Does camera need a corrective lens?


#1

It seems to me that the camera, when looking through the curved pexiglass waterproof container, will experience some severe astigmatism (i.e. horizontal and vertical focusing is different).

Fixing this could be as simple as using a cylindrical lens taped to the front of the camera to compensate.

I have used regular convex lenses in the past on the front of webcams to increase the focus distance to infinity (most webcams have limited auto-focus range that doesn't go out beyond six feet or so).

I think that the effect of the 100mm diameter tube is that it acts as a -6 diopter cylindrical lens. Thus we need a +6 CYL, 0 SPH correction lens. Happily this is a stock eyeglass lens, so they are available for a few bucks.


#2

Yes, I also wonder how the visual distortion of the tube is being handled. Besides correcting the deformation via lens, there might be also the possibility to this software based. I'm sure OpenCV has support for that.


#3

I chatted with the openrov folk at Maker Faire on Saturday and we agreed that there would be astigmatic distortion -- that cannot be removed in software. However, it appears that, in practice, the effect is not too severe.

I now cannot find the place that sells highly cylindrical lenses, but a pretty standard pricelist is http://www.reliableoptics.com/pdf/Stock_Price_List.pdf -- for the simple lenses, it is under 2 bucks per lens.

I think that quite a large cylindrical power is required to compensate for the curve of the tube when underwater -- it might be as high as -6 diopter CYL.

Philip


#4

Hi Philip,

i see two Problems with the tube.

First, you reconised this problem too, the astigmatic distortion.

The curve is not the problem as long as the nodal point of the camera is at the center of the tube (what i think is not), but for normal video i dont think thats a big problem. With wide angel you have always distortions and the quality of a webcam-lens not the best and costs more distortions than the cylinder curve.

Point ist the straight part of the cylinder where the distortion at the edges of the pictures is much bigger, you have another distortion in y tahn in y axis and that corrupts teh impression of the picture.

A kind of dome port is, optical and for pressure reasons, best you can do.

When i have my webcam and the acrylic tube i can try something with an old Ikelite domeport to see if there is a big difference.

But theres another thing that can have a effect to the optics. The light goes through the same acrylic. Maybe i'm wrong and this doesnt have a big effect to the quality of the video.

The light is reflected at the borders of the acrylic so it dos not only go through the acrylic, its spreading inside the acrylic, and the video has to pass this.

Regards

Rob


#5

Ups, i forgot something th the corrective lens.

Normally you need a corrective lens in combination with a dome port.

The dome creates a virtual picture round about 20-30cm in front of the lens.

The lens must be able to focus to that distance. If it couldn't focus to that distance you need a corrective lens to allow the lens to focus to that virt. picture.

Here you have a tube not a dome, so the blurring in x and y axis is different.

regards

Rob


#6

Hi all! I come bearing empirical data!

We're still in a hotel room here on the East Coast after Maker Faire, but using stuff I had laying around and the hotel bathtub, I ran a bunch of tests. Long story short, there doesn't seem to be any noticeable astigmatism when looking at an object roughly 20cm away.

I used the Genius Widcam F100 that is recommended for OpenROV. It has a 120 degree viewing angle and is focused manually by rotating the lens to adjust its focal point. I drew lines on the back of the Maker Faire sign I had so that the camera was the same distance from the target each time. Once the camera was focused for the initial test, I did not refocus it. Here are images from my tests:


Here's what an object looks like with just the bare camera


Same setup but with camera positioned in Electronics Tube (in the same location as it would be in a fully assembled ROV)


Camera and tube mounted on ROV (camera held in place with wadded newspaper since other electronics got in the way for testing with a direct USB connection)


ROV with camera in tube submerged in water. Because objects were being held in place by hand, the camera is not pointing exactly at the same point on the target as before, but it is still easy to see that both the top and bottom parts of the target are sufficiently focused.


#7

Experiment trumps theory every time!

I suspect that the resolution is sufficiently low that the astigmatism doesn't hurt! [Being astigmatic myself, I am acutely aware of what it does to you]

Philip


#8

And furthermore, do we really need to be able to read documents underwater? :)


#9

You never know when you're going to need instructions for connecting Washington Plaza's internet from underwater...