About noise in general

[joconnell]

Yep and it'd be hard to deal with the range of values in a handy manner. Glosiness on the other hand can only ever be from 0 to 1 so it's a better candidate. The vray light lister script is probably fine for the light subdivs, or I could just write in something that'll work on groups of lights you have selected instead of running through the entire scene.

[zeronove]

for my usual kind of subjects (mainly archviz); I find less problematic to set up the lights correctly (they are few, a lot of istances etc.); so my vote goes to something that could speed up the fine tuning material process that, starting from 2.0 release, has became very, very long and boring.

[joconnell]

It'll be very simple to write something to run through all your vray materials and change subdivs based on their glossiness value - you'd need something like VMC to actually change the glossy values though.

[simmsimaging]

Well I won't be able to do that but I can certainly look at a utility to do exactly what I mentioned above for the materials. We can tack on lights too but they're maybe slightly odder ones - admittedly larger lights are softer and will need more samples buit would you use a similar method to determine subdivs by size?

I'm not really sure - I tend to run with everything at default as a result of deriving my own setup from the old Universal approach, so I am not sure how much tweaking is in order for light size. I think it would likely be enough to just have the ability to individually control light sub-d's and material sub-d's, and then maybe a simple global multiplier for each separately (as opposed to the current one that is literally global. That would basically provide most of what you need for fast tweaking of the quality overall and specifically.

In fact, with something like that you could derive a new and fairly simple workflow for lights and materials. Let's say you arrive at a base level of noise/quality you are happy with, for a material, then you could set the glossiness sub-d' in the material to 1 and the global multiplier to whatever 'base' was needed to give that good result, say 10 for an example. If the material is a bit glossier you could set it to 2, or 3, or 4. You would end up with a sort of measuring stick/standard for your materials, basically having materials that are glossy 'type' 1 or 2, or 3 that you would assign as you make them and then never need to really futz with it again because the multiplier would take care of the rest. It would always be a nice simple number to work with when dealing with the materials. You could do the same for lights, figuring out that a base size - say 1mx1m is a "1" and 2mx2m is a '2'. With the gobal multiplier for lights set a 10, or whatever number.

This would give you 2 numbers to change to tweak up or down your whole scene for final quality or making fast preview renders (although you'd probably still need to combine that with the functionality of the Bercon Preview render script for true preview speed). Anyway, you'd be down to tweaking one or two numbers in one or two places.

Just a thought really, but it might be a good workflow to adopt that would make future work faster/simpler to setup. Anyway, it would be possible with such a script but not necessary - and it would be great no matter what if you had time to write it!

/b

[PIXELBOX_SRO]

Yes, thank you stefan, and of course everybody that input on this thread.
I'm doing some tests with your settings on a "real" scene and so far, so good.
One thing I noticed in my tests so far, is that the glossiness amount of a vray material has large effect on render time and memory.
I had an object with a glossy level of .95 and by changing that to .8 the rendering time reduced from about 8 hours to 6...Also, before changing that, the LC was taking closer to an hour, and after the change to .8 LC finished in 8-10 minutes. Will post images later...

how woudl you interpret that behaviour according to that DMC tutorial? Should it not be vice versa? the more blurred reflections, the bigger rendertime?

[lllab]

i think it is logic but not only dmc related.
as when you have not so blurry reflections the LC interpolation needs more retracing/render more precise,
if more glossy the interpolation is easier/faster to be in the range iof the set thresholds, specially if you use retrace this effect will be bigger.

stefan

[RE:FORM_STUDIOS]

My material lister script would give visual access to all the subdivs of your materials... if that's any help on this topic!
http://patrick.reformstudios.com/wor...terial-lister/
Be warned, this script hasn't been updated for a while so I can't vouch for it working with the latest max and vray. Also, it is limited in the number of materials it can handle.
If you try it out and come across any bugs or have any suggestions, please post a comment on the page linked.

[PIXELBOX_SRO]

right ill do some more tests in regards to the usage of LC for glossies computation as well....guess that could speed up the whole thing a bit more....wanna see the quality difference in the definition of the glossies
so far this setup of yours works really good

[zeronove]

I tend to run with everything at default as a result of deriving my own setup from the old Universal approach,
/b

that could be similar to my previous workflow (something derived from universal settings); I'm not able to use it anymore till 2.0; how do you solve noise? still only with DMC image sampler?

[voltron7]

how woudl you interpret that behaviour according to that DMC tutorial? Should it not be vice versa? the more blurred reflections, the bigger rendertime?

Yes, that is what my "sort-of understanding" had been, too.

[voltron7]

i think it is logic but not only dmc related.
as when you have not so blurry reflections the LC interpolation needs more retracing/render more precise,
if more glossy the interpolation is easier/faster to be in the range iof the set thresholds, specially if you use retrace this effect will be bigger.

stefan

FWIW, the scene does use LC retrace, and LC prefilter is at 50.

[PIXELBOX_SRO]

that could be similar to my previous workflow (something derived from universal settings); I'm not able to use it anymore till 2.0; how do you solve noise? still only with DMC image sampler?

i think llab has sorted it...at least according to some of my tests

[zeronove]

yes, I already tested what llab suggests but it's seems to be totally different from what Brett is saying

[lllab]

no it is still all based on this paper/tutorial i think and not contradicting, up to a certain subdivision the AA threshold is the only really counting. i think thats also in the dmc tutorial. additional this special case only considering some extra related things like glossy ray interpolation and retrace value which is not covered by this tutorial

stefan.

[zeronove]

confusion is rising...
lllab method is similar/derived to/from interstation 3d tutorial (high subds because of high AA ...well, more or less), ok?
Brett said that he uses default values for subds, that means 8 (not an high value)
Actually my opinion is slightly different:
tweaking the DMC sampler setting allows me only to solve antialiasing problem, no more noise coming from materials or lighting; for that I always end to raise their subdivision on local basis (very time consuming), I don't like use global multi because it always lead to rendertime waste.
I have also made some test using high AA setting like you suggest (100) but for my scenes is slower; as an example right now I'm working with max 32 AA and 0,03 thershold (subs are various starting from standard 8 up till 256).
To summarize:
glossy and light/shadows are solved by their subds
antialias is solved by AA
crl threshold controls general noise/speed ratio
...pretty basic workflow but actually for me is the only way to go