Tweet
So.
No scene is the same so there such thing as the ideal settings or the best settings. What Vlado's method proposes is something that's pretty good for nearly everything so it's a great starting point for nearly every user. Previously what was happening is people were setting up a scene based on the amount of AA they needed for test shots, then when things like hair or fur were added to the scene or they turned on camera motion blur or depth of field, the AA needs of that scene changed and thus all the sampling that they'd set up based on the previous aa had to be throw out. Likewise if you've got an animated scene, some frames might have higher or lower AA requirements so say if you set your vray settings based on tests of one of the easier frames and then later on in the scene there's more motion blur or a finer detailed part, you'll have a dirty part of the animation. Vlado's settings of something like 1 / 24 or 1 / 32 is enough AA to catch nearly every possibility in terms of fine model and texture detail but also good enough to give you decent motion blur and DOF. Likewise putting on 6 - 8 samples of shading rate means you can get up to 256 shading samples which is a fair bit for most things.
In your case, you've got a specific type of scene and yes you can definitely go that high with MSR. Say for example I've got a scene where it's nothing but clean white walls, a window and a hdri lighting the entire thing. If I've got no glossy reflections on my walls so it really is as simple as a white box with a single light, my only source of noise and my only source of sampling is my domelight and my GI settings. It makes zero sense for this particular scene to try and use a high AA method. Ultimately vray is only trying to clean up noise regardless of where it comes from and in this particular scene all of the noise is coming from the light and the GI only so it makes way more sense that if AA of 1/8 is good enough for the quality of the edges of my walls and then throwing lots of samples only at the noise causing light source, you'll get the best results.
The last thing to think of is scene complexity. Tweaking a scene individually is great if it's manageable - if you've only got a few lights and materials it's very doable to tweak things for the quickest results but if you've got a huge scene with hundreds of lights or materials then this isn't as possible. The universal / global settings ideas work quite nicely here and you can get something on to a farm quicker - if you figure out what your operator time is worth compared to your render time does it make more sense to spend less time perfecting settings so you can get through more scenes and then just let the farm work for a bit longer?
No scene is the same so there such thing as the ideal settings or the best settings. What Vlado's method proposes is something that's pretty good for nearly everything so it's a great starting point for nearly every user. Previously what was happening is people were setting up a scene based on the amount of AA they needed for test shots, then when things like hair or fur were added to the scene or they turned on camera motion blur or depth of field, the AA needs of that scene changed and thus all the sampling that they'd set up based on the previous aa had to be throw out. Likewise if you've got an animated scene, some frames might have higher or lower AA requirements so say if you set your vray settings based on tests of one of the easier frames and then later on in the scene there's more motion blur or a finer detailed part, you'll have a dirty part of the animation. Vlado's settings of something like 1 / 24 or 1 / 32 is enough AA to catch nearly every possibility in terms of fine model and texture detail but also good enough to give you decent motion blur and DOF. Likewise putting on 6 - 8 samples of shading rate means you can get up to 256 shading samples which is a fair bit for most things.
In your case, you've got a specific type of scene and yes you can definitely go that high with MSR. Say for example I've got a scene where it's nothing but clean white walls, a window and a hdri lighting the entire thing. If I've got no glossy reflections on my walls so it really is as simple as a white box with a single light, my only source of noise and my only source of sampling is my domelight and my GI settings. It makes zero sense for this particular scene to try and use a high AA method. Ultimately vray is only trying to clean up noise regardless of where it comes from and in this particular scene all of the noise is coming from the light and the GI only so it makes way more sense that if AA of 1/8 is good enough for the quality of the edges of my walls and then throwing lots of samples only at the noise causing light source, you'll get the best results.
The last thing to think of is scene complexity. Tweaking a scene individually is great if it's manageable - if you've only got a few lights and materials it's very doable to tweak things for the quickest results but if you've got a huge scene with hundreds of lights or materials then this isn't as possible. The universal / global settings ideas work quite nicely here and you can get something on to a farm quicker - if you figure out what your operator time is worth compared to your render time does it make more sense to spend less time perfecting settings so you can get through more scenes and then just let the farm work for a bit longer?
We are mostly dealing with interiors , rarely exteriors , and very rare - animations , and the difference between scenes is really subtle , we have lots of plain surfaces , all the materials have reflections / glossy / bumps. Sometimes displace for carpets and stuff , but i think it will be fine even with 80-100 MSR , instead of giving that piece more AA.
Comment