Complex DSLR system for scientific video -- thoughts on my plan?

I'm getting ready to purchase a pair of cameras for scientific video analysis of freshwater fish behavior. I've done as much reading as I can to put together a suitable system, and would appreciate the opinions of some professional underwater photographers about my plan and anything I might have overlooked.

I'm going to use a stereo pair of video cameras, placed side-by-side on a crossbar on the bottom of a small, shallow, clear, freshwater stream, pointed at a drift-feeding fish (that is, a fish that maintains one semi-stationary position in the river and darts back and forth from there to intercept potential prey from the current). I need to be as close to the fish as possible, with the best sharpness possible, to resolve subtle details (for example, to see whether it "spit out" an item it just captured), yet I also need to have a wide enough view that almost all the fish's prey capture attempts are on-screen in both cameras. The fish will be anywhere from a few inches to ten feet from the cameras. I'll be using the video from the two cameras to make precise 3-D measurements of the fish's activity using software I wrote (http://www.vidsync.org), but I don't need feedback on the 3-D part of things, just maximizing image quality.

To assure adequate camera placement and maximize recording quality, I need to run clean HDMI output from the cameras to external recorders, which I'll be observing from dry land up on the stream bank. I'll be setting the exposure/focus/etc and starting the recording manually, placing the cameras in the stream, and then watching the fish through the monitors on the bank to make sure the camera placement is good.

One important difference between my requirements and those of most videographers is that I don't want the "cinematic" look -- I need every frame of the video to give the sharpest possible view of every detail, including moving objects (like the tiny drifting items the fish are pursuing). This means going for the fastest shutter speed I can (widest aperture and highest ISO) while maintaining adequate DoF for sharp focus throughout the virtual image and avoiding losing important details to high ISO noise. Artificial lighting is not possible, and I expect to work in a range of natural midday lighting on sunny/cloudy days or in moderate shade from trees.

My plan is to use a pair of the following:

Camera body: Nikon D5300

Lens: Samyang/Rokinon/other brands 8mm f/3.5 Aspherical Fisheye for Nikon AE

Second choice lens: Nikon AF DX 10.5mm f/2.8G ED Fisheye

Housing: Ikelite #6801.53, with hole drilled for HDMI bulkhead

Housing port: Zen Underwater DP-230 9" Superdome

Housing port spacer: I don't know

HDMI monitor/recorder: Atomos Ninja Blade

HDMI bulkhead and cable: DiveAndSee

It makes me nervous that Ken Rockwell's review of the Samyang/Rokinon lens says it can focus up to 12" from the image plane, which will hopefully allow it to focus on parts of the virtual image all the way up to the 9" dome, but I'm not sure. The Nikon lens could definitely focus close enough, but it's more expensive and based on reviews I've read seems like its projection and optical performance aren't quite as good. Also I assume it would have a very slightly narrower field of view and smaller depth of field given the longer focal length.

Also, I don't know what kind of spacer (if any) will be needed to approximately align the nodal points of the dome and the lens. Does anyone have experience with this combination?

I'd appreciate your thoughts on the likely performance of this system, things I might be overlooking, options that might be better, etc. Thanks in advance for any help!

Hi Troutnut, sounds like you have an interesting project on your hands.

I'm not an expert in video, and just a learner in still's I'm afraid but one thing that comes to mind reading your post is that if you are using a very wide lens and the fish is 10 feet away it will probably look tiny and you may not be able to see much detail in the footage.

I'm a keen trout fisherman and tried mounting my housing on a pole and taking some photos of trout cruising in a backwater not long after I got my housing. I was using a tokina 10-17 lens and was surprised by how far away the fish looked even though they seemed very close. Below is a photo of a trout 3 or 4 feet from my camera taken with natural light only (just a screen grab sorry). The water had a slight bit of turbidity but you can see in this photo that already the colour and detail is dropping off significantly.

Also I had a bike cable pulling the shutter lever and I noticed that when the trout were very close either the sound of the shutter or the slight movement of the lever spooked them a bit (not an issue with video I guess but worth considering).

