adaptive sampling - huge rendertimes caused by simple textures

[Zxx]

LOOK AT THE PREVIOUS PAGE FOR THE BEGINNING OF MY POST

4. added more filtering (filtering = pyramid, blur = 4):
sampling 1-16, time: 5:22
test4.jpg
diagnostic
4diag.jpg

texture looks weird and rendertime still too big. so filtering is not an option.

[bart]

Filtering that is baked into the mipmap/pyramid may be an option. That is what we are talking about I believe.

[Zxx]

i've converted texture to openEXR with baked pyramid filtering. also tried mental ray map manager. almost no effect.
i made this test on different computer so rendertimes a little bit lower
original render: 1-16 with jpg texture - 9:25
enabled mrmm: 9:22
changed texture to openexr with baked filtering (mrmm off): 8:52
exr with mrmm enabled: 9:20
don't see any significant change

[Remydrh]

What are you using to bake the EXR pyramid?

The problem I see with the minimum samples method is once animated, that texture will probably sizzle.

[Zxx]

i used "imf_copy -p -r -k zip working_texture.tif published_texture.exr" command from your blog

[Remydrh]

I would have to check but depending on the pyramid level it's choosing and maybe the filter after that (maybe not elliptical) it's probably blurring it.

For a still frame your minimum samples look ok but once it starts to move, it will cause problems. mental ray was built for animation to an extent so it doesn't understand that. You may be able to raise the contrast threshold some and find a happy medium.