Serious antialiasing issues with LWF

[pablo scapi]

Hello:

 

I´m tring to work with the archi-discussed Linear Work Flow, particulary using the gamma correction color map (1 - 0,454) to burn the result in a tga.

 

The thing is that i love the results in therms of colors, lighting, etc, but in most cases that i´ve tested the images were horrible aliased, to the point that i can´t use this procedure even when i love the results.

 

I´m very curious to know if there´s someone who is having this problems too, because i´ve been looking for this problems in the forums and hadn´t found anyone mentioning this issue, when in my case is so evident.

 

 

mybe is becuse of the method that i´m using, ¿should this problem disapear if i convert the .vrimg files to .exr and open it in PS without using the gamma color map? ¿is that the procedure to go (using linear color map 1-1)

 

Thank you for your attention.

[alexthg]

I don't know if i got the same problem with you, but i do get some pretty bad antialias/noise result when using the linear colormapping instead of gammacorrection method.

so I just made a little test...

this one with linear colormapping, dark 1.0, bright 1.0, gamma 1.0, rendered in VFB with sRGB turned on.

and this one with gamma correction colormapping, multiplier 1.0, inverse gamma 1.0, gamma 2.2, rendered in VFB with sRGB turned off.

Pls notice that with gamma correction colormapping, the glossy reflection looks much smoother, especially in the dark areas, than using linear colormapping.

Other settings are:

- VRay ver 1.5.RC2, 3dsmax 7

- Image sampler : Adaptive QMC 1,4 (default)

- Antialias : none

- Primary bounce : Irrmap low preset, hsph20

- 2nd bounce : Lightcache subdiv400

- rQMC : default

- Max gamma : 2.2, Input gamma 2.2, output 1.0

- Lightsource : vraylight (plane), stored with irradiancemap

- All materials have glossy reflection with subdiv 8 (no interpolation)

anyone care to explain why the difference? And how to achieve the same result with linear colormapping?

[alexthg]

Ok, here's my further test

I got the 'correct' result by putting gamma 2.2 in the linear colormapping rollout, instead of using the sRGB button in VFB.

thus it leads to further question... by putting gamma 2.2 in linearcolormap, does it mean that I 'baked' the gamma? If the answer is no, then I suppose we don't need to use the gamma correction colormapping method anymore?