Direct Light Sampling Cache Test

Post your tests, experiments and unfinished renderings here.
Post Reply
User avatar
B.Y.O.B.
Developer
Posts: 1312
Joined: Mon Dec 04, 2017 10:08 pm
Location: Germany
Contact:

Direct Light Sampling Cache Test

Post by B.Y.O.B. » Thu Jul 26, 2018 7:07 pm

A quick comparison between the different light strategies at 1 Sample/Pixel.
The used strategy is noted on the bottom left in the statistics.

Note how the Rays/Sample value is especially low (good) in the DLSC image.
Attachments
img-0.png
img-1.png
img-2.png
img-3.png
spots.blend
(162.58 KiB) Downloaded 12 times
Support LuxCoreRender project with salts and bounties

User avatar
B.Y.O.B.
Developer
Posts: 1312
Joined: Mon Dec 04, 2017 10:08 pm
Location: Germany
Contact:

Re: Direct Light Sampling Cache Test

Post by B.Y.O.B. » Thu Jul 26, 2018 7:12 pm

The noise in the areas where multiple lights "overlap" per cache entry begins to vanish after about 70 samples:
Attachments
70_samples.png
Support LuxCoreRender project with salts and bounties

FarbigeWelt
Posts: 101
Joined: Sun Jul 01, 2018 12:07 pm
Location: Switzerland
Contact:

Re: Direct Light Sampling Cache Test

Post by FarbigeWelt » Thu Jul 26, 2018 7:57 pm

Very nice post. It is hard to wait for all the improvements.
Instagrammer, please join #luxcorerender for your renderings.

marcatore
Posts: 93
Joined: Wed Jan 10, 2018 8:04 am

Re: Direct Light Sampling Cache Test

Post by marcatore » Fri Jul 27, 2018 6:55 am

Surely I didn't understand the Samples/sec value. I understood that higher is better meaning that when the rays calculations are easy, the CPU (or GPU) can complete more rays calculation in the same time and it can sample more.

Nice to know that it's not like this.

User avatar
B.Y.O.B.
Developer
Posts: 1312
Joined: Mon Dec 04, 2017 10:08 pm
Location: Germany
Contact:

Re: Direct Light Sampling Cache Test

Post by B.Y.O.B. » Fri Jul 27, 2018 7:38 am

marcatore wrote:
Fri Jul 27, 2018 6:55 am
Surely I didn't understand the Samples/sec value.
No, you understood it correctly. I was talking about the Rays/Sample, which is computed as (Rays/Sec) / (Samples/Sec).
This statistic was not visible until recently. It tells you how long the light paths per sample are on average.
Short light paths usually mean less noise.
Support LuxCoreRender project with salts and bounties

marcatore
Posts: 93
Joined: Wed Jan 10, 2018 8:04 am

Re: Direct Light Sampling Cache Test

Post by marcatore » Fri Jul 27, 2018 8:59 am

Oh..sorry.. I've misread!

And I've gained a tip about scene efficency.

Thanks a lot

Post Reply