Cylindrical camera photomontage

[tricky]

Increasingly, we are creating photomontages where we need to stitch together a series of photographs to give us the FOV to encompass both the site and a bit of context. Some of these'panoramic' images can have FOVs of around 100 degrees or more. Obviously, with a rectilinear projected camera (a 'normal' camera), things look very distorted and stretched around the edges.

The software we use to stitch the photos together allows us to output 'cylindrical' images, which show far less distortion, and I have noticed that Vray supports cylindrical cameras. Does anyone have any experience on using cylindrical type images - for example, is it easy to accurately match a vray cylindrical camera in Max to the cylindrical panoramic backdrop? Does anyone have any examples of images they have produced that they could post up here? I suppose the output from a cylindrical camera would need to be printed on a curved sheet of paper in an ideal world, but I would be very interested to see what an image would look like just printed flat.

[jacksc02]

what stitching programme do you use??

I too have problems with photos with large FOV. I tend to use a speherical camera with a lot of post processing in photochop.

[tricky]

Mainly ptgui

[jujubee]

So your background is coming in distorted after stitching? What does this have to do with the Vray cylindrical camera type and why would you want to use that?

As for creating 360 degree views, the best way to do it is to get a fish-eyed lens and take two-three shots. Here's an:

http://www.cgtechniques.com/tutorials/makepano.php

There's also got to be other links out there...

[tricky]

jujubee: I obviously haven't explained myself clearly. Maybe you can understand the following couple of images more easily:




Both of these are 120 degree FOV, the 1st is rectilinear, the 2nd cylindrical. I wondered if anyone had created a photomontage using the cylindrical photomontage - NOT to be used as a QTVR or anything like that, but just to be printed flat as with a normal image.

As far as I am aware, chosing a Vray camera type does not reflect what you actually see in the viewport, but only at render time. This would suggest that it would be difficult to match correctly.

Is this any clearer?