How I Created a Matrix Bullet Time-Style Rig With 50 DSLRs | Hacker News Reader

How I Created a Matrix Bullet Time-Style Rig With 50 DSLRs

petapixel.com

128 pointsuptown54 comments

I wonder if the comments from "The Guy Who Knows" on the article page are real, or just some very random trolling.

The Guy says:

Well... Braging of you success, right? Guys, I work for the company who ordered this Matrix thingie... Martin Legeer designed this and built this first version. What this article doesn't say is that it DIDN'T WORK AND WE HAD TO COMPLETELY REBUILD THE WHOLE MATRIX. The cables were poorly soldered, when you touched them you got hit by electricity, the place was too small for its purpose... We had to completely demolish it and build it from scratch, this time with guys who knew what they were doing. And FYI you are not authorised to use these pictures Martin, so be as kind and delete this whole article and stop boasting with something you failed at!!!

assuming the author of the article intends this to be useful to e.g. get potential clients, that looks like a comment to be deleted or at least replied to.

Also interesting that the angry commenter blames the OP for building this in a too-small space, I got the impression the space was simply part of the deal, and not chosen by the OP. Weird.

That would at least explain the absence of sample shots from the rig.

http://vimeo.com/55566188

Those are great, stunning even, but they are not from that rig.

Someone commissions a hugely expensive rig like that to be installed in a random garage? Bizarre. It must be for porn.

my initial thought was same. "So what does he plan to do there?"

I don't see how or why. The enclosed space is too small, and what advantage is there to be gained from having a 360° still frame of bouncing boobs? The rig is not even designed to capture motion.

Too small? o_O

No. White background without shadows means the images will be used as product images on websites. He will most probably produce interactive turntables of living models for online clothing stores.. Or he is planning his own clothing store or clothing preview app for iPad, whatever. That's the only way it makes sense as an investment.

Then why install it in your private residence? that's what office space is for.

You can't make porn in a rented space?

You can, but you might not want to for reasons of privacy (if one wanted to record one's own intimate moments, say). On the other, there's no particular reason to use your own home for the mechanical business of product photography.

Aww... I would love to see the shots that this thing took.

EDIT: Found some of a similar rig they built in the comments of the same page: http://www.petapixel.com/2012/12/17/project-uses-a-bullet-ti...

Awesome, thanks. At least it gives me an ending to the write-up!

It seems that in that rig, without an enclosure, they are limited to shooting in the dark using light painting lest they capture the background. In this "shadowless" rig with the overhead lighting and white background they have more options. He even mentions the possibility of painting the inside which makes me think they might be thinking of doing a green screen. They do have to worry about the camera holes which means they'll probably have to shoot with an upward angle.

Wow that looks straight out of Mass Effect!

They have/added a link at the bottom: http://www.petapixel.com/2012/12/17/project-uses-a-bullet-ti...

The images are pretty cool.

There was this Amazon scam where the scammer impersonates an Amazon user who bought Canon DSLR, and asks for replacements to be shipped to some other country.

Exactly two days later, we see a post where 50 Canon DSLRs were used to create a movie effect. Funny coincidence ;)

He does commercial photography. Here is a link to his website: http://www.martinlegeer.us/photo.php

No, this isn't for porn, it's probably for fashion photography.

And to anybody who thinks this is a "lot" of money for fashion photo...a common camera for doing fashion photography costs $35,000+. This is a different niche, but in the same realm of dollar figures.

(If you're curious what a truly "high end" digital camera looks like, google Hasselblad, Mamiya and Phase One.)

> anybody who thinks this is a "lot" of money for fashion photo...a common camera for doing fashion photography costs $35,000+

the money spent on hi-end fashion shoots isn't in cameras. leased/bought equipment works for itself hundreds of times, and when you do the math for cost per shoot, the camera would probably be one of the cheapest items on the expense list after coffee.

> No, this isn't for porn, it's probably for fashion photography.

ummm, more like packshots.