There is a guy on wetpixel who takes awesome photos of spawning salmon (the attached was a wild trout in a back country stream in summer). I think his name is Tom Kliene (sp?) and I think he is also a salmon researcher so suggest you look him up and pm him.

Hope that is of some use and good luck with you project (make sure you post some footage). It's winter here and I'm hopefully going to try again on some spawning rainbows soon.

Cheers

You might also want to post in the "Underwater 3D" section. There are some folks there with quite a bit of experience, who might be able to offer some insight into your plans.

http://wetpixel.com/forums/index.php?showforum=75

Nikon D5300 has not focusing motor, so you have to choose lens which have focusing motor inside. Like any Nikkkor AF-S lens.

Nikon AF DX 10.5mm f/2.8G ED Fisheye has no motor inside so it not autofocus with Nik D5300. I suggest Sigma 10 mm F2.8 EX DC HSM Fisheye instead.

However, If you want shooting video, rectangular lens like Tokina 11-16 f2.8 AT-X Pro DXii, Tokina 12-24 f4 AT-X Pro DXii, or Nikon AF-S DX 10-24 f3.5-4.5G ED would be your choices.

Here is a feeding sequence that I did of Cutthroat Trout. This was downstream of a beaver dam The water is stained. The trout were positioned under a log so are in shadow. I PP'd the images quite a bit already but are still rather poor. This might be somewhat similar to what you are after. I used the Nikkor 10.5mm on a D2x. Cannot see any prey. The action was quite fast.

Tom

Here is another sequence. Same location, different day, slightly upstream viewpoint (looking downstream rather than upstream) compared to the previous sequence. The log casting the shadow is visible in the Snell's window. I shot all these using single shot exposures since I was using a flash. AF was on. To do this in stereo you would want the images synched.

Thanks for the advice so far, everyone.

A few clarifications: I don't need autofocus or zoom. I'll be looking for a fixed focus setting with maximum depth of field. I'll be setting the cameras in a fixed position, as close as possible to the spot the fish feeds (to avoid the turbidity issues seen in Julian D's image), and then hiding back on the bank and waiting for the fish to come back and resume feeding. In my past experience with juvenile Chinook salmon, the fish return to normal behavior within 5-10 minutes after the cameras are placed, and they treat the cameras like just another log in the water. I'm hoping I can get similar proximity to the larger Arctic grayling and dolly varden I'll be studying (in addition to Chinook) in this new project.

I understand there's a tradeoff between field of view and size of the subject on screen, but I have to resolve that in favor of field of view because it's important to capture almost 100% of the fish's activity on both cameras. The angular size of the fish in both Julian and Tom's images would probably be adequate for what I'm doing. The closer the better, as long as I'm getting all the activity on screen. The HDMI monitors should be a big help in framing the shots optimally.

Tom, your work is excellent. I'm surprised we haven't crossed paths before -- where are you based in Alaska? I'm in Fairbanks. Do you ever go to the state AFS meetings?

I took this shot not too far from Fairbanks. There are 3 Grayling in it. One is in the process of grabbing an airborne prey. I find that getting the fish in focus is half the battle - is often the reason for tossing an image (even with a fisheye lens shot at f/8 or so). I rely on AF. This one was done with the Canon 8-15 and is cropped a bit. I have the same experience with the fish ignoring the camera. Sent you a PM.

I dug up this shot from last month's shooting:

That stuff cluttering the shot is pollen - we had quite a bit this year. A Dolly Varden is in the center and in focus the other fish are not. The location is at the outlet of a culvert. There is a beaver lodge in the background to the left. The suspended sediment near the bottom is likely due to the beaver moving around - I did get a shot of it on another day.

Hi Folks;

I'm working on a similar project here in NL. I have a modified GoPro housing that uses an 8" dome (less heat than a DSLR being made in our cold condensation making rivers). It uses a 20,000+ mAh battery bank so it can film almost endlessly and has a raspberry board added so I can control the GoPro and aperture/shutterspeed/exposure metering/etc. Just a thought. find me on facebook and I'd be happy to chat.

Cheers;

Bill Bryden