Shading Rate vs SubDivs | Default Values

[TubeSmokeGuy]

Hi!

I watched this Breakdown and arround time 44:58 the author says higher shading rate, higher min subdivs and lower max subdivs are a lot better for rendertime and quality. So I like to discuss this with the officials here and might ask why the default aren't in this way?

It was posted on "Stack" Facebook Group, tagged with "Arrival" breakdown from ILM's 3dsmax Generalist Team:
https://youtu.be/M6n_E4hSK6M

Thank you and Greeting!

[Muhammed_Hamed]

So I like to discuss this with the officials here and might ask why the default aren't in this way?

Using Higher Min Shading Rate favors secondary rays (For glossy effects, GI, SSS and so on) versus primary rays used for AA and DOF
The default value of 6 works very well, I never needed to change this value in Vray 5.. There is just no need to, Vray is very smart and you don't need optimizations for render times.. the default settings work nicely in 99% of cases
Min Shading Rate has been removed from many Vray 5 plugins btw, for this reason. The default value of 6 and min/max subdivs 1/24 are chosen carefully in Vray, they are univeral settings that will work well for most cases
I'm not saying Steffen is wrong in the video, the scene he is working on is a special case needing min subdivs of 2 .. you shouldn't do that for all your scenes
Same for max subdivs of, 8 is not gonna be enough for GI intensive scenes.. many places will be cut off needing more samples (you will need to add sample rate render element to see the areas that are cutoff represented by red)
His scene uses mostly direct lighting, many of the materials are quite basic in his setup.. having only diffuse color
I recommend sticking to default sampling settings in Vray, and using noise threshold to control noise/render times..

[piotrus3333]

heh, this guy is sorting fine geo (leaves) with min aa subdiv of 6... (just about 45 mins in) - he has that issue because of high msr...
just skip this part of the video and stick to defaults. they do work.

[^Lele^]

Add a noiseLevel and a sampleRate RE to your tests, and it will be very clear, very quickly, what kind of mistake he's making, as explained by the other two posters.
Those two REs are the *only* way one has to visually (and with some color-picking) assess a V-Ray's render effort versus reached quality.

As a side note, CG in VFX can get away with stuff archviz or product viz can't dream of.
This is because of the often very heavy-handed post-processing a given shot will go through.
For size, i have *never* had one shot go to finals with its beauty pass, in a few years of Film: everything was always recomposited from the REs, after *much* accessory work.
This to say that they care less for noise uniformity, or even noise amount, when a shot is going to get post-moblurred, glowed, heavily color corrected and in the end added grain to.
However, they need to pump out a metric ton of renders in a very short time so, shortcuts.

[TubeSmokeGuy]

Allright! Thank you !