Simulating human vision -- field of view, etc.

[davision]

I know this is a perennial question in the CG world, but maybe there has been some progress made or some Vray feature can help...

I've read that a 50mm focal length on a 3d (or 35mm) camera approximates human eyesight in terms of the relative perspective. However there is the serious shortcoming of the field-of-view being very narrow (in comparison to the eye). Now I realize the eye approaches 180 degrees FOV due to peripheral vision and that probably isn't attainable.

However does anyone know how to achieve FOV wider than that of a 50mm camera without shortening the focal length and exaggerating the z-depth?

I don't really understand all the different camera types in Vray -- I thought maybe they could be useful -- perhaps with some post-processing?

Any ideas at all?

[vlado]

Re: Simulating human vision -- field of view, etc.

However does anyone know how to achieve FOV wider than that of a 50mm camera without shortening the focal length and exaggerating the z-depth?

Yes - increase the size of the film. Of course, that will give a very similar result to decreasing the focal length, since the FOV is determined mostly by the ratio of the film size and the focal length.

The eye has larger FOV because it has a spherical shape - the image is not projected on a plane, like in cameras, but on the curved surface of the retina. Given a planar film surface, it is not possible to get a wider FOV without distortion of some sort.

I don't really understand all the different camera types in Vray -- I thought maybe they could be useful -- perhaps with some post-processing?

The spherical camera models may help, although they introduce distortions in the image. If you need larger FOV in one direction only, you can use the cylindrical camera, which will introduce distortions only in the horizontal direction.

Best regards,
Vlado

[mike.edel]

having written my diploma thesis about a similar topic, i am a bit biased

one method among several others is using non-planar projection screens (hemispheres, cylinders, etc.)
example:
http://www.elumens.com/products/visionstation.html

#works quite well with the fisheye camera of vray
#effective fov is about 140°
#gives a very nice sense of depth in scenes with forward movement
#needs optimized content (but what output device doesn't?)
#quite expensive

cya, mike

[davision]

Thanks for the replies. That cylindrical camera idea might help...

I'm wondering if there is a way to post-process these images so they aren't quite so warped.

I thought that Quicktime VR -- QTVR worked something like this...

It's been a while since I used it, but I thought it brought in a very panaromic view, but compensated for some of the distortion at the edges. It looks a little weird, but not too bad as I recall.

[Dynedain]

I thought that Quicktime VR -- QTVR worked something like this...

It's been a while since I used it, but I thought it brought in a very panaromic view, but compensated for some of the distortion at the edges. It looks a little weird, but not too bad as I recall.

Nope, QTVR just stitches togethor a bunch of regular images taken at different angles.

[davision]

oh well.

Thanks for telling me. Saved me some time.