Hi
I just realized that dirt doesn’t show up with qmc AA.
(well, not shown enough)
Is this a bug or feature?
-LarsSon
It will show up if you use more subdivs for the dirt.
Best regards,
Vlado
I’m using 16 already. Should I go as hi as 32 or 64?
Is there any short explanation why dirt with Qmc needs more subdivisions?
I just want to understand how things work.
-LarsSon
because the AA method (adaptive QMC) u want to use is more sharp maybe
Because, when you use VRayDirt in a non-linear way, the result you get depends on how many subdivs are calculated; the qmc aa distributes all the dirt subdivs among the subdivs for the aa itself, but the net result might be different than if the subdivs were not distributed but computed in full for each aa subdiv.
If you are interested in a little maths, here is a more detailed explanation.
Suppose that you are using VRayDirt inside some non-linear texture (e.g. mask with curve or something). In that case, for Adaptive subdivision AA, you will get something like pixel=mask_tex(dirt_tex_full). If you use Adaptive qmc AA, you will get something like pixel=(mask_tex(dirt_tex_0)+mask_tex(dirt_tex_1)+…)/num_aa_samples.
The two values are equal only if mask_tex() is a linear function.
Note that this is not V-Ray specific thing; you will get this with any renderer where an AO texture is used in a non-linear way with a small number of samples.
Best regards,
Vlado
Thank you very much!
I think that I understand it now. Actually, It’s really obvious why
the result looks like that.
Is it possible, that there could be an option to pre-calculate somekind of
dirtmap? Like calculating LC, interpolating it and using it as saved solution.
But same think for dirt. After all, dirt is only AO with some added features.
Then we wouldn’t need to render dirt totally using qmc sampling.
I’m not very good eplaining this, but I hope you get my point.
-LarsSon