Tweet
When you render mutliple passes of irradiance maps, does it spend less time re-thinking the sections it's already processed?
-
-
yeah... if it is saved in cache or in memory -
Just wondering, I didnt know if it would continue to overlap more irradiance mapping ontop of what it just did, or if it would skip or blend into the previous mapping. Thanks
Now if only I had some faster processors.Comment
-
He could also be asking something a little different.
Are you using incremental add?
Even if you're just rendering with multi-pass you may notice the passes get faster each time. This is VRay adapting by not adding samples that aren't needed. If you notice some grey areas on subsequent passes it's because these are the areas that aren't getting new samples.
--JonComment
-
Well, isn't that the main advantage of using Irradiance Maps? Otherwise, it would be more or less the same as direct computation (brute force). Right?Even if you're just rendering with multi-pass you may notice the passes get faster each time. This is VRay adapting by not adding samples that aren't needed. If you notice some grey areas on subsequent passes it's because these are the areas that aren't getting new samples.
--Jon
Regards,
NenadComment
-
well not really. The main advantage adaptive sampling is that it intelligently places samples where it thinks is should... The passes getting quicker is really just a byproduct of how vlado set it up... Of course, who would create adaptive irradiance mapping and NOT set it up of for multipass across incremental frames?
Comment
-
Right now I seem to have a photon-irradiance transition problem
I render nothign but photons and its all nice and bright, though splochy of course, then when I transfer it to irradiance, the irradiance pass is nearly solid black... this is very confusing
Comment
-
Percy, actually, the first pass isn't adaptive at all, but a basis for each next pass. The first pass will be computed completely, all over the image. The trick is that it won't use the full resolution, but a harsh image, in order to make the computation quicker. Then, each subsequent pass will be computed over the one before, with the adaptive method, according to the threshold parameters.well not really. The main advantage adaptive sampling is that it intelligently places samples where it thinks is should... The passes getting quicker is really just a byproduct of how vlado set it up... Of course, who would create adaptive irradiance mapping and NOT set it up of for multipass across incremental frames?
But, if you lower the threshold parameters, you'll be able to notice that each next pass is getting slower than the one before. This is because :
1. It is computed over a higher resoltion and accordignly to the complexity of the scene, V-Ray is adding quite some details to the Irradiance Map.
2. As the thresholds are low, V-Ray will add even more samples where needed.
Even in this case, (each next pass being slower than the one before) the whole computation will be faster than with direct computation.
My point is that you don't need each next pass to be quicker than the previous one to get advantage over direct computation. The whole Irradiance Map intelligence is in the fact that adaptive calculation is done over a harsh first pass image, not the complete resolution, no matter if each subsequent pass is faster than the previous one or not.
Hey, I just noticed that we're saying the same thing... My words are adapting to yours, just adding detail where needed. Shit, I'm thinking like V-Ray. This thing is driving me mad. Somebody, please get me out of this chair!
Regards,
NenadComment
-
Didn't realize this was directed at meWell, isn't that the main advantage of using Irradiance Maps? Otherwise, it would be more or less the same as direct computation (brute force). Right?
Yes, it is an advantage of using Irrad maps, not really the main reason though, but other methods also adapt so I don't get your point. I was specifically talking about multiple passes and how I know them to work with Irradiance maps. I guess Light Cache could also be considered the same way. Brute force only uses one pass though I haven't used it in well over a year so correct me if I'm wrong. I guess that can't really even be considered a pass though.
--JonComment
Comment