texture baking AO, what am i doing wrong. ?

[super gnu]

see attached renders, one with original AO (driven by bitmap in radius and with output map to boost contrast after)

the other is a 2k bitmap baked out using vraydiffusefilter element.

firstly its pretty different in behaviour, and second it has some blocky glitches. this is reduced with secondary ray bias, but not eliminated.

how to get a proper match?

cheers!

[Morbid Angel]

well did you bake it to float format like exr? because it seems to be the 8 bit image clipped value or gamma correction that is not like the actual occlusion.

[super gnu]

I did actually bake to jpeg, maybe thats the issue. Also, the square defects are bucket edges. I swithced to progressive and they went away, although it took an hour to do a 2k bake. Probs need to return to bucket and increase min subdivs.

[Morbid Angel]

visible buckets mean you need to increase min subdivs of dmc sampler to higher then 1.