Re: LuxCoreRender v2.2beta2 released
Posted: Sun Jun 16, 2019 1:41 pm
It works, thanks!
Show your work, get help, participate in development
http://forums.luxcorerender.org/
The only reason I haven't yet removed TILEPATH render engine ... and I want really badly to remove that code, it is literally useless in any other possible case but very high image resolution rendering.kintuX wrote: ↑Sun Jun 30, 2019 9:11 am Well, i didn't went for 16K yet...
but did Sharlybg's "Food" Scene with Tiled Sampler, tiled OIDN...
all in all roughly 50min @ 10K! on Q M5000 & GTX1070
Is that part of the code that bad? Is it lousy sampler? Causing regression in a way? Unoptimised? ... Do you have an alternative in mind?Dade wrote: ↑Sun Jun 30, 2019 9:57 amThe only reason I haven't yet removed TILEPATH render engine ... and I want really badly to remove that code, it is literally useless in any other possible case but very high image resolution rendering.kintuX wrote: ↑Sun Jun 30, 2019 9:11 am Well, i didn't went for 16K yet...
but did Sharlybg's "Food" Scene with Tiled Sampler, tiled OIDN...
all in all roughly 50min @ 10K! on Q M5000 & GTX1070
I agree with you. I'm and a lot of users are still using Tiled PATH (a.k.a. Progressive refine in Cycyles). It is really faster in both case: in CPU and in GPU rendering. I think you must re-think again. This is just checkbox option, you knowkintuX wrote: ↑Tue Jul 02, 2019 8:41 pmIs that part of the code that bad? Is it lousy sampler? Causing regression in a way? Unoptimised? ... Do you have an alternative in mind?Dade wrote: ↑Sun Jun 30, 2019 9:57 amThe only reason I haven't yet removed TILEPATH render engine ... and I want really badly to remove that code, it is literally useless in any other possible case but very high image resolution rendering.kintuX wrote: ↑Sun Jun 30, 2019 9:11 am Well, i didn't went for 16K yet...
but did Sharlybg's "Food" Scene with Tiled Sampler, tiled OIDN...
all in all roughly 50min @ 10K! on Q M5000 & GTX1070
Because without a simple "click the check box" option it takes a day to get 10K denoised image doing it the old school way (tiled animated camera then stitching tiles in image editor) for experienced user. New comers will simply turn away. Demands for large scale imagery grows in archviz & carviz arena - 8K becoming the norm.
W/o bucket rendering, it's quite a problem for Corona dev. (ie. 30k render (strip renders) do not match near shadow catcher plane, Increase of rendertime with size of render, Render Big Resolution...)
So you're running a marathon, suddenly have a huge advantage, but because you're alone in front you wanna shoot yourself in a leg... ah, well. Please, visit other forums, check user requests, demands & think about it. Buckets still present huge advantage, especially for serious production (VRay, Arnold, PRMan, RedShift, Cycles...).
& You also wanna cut off the Light Groups... oh, man This community is lacking more pro-users to give you force, motivation, high, healthy spirit to go on. You're making one of the greatest open source engines ever! but is lacking some very fundamental features (displacement, simple material settings as was before in Blender with classic Lux, Texture sampling filters, then Classic GUI - Standalone app alike Studio...)
I really hope you see where you're at, cuz for many artists i talk with, show the capabilities, this stuff is beyond their wildest dreams.
According to your links, the pictures are from Corona page. Please use a note and limit the number to max a handful of examples.