Wondering what setups you use when testing your scenes lately with new developments?

[pixela]

I have been testing with progressive image sampler, trying to go deep into method with adjusting MSR and playing with all these stuff recently.
Final render time is a secondary worry for me as I am usually dealing with stills. But my main concern is the speed of test renders and it is crucial to find a nice spot with adequate quality and high speed. Especially for testing the materials and surface quality.
So I was wondering what setups you guys are using for testing scenes.
Do you prefer progressive image sampler? Do you tweak MSR and utilize subdivs per pixel information? Or what methods, techniques ....?

[joconnell]

I'm slightly different to most but what I'll generally do is use a combination of crops at full quality for specific areas / materials, half resolution for scenes where the final render is fairly quick or use the colour / dmc threshold at higher values - at 0.05 or even 0.1 so that as adaptive sampling is being done, each bucket will meet my 0.1 threshold on the first round of sampling. It also means that I'm only using one control to switch between test and final quality. You can apply the same principle anywhere in vray - just use whatever setting that determines what final quality it has to work to, be it time or colour / dmc threshold and then raise it so that vray doesn't have to work to as high a standard.

[pixela]

I'm slightly different to most but what I'll generally do is use a combination of crops at full quality for specific areas / materials, half resolution for scenes where the final render is fairly quick or use the colour / dmc threshold at higher values - at 0.05 or even 0.1 so that as adaptive sampling is being done, each bucket will meet my 0.1 threshold on the first round of sampling. It also means that I'm only using one control to switch between test and final quality. You can apply the same principle anywhere in vray - just use whatever setting that determines what final quality it has to work to, be it time or colour / dmc threshold and then raise it so that vray doesn't have to work to as high a standard.

I understand but I can't manage the high thresholds such as 0.05-0.1 when testing my materials, how texturing, bump maps, normal maps has an impact on little details of geometry. For lighting tests, I can survive with such high thresholds having an eye trained to filter the noise in the render. But for shaders, I can't.

[joconnell]

It sounds like you can't afford to compromise on render quality in that case, the only option seems like you'd need a good lookdev scene that you use which shows off the shader only without the overhead of a full scene. If you always have to view things in the context of your scene lighting then you're in a slightly nastier situation, you might be able to try the new subdiv multiplier in the right click > vray properties menu to temporarily reduce everything else in your scene so that all the quality is focused on the object that you're currently tweaking.

[pixela]

It sounds like you can't afford to compromise on render quality in that case, the only option seems like you'd need a good lookdev scene that you use which shows off the shader only without the overhead of a full scene. If you always have to view things in the context of your scene lighting then you're in a slightly nastier situation, you might be able to try the new subdiv multiplier in the right click > vray properties menu to temporarily reduce everything else in your scene so that all the quality is focused on the object that you're currently tweaking.

Yes. Until now, I was trying to do something similar- with "selected" parameter as the render mask and rendering the image.
I am trying to test if the progressive image sampler will help me speed up things for such purposes.