Dade wrote: Fri Jan 25, 2019 11:22 pm
As far as I have seen, yes, there no particular problems with clamping.
Don't know if it is a clamping issue, but for the caustic cache clamp value = 1 is required and in my test scene caustics are muted and the light balance is very different when compared with a bidir metro render.
But for the first time we get fast caustics with correct SDS reflections, yay!
Also the innovative metropolis photon distribution works great, no problem with insane photon counts for smooth caustics. .
photoncachetest.gif .
cachetest_LuxCore.zip
It's rather easy to add to the addon, I'm just waiting until the parameters are somewhat stable (don't change every other day) and the whole thing is in a state that is usable by end-users.
I think the time has nearly come where this is the case.
B.Y.O.B. wrote: Sat Jan 26, 2019 2:51 pm
It's rather easy to add to the addon, I'm just waiting until the parameters are somewhat stable (don't change every other day) and the whole thing is in a state that is usable by end-users.
I think the time has nearly come where this is the case.
B.Y.O.B. wrote: Sat Jan 26, 2019 2:51 pm
It's rather easy to add to the addon, I'm just waiting until the parameters are somewhat stable (don't change every other day) and the whole thing is in a state that is usable by end-users.
I think the time has nearly come where this is the case.
I'm also exposing a lot of parameters for testing but most of them shouldn't be exposed to end users: for instance, we have direct, indirect and caustic caches but all the direct cache related stuff can be hidden (i.e. it will only be internally used for building indirect cache).
Support LuxCoreRender project with salts and bounties
Dade wrote: Sat Jan 26, 2019 1:32 pm
Added the multi-threads support for visibility particles tracing.
A 5 minutes rendering with the latest version:
test.jpg
About 10 seconds of pre-processing
BTW, we should (plan to) work on a DemoReel with 3-4 small ArchViz animations with scenes like this one: 5 minutes + denoising starts to be a viable rendering time, on Azure, it should be possible to render everything in one night.
Support LuxCoreRender project with salts and bounties
By the way, 5 minutes per frame is still pretty long.
If we say we want to render 1 minute of footage, that is:
60 sec * 24 frames/sec = 1440 frames
1440 frames * 5 min/frame = 7200 min = 120 hours = 5 days rendertime.
Or did I make a mistake?
By the way, I'm not sure we actually want 24 fps, it's really the bare minimum and forbids to make any camera panning or fast camera movement. 60 fps would be much better.
By the way, 5 minutes per frame is still pretty long.
If we say we want to render 1 minute of footage, that is:
60 sec * 24 frames/sec = 1440 frames
1440 frames * 5 min/frame = 7200 min = 120 hours = 5 days rendertime.
Or did I make a mistake?
By the way, I'm not sure we actually want 24 fps, it's really the bare minimum and forbids to make any camera panning or fast camera movement. 60 fps would be much better.
60fps in cgi looks usually quite fake imho. I'd say 30 is best and will be faster to render. I would also suggest non-denoised footage and remove noise with video editing rather. BCD needs too clean footage to work and I wouldn't be surprised to see all kinds of artifacts otherwise...
Also avoid abberation and image filtering so that the noise is fine and easy to remove with some cross frame denoising in video editor. If you still want abberation it should be done as a last step in post(I would still avoid it though, keep things clean)
