track-matching tool of choice. I also looked into HoudahGeo, which seemed just as good (and avoided some of the GeoPhoto usability quirks). I don’t find the embedded Google Earth clone compelling, although I do appreciate the (fleeting) view GeoPhoto gives me of the imported track overlaid on a map (HoudahGeo only indicates which pictures overlap with the track, giving no way to know if they matched properly). At the same time I was put off by GeoPhoto’s insanely crippled demo which doesn’t allow testing of the single most important feature of the application: how well and compatibly it writes geocoding information to the files. I almost didn’t buy it for this reason. However, thanks to a Euro-US$ 1:1 conversion in their pricing structure, GeoPhoto was almost half the price of HoudahGeo, so I went with GeoPhoto. Not exactly a ringing endorsement, and I have quite a few problems with GeoPhoto overall (see the Wishlist section), but do note that all those nits so far as I can tell also would have applied to HoudahGeo.

As glue for the repetitive tasks, I have two scripts.

The first script is used to import photos from my card, encode them into DNG format, and import them into Aperture.

The second script is used to pull the log files from my GPS tracker into an organized folder structure on disk.

The real fun, of course, comes in viewing these geocoded images. For that, I use Google Earth.

Wishlist

Geocoding on the Mac, and especially when trying to keep the geocoding information attached to a permanent Master file rather than a fleeting exported version in Aperture, is very rough. I am told that smoother workflows exist on the Windows side of the fence, which is a shame because so much else about organizing and editing photos is so much better on the Mac side.

Aperture Wishes

• Keep project-level non-photo information. Specifically, give me a place to put the damned .gpx track log! Bonus points for specifically handling the .gpx attachments to a project below, but a general “project-level data store” mechanism would be highly useful in many many ways!
• Support geocoding internally. Really. Embed the likes of GeoPhoto right into Aperture, or even just “blind” track log matching.
• Alternatively, better integration with geocoding tools like HoudahGeo or GeoPhoto.
• Alternatively, allow me to drag a reference to the Master file directly from the interface to the likes of HoudahGeo or GeoPhoto.
• Better integration with a visualization tool (ie, Google Earth or Google Maps if you must). Bonus points for being able to click on a handful of versions, click a “Show in Google Earth”, and have the .gpx track loaded into Google Earth along with the photos!
• Better location-based searching, such as “all pictures within the boundaries of Sacramento County, CA”, or “all pictures within 2 miles radius of Lake Tahoe”.

GeoPhoto Wishes

• Better integrate with Aperture. Bonus points if you could take a .jpg dragged in and automagically translate that to the matching master image (go up one folder and find the .DNG or .CRW etc file).
• Support Applescripting. Every one of the complaints I have here could be remedied in Applescript, if only you supported it!
• Remember the time zone I selected for my camera. This is the one item which would not apply to HoudahGeo. There is no reason at all, given the fact that every single one of my photos has “PDT” set as its timezone, that you should have me pick from a 1,000-item drop down which city best represents my location. There is certainly no reason why I should have to do that every single time I use the tool!
• Support “pasting” of images from Finder. I can only drag images in. If I could paste them, then I could put together a quick applescript to put the master file references on the clipboard.
• Allow information to automatically be written to the original files once the “match” sheet is dismissed. There’s no reason I should have to tell you again that I want that done. I know, I could set the option to do this when I quit the app, but I like to make sure everything was written and readable by Aperture before clearing an app from memory.
• Rethink the UI to be more workflow-based (or allow for a workflow-based view). It’s a nice interface for viewing photos on a globe, but it really sucks for actually coding those locations. HoudahGeo’s UI is more along the right lines there, although its inability to show the track we are matching up to counters the workflow benefits.
• Include elevation data (altitude) when saving latitude and longitude.
• If an image already has geocoding data and is added to your database, use it’s location information!
• Fix the stupid, short-sighted, crippled trial demo which keeps us from actually trying GeoPhoto in any geocoding workflow. Anyone buying your product is buying it completely based on the faith that you actually are capable of writing EXIF data out to their files and that this will work in their imagined workflow. Imagine how many more customers you’d have if they could verify your product worked before shelling out their money!
• Give error messages when writing to files doesn’t work. After buying GeoPhoto I thought it just didn’t work because it wasn’t writing lat/long to the CRW files. It wasn’t until I tried the same in HoudahGeo, which gave a meaningful error message that I learned of the CRW file limitation. Driving your customers (who buy your product on vague descriptions and good faith) to your competition to figure out what went wrong is not good!
• Get the forum your web site talks about up and running.

Geocoding Wishes

• Add “accuracy” information to geocoding tags
• Somehow, allow the location not only of the photographer, but also of the subject(s), or at least the directionality of the picture. Often, these are close enough to not matter; sometimes, though, the picture is of a distant mountain or a sunset over the water.

You need the mixture.

There is no shortage of nuts and seeds in software development. These are the tasks which provide non-sexy baseline features for users, which build frameworks for future development, and which fix past sugar-induced errors in judgment. Looking at a good bag of trail mix, nuts and seeds are at least half of the volume; the same should be true of a good engineering project.

Then there are tasks which provide an immediate and visceral thrill, the sugar-rush of getting something out that just makes you smile. For some people these are the bits of flash that users look at and gasp. For others, it’s getting .01% better throughput on a database transaction. These aren’t necessarily the bits that will make the application work for your users, but they are fun, challenging, and make you remember you are not a mindless automaton in a corporate machine.

In between these lay the raisins. These are the deeper architectural problems which feel so great to solve but which require weeks, months, years of slogging through the morass of data and hitting dead end after dead end before finally hitting upon the solution which, in hindsight, should have been obvious from the start.

For a successful product, you need engineers hitting all three types of issues, and knocking them all out of the park. For a successful career as an engineer, you need to play all three parts, although you might find you favor one over the others. As a product manager, if you are heavily favoring one of the flavors over the others, you will soon find that your engineers yearn to work on more/less exciting/challenging products (and will be leaving your company in droves if your company can not provide that for them).

Look at your product. Look at your work plan for the next year. Do you have a healthy and sustaining mixture? Is it overly sweet, with lots of glitz and glam but no substance? Is it chock full o’ nuts, keeping everyone satisfied but no one happy? Is it raisin-heavy, overloading your engineers with intractable issues while delivering nothing to your customer base?

Again: You need the mixture.