Robotic Photographer 107
Boojum137 writes " ExtremeTech has an article on a robotic wedding photographer named Lewis. Lewis's hardware stats are modest, so he uses a clever trick to locate people based on skin tone, which is also independent of race. After locating potential subjects with a myriad of sensors, including sonar, laser range finders, and infra-red, he tries to frame the shot according to photographic rules of composition. But the real innovation behind this "red trashcan" is its ability to fade into the background. According to Lewis's creators at Washington University's Media and Machines lab, because of the robot's passive nature, people tend to ignore it after short period of ogling. This allows for some great natural shots, instead of the typical forced and self-conscious shots from human wedding photographers. And, in case you were wondering, Lewis is going to live up to his name in November."
Easy way to get good wedding photos (Score:1, Informative)
But the funny thing is (Score:1, Informative)
It's nonsense to say that wedding photographs have to look posed. Many photographers are experienced at taking candid pictures - that's what street photography's all about.
As a former wedding photographer, (Score:5, Informative)
Problems? You want problems? We got problems!
1. It's too short. Candid shots from that level, looking up at the subject, are not flattering. And people don't buy wedding photographs that don't flatter them.
2. It's digital, I guess, but in any event uses some kind of video camera to capture images. That's (probably) crap. Still captures from video cameras are of lesser quality than a good digital camera and good digital cameras (that is, "good" by pro photo standards) are as expensive as your house. Weddings are one of the few times in their lives that most people will actually pay good money for high quality photographs. Digital has its place at the low end, but is something as expensive as this is sure to be going to be cost-effective at the low end? I kinda doubt it. Good wedding photography still requires film, preferably nice, wide rolls of it. Show me a machine that can operate a Hassy and we'll talk again.
3. Yes, adult humans tend to ignore something like this after a brief period of familiarization. But there are lots of kids running around at weddings. Ever show a 4 year old the hamster dance page? After 30 seconds, you're bored but they're just getting started. They'll giggle for hours. Same principle here. Just wait till a few unattended kiddos (and there's always at least a couple of kids at every wedding whose parents are nowhere to be found while they tear a path of destruction through the place) notice this thing and decide to play "Let's push over R2D2!" with it. It'll happen.
4. Good wedding photography (Wait - this comment might not be a good one since we've already established that this device is only worthwhile at the low end - but I'll go ahead anyway...) requires making art (sometimes hack art, but art nonetheless) quickly. That requires aesthetic sensibilities and brainpower that this thing just doesn't have.
One last note - I can understand the concept. There are WAY too many wedding photogrpahers in this world with gigantic egos who act like a wedding is a get-together for the purpose of taking pictures. They take over and try to run the whole show. After dealing with them, I can certainly imagine being motivated to invent a machine that would just shut up, do the job, and not get in anyone's way. But that's a by-product of pain-in-the-ass photographers, not really a good reason to develop a new machine. A truly good photographer knows how to be totally unobtrusive 98% of the time. The solution is to not hire bozos. The solution is not to try to replace photographers, even if it's just a few at the low end, with a machine that will necessarily produce substandard output.
This is a good thing (Score:2, Informative)
What it does show is that it is feasible to do more work in this area - despite its size, people lose interest (although I would imagine this is more to do with wedding-associated alcohol than anything else...) and let it click away without ogling at it - all it is is fairly simple technology appliying simple rules (e.g. face should be in centre or photo or wherever) and then takes a few pictures.
While a lot of these will be crap (same as for a human photographer) what this does allow is more natural photos, where people are not posing, or forcing smiles, or having to hold their smile just a bit too long as the cameraman waits for whatever moment all cameramen seem to thnk comes about 5 seconds after asking everyone to say cheese (don't you hate that?).
So we get lots of natural photos, discard the rest, and end up with some good ones which we might not have achieved with a human.
In the long term, though, this might have much more application - we can get it to recognise terrorists, criminals, finally get the intrusive Minority Report style advertising, and maybe we can train Lewis to recognise cleavage.
