automatic calculations of subdivs is great and computing is much faster. But what I can do when I wanna better quality - less noise? I have an exterior scene, can see to interior and there is too much noise for me. I tried to lower Noise treshold, but nothing really change (and it was a greatest gamechanger in standard subdiv mode). Tried to change Min shading rate, but again nothing happen. Tried to change Min and Max subdivs, there is a change and less noise, but not as good as I expect. And when I change those parameters more it already not occurred such a change.
Can you show me, or even share that part of the scene? The things that you mentioned should affect the noise levels, so maybe there is something else going on.
If you have an exterior and you can see into the interior, check that you enabled “affect shadows - all channels” on refraction for the glass. If you forgot this, V-Ray will have a harder time cleaning up the interior as less light is “streaming” in through the windows
There is still the same amount of light in interior. I have lights in inteior where windows are to light the interior. If I delete those light, interior is almost black. All channels option didnt help, interior is still almost black wihtout supporting lights.
You can see how it looks with supporting lights at windows. Wihtout them interior is almost black (dont have picture saved, sorry).
I would actually really really (really!) leave the MSR alone In fact, I’ll probably hide it in some of the next builds.
Attached is an .ods spreadsheet where I rendered a bunch of scenes with varying complexity (interiors/exteriors and so on) with different MSR settings. You can see that 6-8 is a good value for many scenes. While for some scenes lowering it or increasing a little it may reduce render times by some percent, for other scenes incorrect MSR can lead to huge render time increases.
The spreadsheet does not reflect the most recent changes in V-Ray for some of the scenes, but it can still show you that there is not much benefit to be gained by straying too much from the default.
As I wrote, change Color Threshold from 0,01 to something lower solve it - noise was much less. If I have any trouble when I compute final scene, I will let you know.
Because it didn’t really make any sense. The noise threshold for the DMC sampler is used only for secondary effects (reflections, GI) and not always at that. The default value of 0.005 is good for practically all cases and there is little to be gained by adjusting it.
The color threshold for the image sampler on the other hand makes a lot of sense to be adjustable.
I find it impossible to remove the noise in automatic mode using BF+LC. Even lowering the thresholds to 0.001 still produces extreme noise (which appear related to secondary rays). Testing a scene with a mid grey material illuminated with sun+sky. Same scene rendered clean with a dome HDR using v-ray 3.20
There is no use decreasing the threshold if you don’t have enough max. subdivs. Look at the sample rate render element - if it’s all red, you need more max subdivs.
I know that. I tried that ultra low thresholds on 1/100 subdivs. I can get a clean image with any GI combination with the individual approach. Setting the lights/mats/GI subdivisions manually. One of the numerous novelties in SP3 is the “automatic” settings mode which should provide production grade quality for those users who don’t want to meddle with various settings or doesn’t care for render times. Perhaps I just switched to SP3 on a tricky scene as it is… My point is, I could have used the universal settings on a scene with SP2 and get decent results, but when loading the same scene with SP3 I get unacceptable results (same settings) and end up spending a great amount of time tweaking the settings to get the same results as before. Isn’t that the opposite of what SP3 aimed to be?
I am confused now as to who you’re answering to.
If you refer to me on the other thread, please keep it there, or ideally entirely away from an announcement thread, and start an issue thread, where we can all delve into minutiae without polluting what should be a reference thread for everyone else.
And if you provide us with a scene to look at, we may perhaps offer you an explanation as to why were you experiencing what.
A process to intentions may be fun on a lazy Sunday afternoon, but it doesn’t get images rendered.
And i feel i exhausted my options, without some solid data to work on.
Yes, in the particular scene that you sent (empty room, flat walls and floor, sunset lighting), higher MSR does produce slightly lower render times.
However, if I do a few seemingly minor modifications to the scene - f.e. I added slight bump on the walls, a bump map on the floor, displaced rug on the floor, added some plants and moved the sun to a different time of day - the situation is quite different and the optimal MSR gravitates back to the default value of 6.
So my thoughts are these:
*) In my view, it is unlikely that you will have to render a final image consisting of bare walls and nothing else. Except for the initial development stages (when you need mostly just preview images anyways), most scenes would have at least some other stuff going on.
*) Is it worth it to spend time playing with the MSR value for the few extra percent? If you want to squeeze as much render time as possible out of the scene because you have 1000 frames to render - then perhaps yes. But if you have a few stills to render, it might take more time to find the optimal MSR value that it would be to just render with the defaults.
My goal with the next releases of V-Ray is to reduce the render time differences between different MSR values, so that it would be even less necessary to play with the MSR value.
Guys, can you start different threads for your own issues?
