In your pixel-to-world-ray shader, you have this :
miScalar s = sqrtf(2 – r2);

In your inverted (world-point-to-pixel) shader, you have this :
float scale = 1 / sqrt(1 + c);

Plots of each show that, indeed, this is not a perfect inversion. If, however, I change the pixel-to-world-ray shader to sqrtf(1 – r2), it is a perfect inversion.

I presume you’ve increased the “fisheye’ness” of the pixel-to-world-ray shader since maybe it looks nicer for your needs, but wanted to make sure you take this modification into account in the inverted code.

Thanks for the nice post!

Is it possible to send the .dll to me, i’m not a programer and i don’t have any idea of compile, i’m just need the shader to my project in 3Ds max, thnx

after several hours of research, i managed to get it to work in Max. my ignorance of using custom shaders kept me from realizing that i was trying to compile for a 64-bit environment with a 32-bit compiler. >.< after i acquired the proper tools, everything worked fine. thank you for posting this. i have been searching for something similiar for a few months for my project. now, on to the animating!

We are using it in Maya 2011. I don’t know how it works in Max 2012, maybe they changed the interface and some new things are necessary? This could be a good learning opportunity…

i got all this code from autodesk’s mental ray 2012 help file…

i compiled your .c as:
cl /c /O2 /MD /W3 -DWIN_NT -DBIT64 witness_fisheye.c

then tried to link the obj with:
link /nodefaultlib:LIBC.LIB /OPT:NOREF /DLL /OUT:witness_fisheye.dll witness_fisheye.obj “C:\Program Files\Autodesk\3ds Max 2012\mentalimages\dev\lib64\shader.lib”

there i got fatal errors because four mi_ functions were not recognized. it did not error using the 32 bit shader.lib, but then it wouldn’t load. i did also change the include file in the .mi to match my file name.

the JS_fisheye loads, but on render, i get error 051011: shader “JS_fisheye” not found.

i hate my life. >.< am i doing something wrong or is it this stupid 2012 version of max?? any help would be greatly appreciated!!

Thanks, and nice work guys.

If they are encoding a fish eye view of actual gameplay into an AVI, maybe this is later (or earlier in game time…?) projected as reflections and/or shadows into the live scene. So you *witness* yourself (or someone else) doing things at a different point in time, like some kind of spectral apparition.

Pierre: What I�m wondering is what the massive spoiler could be.By my limited understanding, the player does not normally view the world through this, it�s just a technique used for rendering things like shadows, reflections etc. What could be so spoiler-ey about that? Perhaps some secret message that�s only revealed when you view the scene through a fish-eye lense? Like anamorphic drawings?(http://tinyurl.com/6j2msq9)I can already tell this game is going to be something special. Keep up the awesome work!

http://artis.imag.fr/Publications/2008/GHFP08/

At the time, I was working on computing the fisheye view from each polygon in the scene along its normal to figure out what it could see. This “cube” of data then got passed along to a coworker for more calculations, so my experience is somewhat limited.

I ended up going with a straight forward cube map method in the end as it was fast enough for our needs, but it sounds like cubemaps our out for you guys. Not sure if the non-linear projection shader code/paper at the link will be of any use, but it was some interesting reading at least :P.

By my limited understanding, the player does not normally view the world through this, it’s just a technique used for rendering things like shadows, reflections etc. What could be so spoiler-ey about that? Perhaps some secret message that’s only revealed when you view the scene through a fish-eye lense? Like anamorphic drawings? (http://tinyurl.com/6j2msq9)

I can already tell this game is going to be something special. Keep up the awesome work!

What we are talking about here is for, say, drawing reflections on reflective objects (that’s not actually what we are using it for but it is similar in implementation.)

good luck with other experiments and innovative ideas, jonathan and team!