I remember from my old Blender days a rendering mode called Approximate Ambient Occlusion. It rendered as an AO pass but extremely fast and smooth. This was extremely useful in scenario where you need to fake GI and to get a very quick render time. I find getting an AO pass via VrayDirt slightly cumbersome but mainly quite slow, especially if you want to remove all the grain. Could something like that be implemented in Vray?
Are you looking for something like the effect from VRayDistanceTex?
Best regards,
Vlado
Basically I’m looking for the smoothness and speed of the DistTex, but in an AO pass.
This link probably explains it much better than I do. It also includes links to the relevant papers:
http://www.blender.org/development/release-logs/blender-246/approximate-ambient-occlusion/
Looks like single-bounce IR to me.
this is a handy little plugin which saves a lot of time (it leaves all your settings as they are) http://plugins.angstraum.at/vrayao/
but no idea if it does what you are looking for
How do you render AO pass with VrayDistanceTex?
It looks like Approximate Ambient Occlusion, but I don’t understand how I could get AO pass from whole scene with it.
You can’t for the moment - I just gave it as an example for the effect that is needed.
Best regards,
Vlado
yeah it would be great to get a faster ao even at the expense of a little accuracy. been using mr for that recently as it is faster than vray methods.
V Miller
Duke has the answer!
Just setup a single bounce GI soulution using the irradiance map-[TURN OFF secondary bounces]
This will give you an ambient occlusion pass that is super smooth as it is using the smoothing/interpolation of the irradiance map. You also have Detail enhancement if using this method which can help define small details that the irradiance map often smooths too much.
Cheers
Cheers
but in animation won’t the IR method result in flickering?
V Miller
bumping this thread as im finding it very hard to get a grain free result with vray dirt with vray ambient light..
simple abstract scene with a clump of intersecting 3d polygons. with everything else turned off, im finding i cant even clean the grain with 0.001 noise threshold and 900 subdivs on the dirtmap..
at default settings it takes 10 seconds,but looks terrible.
with only noise threshold dropped by a factor of 10 (0.001) the render goes 1 second faster for some reason, but looks identical..
with 0.001 NT and 900 subdivs, it looks arguably worse, and rendertime goes from 9 seconds to 2 minutes.
increasing the aa to 3/10 cleans things up significantly, but there is still visible grain, and now it takes 10 minutes for the same crop!
so i tried some “universal settings” tricks. dirt subdivs back to 8, and AA to 2/100 now it looks much cleaner, and only takes 2 min 54 secs. whaaaa? surely it should be more efficient to improve dirt with the dirt subdivs rather than the aa system?
maan sometimes the DMC sampler makes my head hurt.
anyway, its still a bit grainy, it still takes 20x longer than the original test.. just seems a bit much for such a simple scene. the whole frame is 720p and will take almost 30 mins at these settings. -and i need to add DOF.
flicking through those, if you look at the difference between the first and last ( one using aa to clean things, the other using high subdivs) you can clearly see a large difference in the look.. any suggestions?
I’ve got a theory that the aa sampler in vray is either too nice for it’s own good, or else it’s an utter asshole which thinks that it’s way to clean up your image noise is the best way (and it’s not). I kind of think with the universal approach (old style) it’s a bit wasteful as AA might not be the most efficient way to clean up a problem being caused by something else (in your case the dirt texture). I’ve some more tests to do but so far it seems to hold firm as a method.
The AA sampler should be for doing sampling on geometry edges, textures, highlights and nothing else really (I think anyway) but if you want you can use the universal method to force everything else through it too. In the case of your scene you’ve got simple geometry with no textures and no highlights that are going to cause really blown out areas and so you could easily get away with low AA. I’d try a default 1,4 with the standard 0.01 threshold on your AA sampler, and then just start upping the dirt map samples from the standard 8 up to 32 first of all, then up to 64 afterwards, and use the threshold in the DMC sampler instead to control how fine the grain in your dirt map is.
well if you look at the sample images thats exactly what i did 
i got all the way to 0.001 on the noise thresh. and 900 subdivs for the dirt. it looked -more- noisy and took 12 times longer to render than the defaults. increasing the aa on top of that boosted the rendertime through the roof, as expected. but still didnt look great.
what baffles me is relying on the aa to clear it up is actually faster and cleaner.. (but, seems to produce a different “quality” to the AO.)
Is it possible that you’ve set the “Adaptive amount” in the DMC sampler settings to 1.0? If yes, then this is probably the reason why you are not seeing much of an improvement when you up the samples.
Best regards,
Vlado
Where do you normally start from vlado with that value?
as a matter of fact.. i do. i rendered it now, but good one to bear in mind.. i thought the adaptive amount at 1, only really gave issues with the imap.
With the default 
I wouldn’t normally increase it past 0.9
Best regards,
Vlado
Grand so - It seems like the 1.0 value has been slightly debunked!
Where does something like motion blur subdivs fit into all this also? I can totally understand The AA sampler being able to reduce the number of material and Light subdivs of an object since the eye rays will hit a surface, but for motion blur subdivs how does that tie in with the various unified sampling parts? Would it be considered almost like extra geometry (eg AA style) sampling rays being added in with a motion blurred object almost being treated like duplicates of the same mesh at different positions in time?
What would be the way to think about your motion blur subdivs and how they’re affected by aa reducing other samplers - would it be the same thing of your subdivs being reduced by the rate of the aa?
Much appreciated for the advice again!
The rays for sampling motion blur are very similar to the rays for AA. I usually leave motion blur to the adaptive DMC image sampler.
Best regards,
Vlado