Why would you need a bullet time setup for packshots? The only scenario I could possibly think of would be if you were selling animals.

the only kind of studio photography that comes to my mind where you don't work with your legs, keep camera in a constant position and don't want any drop shadows is packshots. the setup is basically a huge light box with more cameras. probably for some crazy demos.

I can't imagine why somebody would shoot people with 50 cameras constantly leveled up on the same height. maybe for one experimental project, but what next.

Maybe they want something shot with the product falling in midair, or a model with a fan blowing his/her hair and clothing? Or it could be some appliance/machine that they want to capture in action (a popcorn popper would be a mundane example).

Heck, maybe they just want to shoot a product sitting in the middle of a cloud from a fog machine.

I'm bit naive at cameras and thought best ones were being produced by Nikon or Cannon so this new brands are enlightning. I was just looking at one of the Hasselblad: http://www.hasselbladusa.com/media/2081132/uk_h4d-60_datashe....

This thing has massive 54x40 mm sensor VS current full frames which have "only" 36x24 mm. That's 2.5 times bigger sensor than full frame! And that's 60 mega pixels. One thing that pops out is ISO which is listed only at 800 (vs Canikon's 32000) so I assume these cameras are designed for very bright light studios and to yield minimal amount of noise possible.

high iso is for where people move - journalism and sports. model shots are low iso, because you use lighting anyway (natural or/and artificial) and the quality needs to be perfect.

> This thing has massive 54x40 mm sensor VS current full frames

because it's a medium format camera which by definition is bigger than full frame.

Indeed. There's the world of 35mm cameras, which is what full frame digital SLRs are (think Canon 1D X and 5D III, and Nikon D4 and D800, having a sensor fo 36x24mm). Many digital SLRs have cropped sensors — the Canon 7D, 60D and 600D have a sensor that is 1.6 times smaller than a full frame (22.2x15.8mm). More: http://en.wikipedia.org/wiki/Full-frame_digital_SLR

The next level of cameras is medium format, which are vastly more expensive and have much larger sensors. They're generally used for studio work, where light isn't a problem. They're popular for fashion and product photography.

Then you move up to large format cameras with even larger sensors, although most of these still use film (but digital backs are available, and tend to have hundreds of megapixels). Large format cameras are still popular for landscape photography.

Great writeup, but utterly incomplete without a demo.

Interesting to see the future of Bullet Time, too. RED camera recently did a shot that they can't yet talk about, but the setup pictures of 50+ RED Scarlet cameras is very cool.

http://www.reduser.net/forum/showthread.php?87124-Scarlet-De...

I thought this line in the article was pretty funny: "There were potential programs we could use, but they were written for Linux." ><

yeah. really suprising how someone who builds a 50 camera rig can see installing an OS for free as an obstacle.

> We eventually found a single program for Canon DSLRs, which turned out to be a huge waste of money. In the trial version it had a limited options, so we did have to buy the full software.

karma.

Perhaps my ignorance, which is likely, but isn't this pretty much the exact same way it was first done? I vaguely remember, at the time The Matrix came out, some artist showing off a sequence done exactly like this capturing molten metal sparks from a anvil, or some thing visually similar. Makes me wonder what the big deal is here 13 odd years later.

Here's a very similar thing from nearly 20 years ago. I remember watching this as a kid and it blew my mind. Distinctly remember the dog: http://www.youtube.com/watch?v=fIpmUi8HI1k

I'm no expert but it strikes me that the shadowless full-light exposures that he talks about are the real challenge here and maybe that sets it apart from other popular examples.

All the commentary here and no one has mentioned Google's effort to automate a similar task: building 360° rotations of children's toys for holiday shopping. [1]

[1]: http://www.google.com/shopping/product/4502510368537277191?p...

That's really cool. But if it's a static object then surely you use one camera and a fixed speed turntable and just shoot a continuous stream of shots.