I've seen/played with it... (Score:5, Informative)
The photos it "composed" were actually quite good as it would use the rule of thirds when trying to set up a photo. The only real problems I saw with it from a tech standpoint was that the lighting in that area of the convention center was a bit yellow so their white balance was off and the robot had the embarasing habit of totally ignoring anyone with very dark skin. At the time I was talking to a student from a local school who was there who happend to be black and we let the robot take a few pictures of us... when we went to look at the pics every single one of her was cut off as she was basically being ignored as "background".
The other problem were the batteries - I saw them change the batteries on the 'bot once and there were a *LOT* of what appeared to be very heavy lead -based batteries - they looked like motorcycle batteries, but I'm not sure. Apparently the bot lasts for a few hours on a single charge, but then you have to swap out all of its batteries (I think there were at least 5 of them)
neat but primitive.. (Score:2, Informative)
next most wedding receptions are at dark places... so without the ability to flash (as I highly doubt the low cost video camera is a 3ccd 0.05lux color array with plossl lenses costing around $15,000.00 (for the CHEAP one.. Ours was $35,000.00 but we wanted Zoom)
It's a great proof of concept, and with some upgrades it can be a really cool device. but it needs to get away from video for the camera and use a regular Off the shelf 3megapizel digital camera with flash or 35mm film to do the job.. I certianly dont want my wedding photos to be nothing but low rez,underexposed,blurry NTSC captures..
A few more details (Score:5, Informative)
So I'm a student at Washington University, and I know the people who programmed Lewis. A couple of points:
Anyway, it was a lot more work than you'd imagine to get Lewis to function properly. Lots of things you don't think of crop up (The laser's can't see the legs of a table. Micheal is pale and other people aren't. Wooden doors are approximately skin toned.), and the problem is intrinsically difficult. The skin tone stuff alone eats up most of the processor, nevermind the path planing and mobile obsticle avoidence. For SIGGRAPH he was running on reduced hardware too: he has a dual mobo, but it isn't as reliable as the singe-processor one used at SIGGRAPH.
Technical details, clarification (Score:2, Informative)
As a member of the team working on the Lewis project, I'd like to provide some additional technical details. It should also be noted that the Lewis project is not intended to replace human photographers. It's an easily accessible research-oriented endeavor to explore human/robot interactions in a real-world environment.
Specifications:
Processor - Lewis is a B21r mobile research robot from iRobot Corporation [irobot.com]. It's powered by a single 800Mhz Pentium IV processor. This CPU must handle all of the motor drive and low-level robot tasks such processing the data from the large array of sensors. On top of this CPU load is the task of finding faces, navigating crowds, and taking and processing the photos. The two additional processors to be installed in the future will allow Lewis much more power for its photographer duties.
Camera - Lewis currently uses a Sony DFW-VL500 [technical manual [sony.co.jp]] digital 1394 (Firewire) camera. This has a 1/3" CCD that produces 640x480 color images at up to 30 frames per second. Image output is YUV 422 format and is not compressed. The built-in 12X zoom lens is sensitive to 14 lx (F1.8). Higher-resolution 1394 cameras are available, but these do not have built-in lenses; this is bad because focus, aperture, and zoom must be fixed.
Safety - The entire enclosure is lined with bump-sensitive panels, so that if the robot runs into anything, the currently executing program is terminated, the motors are halted, and the brake is applied.
Operating System - The operating system on Lewis is a standard Linux distribution using kernel version 2.2. Various libraries for control of the motors, sensors, pan/tilt unit, and camera are used.
A couple of other comments: the camera is not an NTSC video camera. It was chosen because of the easy ability to control zoom, focus, and aperture from software. Since our goal at the moment is not film-quality pictures, this camera suffices.
Sample photos are available on our website [wustl.edu]. We have been slow in posting samples due to privacy concerns, not because the pictures are bad. We have over 3,500 photos, and I'd say well less than 2% are false hits -- photos of doors, walls, elbows, etc.