I’ve been trying to remove the reflection noise with not much effect.
I attached my rawReflection pass, which has the following settings:
dmc sampler 2/10, 0.003, and different AA samplers with no visible difference
Reflective material: glossy 0.7 with 32 subdivs. I tried 64 with no significant difference.
Right now my render times are at 30 minutes on a 12 core i7.
Is the reflection interpolation the way to go? I can’t get it to work without increasing insanely the render times. Thing is, I can’t live with 3 hour render times or I won’t get there in time nor budget.
What are the options?
Thanks!
PS.How does alex roman render all those glossy things anyway?? That’s just insane!
Never reflection interpolation. Try DMC 1/10, 0.01 and on your reflective material try multiples of 10 (use multiples of your max aa always). Try going 80 / 120 / 160 on a tiny region. Also make sure you check your normal reflection pass as well as your raw one. If the final strength of the reflections in part of your render is only a small amount (like the front facing bits on a fresnel material or on a material that’s only got a really weak reflection) then vray won’t bother heavily sampling this as it won’t be seen in the final picture. If your raw ref is dirty, but your ref pass looks okay, then the ref pass is showing you what the actual final reflection contribution looks like and you’re good to go.
The entire scene is a kitchen with fully reflective metals, so no fresnel happening there. For animation, with a POV scene with lots of motion blur happening. 4 area lights so far.
I don’t think I can go for 128 samples. That simply won’t render. Seems to be more complicated than I thought. Are there any other options?
let me try dmc 1/10 0.01 and 80 subdivs
how about subdivs mult in the dmc sampler settings. obiously I have to go for 1?
Did u optimized lights to give enought samples?
Are u using IBL ?
There are 2 great optimization topics/tutorials that give good introduction to how to do it on this forum - did u read them maybe ?
Its really hard to just say what to click to drop down render times…
I would assign gray shader and optimize lights 1st. Then work on reflective materials, then transparent materials then GI settings - u should bake IRR every 5-15 frame depending on camera movement and same for LC. If you have moving object then either render them separably or use Brute Force to avoid flickering.
From experience I usually leave all vray settings to default except for DMC amount where I lower it from 0.850 to 0.650 or something like that depends on the scene. Also make sure ur gamma is properly set up or u will get noise.
Can u show us Sample info pass? That will tell us straight away how well ur scene is setup.
PS everything has Fresnel. Not having Fresnel ure breaking physics and can be exposed to VERY long render times.
Basically light cache (the secondary gi method) can be used to also speed up glossy reflection calculation. If you are using lightcache, then check those on and see if this helps.
Otherwise post part of your scene so people can poke at it.
Actually we never get flickering in our reflections when using this method, since retrace threshold relies on brute force a lot, its not very visible in any case.
Ure lucky I recently did animation and forgot about this setting. My floor was flickering all over the place as the retrace treshold was kicking only on contact with other objects and not in the middle of surface.
Folks, for interior fly throughs (think sci-fi film vfx) in the case of having moving objects in theory brute force and light cache or brute / brute is the way to go if you’ve got the render power. What’s the typical approach to these shots in LA / Vancouver studios?
You know everyone has different methods. Big studios have so much render power to throw at things, that it does not matter if your render takes 20 hours per frame
I recently did animation and forgot about this setting. My floor was flickering all over the place as the retrace treshold was kicking only on contact with other objects and not in the middle of surface.