PDA

View Full Version : focal length for camera mapping


bcottman
11-05-2004, 10:49 AM
I am in the process of camera mapping a painting and I was curious about how to determine the apropriate focal length of the projecting camera. I have it all set up with a long focal length thinking that it would reduce the stretching artifacts. however, after doing some tests I am begining to think that I should have went the opposite direction with it. Or maybe left it somewhere in the middle. I also wondered if it was necesary to keep identical settings on both the animated and projection cameras. (The paintings focal point is rather deep and I am attempting to push pretty far into it.)

If anyone has any relevent information on recomended camera setting for this sort of thing I would be very interested.

I have camera mapped paintings before and am familiar with the process. I am just drawing blanks in this area.

Thanks
Brenton

jfrancis
11-15-2004, 08:31 AM
This brief tutorial might answer your questions...

http://www.digitalartform.com/archives/ ... cti_1.html (http://www.digitalartform.com/archives/2004/11/camera_projecti_1.html)

http://www.digitalartform.com/archives/images/forBox.jpg
In it, I turn this photo . . .

http://www.digitalartform.com/archives/images/camProjSwoop.gif
. . . into this animation.

The main thing is to match your wireframe to your painting at the moment of projection. After that, feel free to separate one from the other.

bcottman
11-15-2004, 10:54 AM
Thank you for the reply jfrancis. Your tutorial very clear. It doesnt really address my question though.

My question is... that altough you can essentialy project an image onto any geomety through any focal length as long as they match. Using different focal lengths require different geometry to match it. for instance a long lens would require the ground plane geomety to stretch much further in space than if you were to set up the same projection with a short lens. So my question is how you determine an appropriate focal length to start with.

I am done with the project but I am still curious how others solve this problem for future reference.
Thanks
Brenton

jfrancis
11-15-2004, 11:06 AM
"You can project any image onto any geometry with any lens as long as they match" -- that's a big "as long as"

If you've matched the relative position and angle of a camera then there is only one focal length that will allow a photographed object and its match-modelled wireframe to line up.

Lining up the wireframe to the photograph is the bulk of the job. If you've done that, then you have already made a decision about focal length, camera position, and camera orientation.


http://www.digitalartform.com/lenses.htm

According to a common misconception, image distortion results from the kind of lens one uses. In many people's minds, a wide-angle lens produces great distortion, while a telephoto lens "flattens images."

In fact the distortion arises from the distance between camera and object, and it becomes most apparent when a distantly-photographed object is unnaturally magnified by a telephoto lens, or when a closely-photographed object is unnaturally reduced by a wide-angle lens.

Close objects display distortion by virtue of their closeness. People associate distortion with wide-angle lenses because such lenses are commonly used to pleasingly frame close objects. The wide angle lens is not the cause of the distortion.

Distant objects display flatness by virtue of their distance. People associate flatness with telephoto lenses because such lenses are commonly used to pleasingly frame distant objects. The telephoto lens is not the cause of the flatness.

You can't DISTORT a cube without getting VERY CLOSE to it. No matter what focal length lens you use.

You can't FLATTEN a cube without getting VERY FAR from it, No matter what focal length lens you use.

http://www.digitalartform.com/assets/lensGeometry.jpg

http://www.digitalartform.com/assets/lensGeometry2.jpg

http://www.digitalartform.com/assets/lens_demo.gif

bcottman
11-15-2004, 06:53 PM
Essentially what you are saying is that the distant object will appear as flat through the wide angle lens as it does through the telephoto lens. The telephoto lens only enlarges or "magnifies" that portion of the scene. So determining focal length is just a matter of figuring the distance that objects appear to flatten.

My primary confusion was with wide angle lenses but you're saying that all lenses are the same just cropped differently. Does that sound right?

"If you've matched the relative position and angle of a camera then there is only one focal length that will allow a photographed object and its match-modelled wireframe to line up."

I usually arrive at a decent match through a long trial and error process and assume there are other settings that would also have worked (possibly better). This should really help to have a clearer approach to that trial and error and to be confident to stop when it looks correct.

Thanks alot
Brenton

NickATnitE
11-15-2004, 07:57 PM
I recently came across this tutorial on how to find the camera information from a still picture, It might help you out for what you are doing. I usually do it by eye but if you really need to find out where the camera was in your plate this should help or give you a place to start. as for the camera settings, the projecting camera is the one that has to be set to the image size etc. the animated one should be whatever your final render settings will be. http://www.fxguide.com/fxtips-233.html hope this helps,
good luck
Nick

jfrancis
11-15-2004, 09:29 PM
Yes, that's right - all lenses are the same lens only cropped differently, and a distant cube seen through a wide angle lens will look very tiny, but if you look closely at that tiny cube, you'll see it to be flattened, not distorted.

http://www.digitalartform.com/archives/images/allLenses.jpg
http://www.digitalartform.com/archives/ ... and_w.html (http://www.digitalartform.com/archives/2004/12/telephoto_and_w.html)