Hardware for camera array trigger

30_08 wrote:

Hello!

I read thread about "bullet-time effect" (https://www.dpreview.com/forums/thread/4031596) and i had a question about this. Which hardware should be use for remote camera array in time? Simple USB cables bad because it can delay cameras about 1/10sec. BreezeSys advice to use TriggerBox by Esper for avoid delay between cameras. It's good, but maybe there are others solutions for this?

General question - what are the hardware solutions for remote cameras array?

There is no general answer. It depends on the cameras used.

Canon PowerShot cameras that are supported by CHDK can sync very precisely (to about 1/1000s, sometimes better) by triggering when they see 5V applied to their USB.

Lots of older cameras have a three-conductor wired remote that works as a ground and half-press and full-press; you connect ground to half-press, wait a bit, then also connect full-press to ground to fire the shutter. That's what BreezeSys is talking about here .

Basically, the issue is that most cameras have a little lag as they do the half-press (focus and set exposure), and normal tethered-USB operation asking for the shutter to be pressed can suffer that delay. The actual USB delay can be pretty small, and for USB2 the communication protocol delay is less than 120us (i.e., 0.00012s), so it really just depends on how the camera handles the shoot request. With Canon PowerShots under CHDK, you can actually control things well enough that you can trigger quite precisely using an actual USB message (rather than just detection of 5V), but then there's the scaling problem that USB commands to multiple cameras sharing the same USB connection will get serialized, adding a little delay... whereas any number of cameras can detect the 5V simultaneously without any scaling issue.

The other option is to do sloppy sync, but flash the light -- dark room, long exposure, fire a strobe to capture from all cameras at the same moment (usually about 1/1000s duration for a strobe).

I've recently been working on a solution to this problem. I built a device that can control a 13-camera array to do bullet time, as well as genuine simultaneous due to how I set up the circuit. I also built an iPhone app to control the device remotely. I built a really quick website with some information, but I also work full time so it's a work in progress: https://shutterdeck.com

I am looking for test users who can give me feedback on it, and I'd be happy to let you try it. Also happy to answer additional questions about it here.

shutterdeck wrote:

I've recently been working on a solution to this problem. I built a device that can control a 13-camera array to do bullet time, as well as genuine simultaneous due to how I set up the circuit. I also built an iPhone app to control the device remotely. I built a really quick website with some information, but I also work full time so it's a work in progress: https://shutterdeck.com

Excuse me, but how can you claim "microsecond accuracy"? Conventional cameras cannot trigger that accurately even if you make the wired remote connections at the exact same nanosecond. I'm assuming you are using the 3-wire remote, right?

I am looking for test users who can give me feedback on it, and I'd be happy to let you try it. Also happy to answer additional questions about it here.

Don't have many cameras supporting the 3-wire remote any more. Have dozens of Canon PowerShots that can trigger using detection of 5V on the USB connector....

Hi, pardon some of the language on the site. I threw that up quickly more as a portfolio piece. It's not a mass market product. I'm a photography hobbyist and designer.

What it means is that the device triggers the pulse to all cameras simultaneously, rather than slower ways via usb hubs, etc, where pulses to a camera array are sent 10s to perhaps 100s of milliseconds apart at unpredictable intervals. So of course cameras in an array connected to my device would still be subject to the small mechanical differences in each camera, if that's what you meant. But the signal that tells each camera to fire is sent to all cameras at the *same exact time*, unlike other solutions I've seen out there. And if you see the sample images on the site, you see that the results are pretty good. Those examples were shot with an array of 10 Nikon D80s.

You might be right that not all cameras support 3-wire remote anymore, but I've been using Nikons up to D610 and and we have not had support issues. What I aimed to achieve with this version is support for a lot of cameras, not necessarily the very newest cameras.

While perhaps not a panacea, I wanted to create something that would be easy to use to people who need this kind of shot right now. But I appreciate your feedback and I'm happy to answer more questions about it. I would certainly love to support all cameras, but at this point I'm just one guy with limited resources to put into it.

shutterdeck wrote:

What it means is that the device triggers the pulse to all cameras simultaneously,

That's what I thought, but you need to say it that way because actual sync is a lot less precise. It's very difficult to get cameras capturing +/-1ms; +/-1us never happens.

You might be right that not all cameras support 3-wire remote anymore, but I've been using Nikons up to D610 and and we have not had support issues.

You should take a look at the various sites talking about KAP -- Kite Aerial Photography. They explain all the different ways you can trigger and give circuits to do it.

It's literally a decade old (has it really been that long since I updated it? D*mn. It has.), but there is also my pages on Capture Control Of Digital Cameras .  Probably most people triggering fleets of cameras are using piles of Canon PowerShots with this 5V USB detection trigger ... we certainly do in my research group (I have a paper submitted to Electronic Imaging 2018 overviewing a bunch of our camera clusters, but I'm still waiting to hear about acceptance).

The more recent change is that many cameras now work quite well via either USB or 802.11 wifi tether.  Typical USB uses a slotted protocol based on 120us intervals, so latency to signal the camera is less than 240us... and that's significantly less time than it usually takes for the camera to do it's thing. Typical operating systems inside cameras use a scheduler jiffy of 1/1000s (1ms), so that's about as fast as they can respond. Of course, focus, etc., takes much longer, so you have to prefocus and set exposure (like the half-press signal on 3-wire) long enough beforehand so that it's really ready to trip the shutter -- otherwise, you'll wait random amounts of time for that.

That's really interesting, thanks. I'll take a closer look at that info over the next day or so. When I started working on this project, I did it for a couple friends who have a little photo booth side project. Using an Arduino with all the cameras connected to the same input line worked just fine to trigger a pulse simultaneous, but they wanted to do more. They wanted to be able to do linear sequences too.

The easy way of doing both on an Arduino means still firing them several microseconds apart at best, enough elapsed time such that if a fast strobe light were plugged into the circuit and was fired first in the array, the light would change by the time it got to the end of the array. So I built a more complicated circuit using a different chip.

The device does more than just triggering a pulse simultaneously. It also allows you to set them off in sequence at whatever interval you want, or individually on an ad hoc basis if desired. It hit all the needs my friends needed to get going, and since then I've worked on it myself, building the iPhone remote control app. It is set up to be portable and run on 5V power, so you don't need to have it plugged into a laptop to use it.

I'd love to take a look at your research paper when it comes out.

Hi Mike,

Yes that is a great idea. I've thought about it a bit. My initial thought was to open up another 3.5mm input that just responds to high signals, so that the device could respond to just about anything, but it would be particularly great for Arduino tinkerers who want to trigger cameras in response to sensor readings, etc. I think it'd work well for stop action shots. I've purposely left a few I/O lines unused on the chip specifically for potential use based on feedback like yours.

It'd be easy to do, and I could easily add support for it on the pcb and chip. The main reason I have not added it to the version you see on the website is just to keep things simple and gauge interest of a minimal set of features. My hope is that there are enough people out there right now who need the features I've already built and who can validate the idea, after which I'd tackle the things you've described.

But we're totally on the same page. I've built out the mobile app as a 'lite' version of a yet-to-be-built desktop version where the user would have much more granular control of the device. A desktop version could connect to the device via serial USB as well

Thanks for your response, I really appreciate it!