LUTs interpreted differently in V-Ray 7?

[yavor.rubenov]

I found a version for sRGB:

Sort of a side question to those:



If I understand it correctly - here you are using the Curve layer to extract only the R component of the color right ?

Any reason not use the color multiplier in the render element layer itself :



It should give you the same result and be much faster

[Pixelcon]

I'm pretty sure I wasted 2 weeks of my life trying to get around this issue without knowing it was an issue

[piotrus3333]

looks like it is the right time to change some things in VFB. this is getting quite messy.

limited support for lut formats - open domain cubes and csp luts (like in Photoshop) would simplify Lookup table layer significantly. no work in.. options, just apply the lut and done, zero points of possible failure and Save all CC as lut could finally do what it is expected to do.

and just turn Display Correction into a regular layer - for example: what is official Chaos approach to users calibrating their displays with colourimeters and icc profiles and also rendering in ACEScg? I bet 95% have no clue their VRay VFB is not displaying what they expect.


if Display Correction can't possibly be a layer (I guess there is a lot of potential for more confusion not less) - a least add 3x3 colour matrix as a layer.

[piotrus3333]

Sort of a side question to those:



If I understand it correctly - here you are using the Curve layer to extract only the R component of the color right ?

Any reason not use the color multiplier in the render element layer itself :



It should give you the same result and be much faster

haha, thank you for that.
there is nothing better than hanging out with smarter people..

[dimitar_panayotov]

looks like it is the right time to change some things in VFB. this is getting quite messy.

limited support for lut formats - open domain cubes and csp luts (like in Photoshop) would simplify Lookup table layer significantly. no work in.. options, just apply the lut and done, zero points of possible failure and Save all CC as lut could finally do what it is expected to do.

and just turn Display Correction into a regular layer - for example: what is official Chaos approach to users calibrating their displays with colourimeters and icc profiles and also rendering in ACEScg? I bet 95% have no clue their VRay VFB is not displaying what they expect.


if Display Correction can't possibly be a layer (I guess there is a lot of potential for more confusion not less) - a least add 3x3 colour matrix as a layer.

Hi, what do you mean by "open domain cubes"?

[piotrus3333]

Hi, what do you mean by "open domain cubes"?

well, "more open domain" would be appropriate term. luts that can sample range broader than 0-1 float. cube, csp or spi1d formats etc.

like this .cube example:
...
LUT_1D_SIZE 4096
LUT_1D_INPUT_RANGE -0.0056508194 13.5216945686
...

or spi1d lut format that can for example adequately cover range from -0.02 to 470 while being a 1MB file.

ideally I would like to see support for all the lut formats supported by OCIO with options for interpolation and obviously direction of transform (for lin to log transforms).

edit: .spimtx lut format would solve the missing 3x3 colour matrix layer issue (colorspace transforms outside of ocio, ocio could stay ot top as a final display transform)

[Alex_M]

I found a version for sRGB:

After using it for a while, I noticed that the background no longer shows unless I change the "Source" at the bottom of the VFB layers from "Composite" to "RGB". It stays black. Is there anything I can do to make backgrounds like the V-Ray sky show?

[piotrus3333]

After using it for a while, I noticed that the background no longer shows unless I change the "Source" at the bottom of the VFB layers from "Composite" to "RGB". It stays black. Is there anything I can do to make backgrounds like the V-Ray sky show?

put the VRaySky in a dome light that affects alpha.

[piotrus3333]

or change the blending mode of folders to overwrite.