PhotonGI cache

Discussion related to the LuxCore functionality, implementations and API.
epilectrolytics
Donor
Posts: 703
Joined: Thu Oct 04, 2018 6:06 am

Re: PhotonGI cache

Post by epilectrolytics » Wed Jun 19, 2019 6:01 pm

Dade wrote:
Wed Jun 19, 2019 4:25 pm
It depends of the importance of light paths and eye paths: if you need to render 99% light paths and 1% eye paths, you should rendering with LIGHTCPU (not even BIDIRCPU or PATHCPU/OCL+PGI).
You are pretty much doing the test with a caustic covering the 99% of the image and a ("hyper")cube over a plane.
I'm not getting what you are referring to :?
It's about these artefacts (SDS-paths) which are caustics seen through (refracted) by the prism.
Screenshot3.png
Screenshot3.png (25.76 KiB) Viewed 1544 times
While the caustics seen directly converge nicely with cache update, these do not, their pattern remains stable through updating.
Probably I was wrong and it is no issue with visibility particles, I'm at a loss here.
Could you suggest PGI settings for the wallpaper scene that help avoid this issue?

Because if this appears here, it will likely do in other scenes too.

Good idea to activate clamping.
Dade wrote:
Mon Jun 17, 2019 10:34 pm
epilectrolytics wrote:
Mon Jun 17, 2019 8:23 pm
It could get complicated like this:
Rendering an animation with persistent indirect cache plus a cyclic caustic cache.
This should already work (see above).
I just tried indirect + caustic + periodic + persistent and get a crash within the first seconds (probably at the time of the first update).

User avatar
Dade
Developer
Posts: 4723
Joined: Mon Dec 04, 2017 8:36 pm
Location: Italy

Re: PhotonGI cache

Post by Dade » Wed Jun 19, 2019 6:08 pm

The visibility particles are not retrace at each update, you need to trace more of them at the start to solve that problem.
Support LuxCoreRender project with salts and bounties

epilectrolytics
Donor
Posts: 703
Joined: Thu Oct 04, 2018 6:06 am

Re: PhotonGI cache

Post by epilectrolytics » Wed Jun 19, 2019 6:12 pm

Ok thanks for confirmation.

Then we would need to expose these in Blender as an additional setting.

epilectrolytics
Donor
Posts: 703
Joined: Thu Oct 04, 2018 6:06 am

Re: PhotonGI cache

Post by epilectrolytics » Wed Jun 19, 2019 6:26 pm

epilectrolytics wrote:
Wed Jun 19, 2019 6:01 pm
I just tried indirect + caustic + periodic + persistent and get a crash within the first seconds (probably at the time of the first update).
Here is the file that crashes.
Rendering indirect + periodic caustic works, writing persistent too, reading persistent not.

First I'm moving the camera to the last frame #100 to write the persistent cache.
Then I disable "compute and overwrite" so that in the next render the existing cache is reused.
When I hit the animation button in Blender, rendering starts and an image appears, but after few seconds it crashes.
concrete4-.blend.zip
(261.83 KiB) Downloaded 55 times

User avatar
lacilaci
Donor
Posts: 1940
Joined: Fri May 04, 2018 5:16 am

Re: PhotonGI cache

Post by lacilaci » Thu Jun 20, 2019 9:34 am

Anyone had success with the caustic cache and something like pool water? Mine looks like trash.

User avatar
lacilaci
Donor
Posts: 1940
Joined: Fri May 04, 2018 5:16 am

Re: PhotonGI cache

Post by lacilaci » Thu Jun 20, 2019 9:46 am

So I've figured what's the problem.

this is what I see when looking from outside:
outside.jpg
and this is from under the water level:
underwater.jpg
wtf ? :D

Also, those caustics converge extremely slowly. I can imagine a full project render would be done in high res in +/- 20 minutes and then another 15 hours to get to see some neat caustics. Might as well just render bidir+metro and wait forever

CodeHD
Posts: 351
Joined: Tue Dec 11, 2018 12:38 pm
Location: Germany

Re: PhotonGI cache

Post by CodeHD » Thu Jun 20, 2019 10:23 am

lacilaci wrote:
Thu Jun 20, 2019 9:46 am
this is what I see when looking from outside:
I had this problem a while back, where I could only see the casutics cache from the outside if the indirect cache was active also.
It was shown in this post:

viewtopic.php?f=4&t=1076&start=10#p12639

But I don't remember if that was supposed to be fixed. There was also the issue at the time that viewed from sinide the water the behaviour was quasi opposite, but that had a different reason and was fixed.

I don
t know the settings you used here, but maybe it is because of that?

User avatar
lacilaci
Donor
Posts: 1940
Joined: Fri May 04, 2018 5:16 am

Re: PhotonGI cache

Post by lacilaci » Thu Jun 20, 2019 10:31 am

CodeHD wrote:
Thu Jun 20, 2019 10:23 am
lacilaci wrote:
Thu Jun 20, 2019 9:46 am
this is what I see when looking from outside:
I had this problem a while back, where I could only see the casutics cache from the outside if the indirect cache was active also.
It was shown in this post:

viewtopic.php?f=4&t=1076&start=10#p12639

But I don't remember if that was supposed to be fixed. There was also the issue at the time that viewed from sinide the water the behaviour was quasi opposite, but that had a different reason and was fixed.

I don
t know the settings you used here, but maybe it is because of that?
Nope, disabling indirect cache doesn't help. I think it's the same problem as observed by epilectrolytics looking at caustics through specular surface.

CodeHD
Posts: 351
Joined: Tue Dec 11, 2018 12:38 pm
Location: Germany

Re: PhotonGI cache

Post by CodeHD » Thu Jun 20, 2019 11:02 am

lacilaci wrote:
Thu Jun 20, 2019 10:31 am
Nope, disabling indirect cache doesn't help. I think it's the same problem as observed by epilectrolytics looking at caustics through specular surface.
Other way round, eanbling the indirect cache helped in my case.
But as usual, it is hard to judge the differences here anyways, just a suggestion cause this looked so familir, and the case presented by Epilectolytics might be different... I can't tell.

User avatar
lacilaci
Donor
Posts: 1940
Joined: Fri May 04, 2018 5:16 am

Re: PhotonGI cache

Post by lacilaci » Thu Jun 20, 2019 11:05 am

CodeHD wrote:
Thu Jun 20, 2019 11:02 am
lacilaci wrote:
Thu Jun 20, 2019 10:31 am
Nope, disabling indirect cache doesn't help. I think it's the same problem as observed by epilectrolytics looking at caustics through specular surface.
Other way round, eanbling the indirect cache helped in my case.
But as usual, it is hard to judge the differences here anyways, just a suggestion cause this looked so familir, and the case presented by Epilectolytics might be different... I can't tell.
Yeah I understand, indirect on or off doesn't change the situation. And I think even if it worked it is just way too slow anyways.

Post Reply