CGeometry_WorldPosition Exponential Values

[rfletchr]

Hi

I tested out CGeometry_WorldPosition today to see how it works now that negative values are being output.

It seems like the values are exponentially scaling as they become distant from the world origin.

I've attached images the scene turned into a point cloud in Nuke. one is done with mentalray and the other corona. both are created using 32bit floating point EXR files.

I'm also going to like the test scenes
https://drive.google.com/file/d/0B6mym9MXR7orT0lETTZ1WTNJYTg/view?usp=sharing

[racoonart]

looks more like wrong gamma to me. Are you saving 1.0 exrs?

[rfletchr]

The gamma is set to Override: 1.0 on both mentalray and Corona. Just to be sure I just rendered with automatic gamma settings and its exactly the same.
didnt really expect any difference as the render elements should never be gamma corrected. I also just tried manually adjusting the gamma up and down. all it does is make things worse.

I'd really appreciate it if someone could try a render and see if they're getting the same results.

[Ondra]

didnt really expect any difference as the render elements should never be gamma corrected.

wel... you can tell that to 3ds max ;)

[rfletchr]

Thanks for looking at this Ondra.

I've affixed a point cloud of the mentalray point poisition pass with a gamma of .454 applied. It has a similar effect but its no where near as extreme

(The compression is reversed on the x because the channel was inverted in nuke)

[maru]

moving - reason: old
revive if needed

[Ondra]

fixed (but now LDR data elements need to be explicitly saved with gamma 1.0)