Which has me thinking, can a DSLR take images quick enough that you'd only need to have a rig that orbits the camera around the object at high-speed in order to be quick enough to capture models. The speed needed would be too great for moving subjects I'm sure.

I've spent a bunch of time thinking about ways to do something like this on a budget. The general idea would be to have a bunch of bare camera sensors directly connected to commodity PCs (maybe with Arduinos interfacing between them) with each PC driving multiple camera sensors.

The bandwidth requirements are quite ridiculous when you want to take a hundred photos simultaneously, so I'm not sure how many sensors I could really drive with one PC.

Not sure what you mean with "bare sensors", a sensor won't do you much good without shutter, lens, and a way to control the aperture?

To go cheap, I'd replace the DSLRs with webcams. It sacrifices resolution (obviously), but reduces bandwidth and computation requirements, and lets you play with the effects until you get a client who can foot the bill for a high-res version.

You are very right. I really meant "the parts of the camera minus the embedded system that drives the whole thing", but that wasn't very succinct. :-P The application I was thinking of was indie movie making, so I would want a decent resolution up front, and I wouldn't expect a high-paying client to be forthcoming.

Along this tangent, I talked today to a friend of mine who is convinced that I would be better off using some of the lower-end consumer point-and-shoot cameras, or even purchasing old film cameras.

Will you shoot video with the cameras? Or stills?

That question is about time lapse with a single camera, but it includes some camera suggestions, including the GoPro action cam and Canon P&S with the CHDK hack: http://chdk.wikia.com/wiki/CHDK

I agree with your suggestion on the webcams. Another approach is to use machine learning to generate the high-definition image from the mutual information contained in the multiple images by using probabilistic estimators. As an example, [1] has empirically shown decent results using MAP estimators.

[1]: http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.130...

Interesting, thanks for the link.

I couldn't find anything in a cursory search of google and nothing in any of the links currently here explain it.

Why is it 50 cameras for the effect? Each link I've seen has the same (50 cameras). I am just curious if there is a reason that the number is used instead of say 60 (every 6 degrees) or something like that?

Edit: It would have been great to see it in action but there is a link elsewhere on the page that had some awesome views of what can be done.

The last thing I have to say will be a disappointment to many of you: I don’t have any resulting images to share with you due to the client’s request to not share any sample photos outside.

-- Why so top secret? But blog the technique...?

Usually its the other way around. "awesome image, how'd he do that"?

I'd guess that it's a technical feat, but not necessarily a new idea or new technique. There's no harm in sharing the how in that case.

The client didn't share sample photos ilkely because they want to keep things quiet about their product/art till launch.

or perhaps because as the commenter posted on PetaPixel, it didn't work...

I've seen a rig similar to this used for capturing images to be translated into 3D objects. No shadows would definitely help with the processing and it wouldn't matter that you couldn't capture video with it.

BreezeSys makes software that can do this.

http://breezesys.com/MultiCamera/index.htm

I wonder if that's the software that the chap tested.

It irks that a writeup for a technically clever and elegant setup for stunning visuals sits next to such a mess of irritating, flashing ads.

Just in case you were wondering what kind of shots this rig can produce....

http://24x360.com/gallery

This would be more impressive if you didn't know that the original 'frozen pan' rigs were jerry-built from film SLRs.

What was the budget to set this thing up?

He's using 600D (which would be my choice as well for that kind of setup).

The cheapest one I can find is $439.

$439 * 50 = $21,950

Then lenses. Not sure which ones he uses. One of the cheapest lenses is 50mm/1.8, and it's a pretty good lens. They cost around $100.

$100 * 50 = $5,000

Then add a could of hundred for wiring and connectors.

Then add a few thousand for building the structure.

All in all, should be under $30K.

He seems to be using 20/2.8 lenses, which seem to be at $500 at amazon, so total would be closer to 50k than 30k.

So the exact same way they did for The Matrix?

Not exactly exactly but yeah, the same idea.

Reply on news.ycombinator.com