I think that is half of hour should be enought to clean the image,but it isn’t…Do i have to let it render a little bit longer,or do i have to set something in render?
What was the exact amount of passes and rendering time here?
If the quality of DOF is acceptable and you are getting noise only in GI, then you should increase GIvsAA samples (16? 32?). You probably followed the guide to decrease GIvsAA when rendering DOF - well, it depends on the kind of scene. You can also try lowering light samples multiplier to 1 (if it won’t introduce noise in direct light/shadows). Also revert UHD cache precision to 1 as with 2 there should be virtually no difference and it will probably be slower.
I think the noise is uniform, not only in out-of-focus areas, just you don’t have much diffuse materials in the focused area so it is not visible.
The “50 samples” in material settings refers to “rounded corners” option only. You can see a tooltip when you hover your mouse over the samples text field (as everywhere else in Corona).
I have better noise in second and third render but very bad in DOF…( Yes! I followed corona guide about dof.)
Time of render is the same, but number of passes is different.
If you want, you can try this scene when you had the time.
Maybe you can find better solution than I am.
Maybe I demand a lot of Corona for such a short time?!?!
You are not only using DoF, but also motion blur - I believe this adds some extra render time. You are also using Corona sun and no skylight. This may also add some noise as there is a huge difference between bright and dark parts of the scene. Disabling caustics slightly reduces noise too. I am checking the scene right now so I will inform you if there are any useful observations.
yes,yes…Maru told me about that sample for corners…Thanks again.
I tried to set default and set GIvsAA to 4 and disable caustic ,but nothing:(
Noise still here…Maybe little better ,but not clean!
render time is 30min about 300passes
After that I listened Maru’s advice and add corona sky,and disable motion blur…
30min without changes:( ,buy the way lighting is fantastic!Corona is really,really good!
This shouldn’t be a problem, especially if fresnel IOR is set to some low value.
It’s noisy because it’s a problematic scene - small windows, high contrast = GI problems + DOF + MB. You need a lot of GI and AA samples here. The biggest problems are with DOF/AA sampling on glass/metal (because it’s excessive) and sunlight bounces. I would say this kind of scene simply needs a lot of time to refine. I tried optimizing it yesterday, used buckets but the result was “acceptable” after about 1h. Still there was some noise on the glass teapot. So my advise would be to change lighting, change materials, set dof and mb to acceptable minimum.
