Page 1 of 1

Path tracing 12M-spp vs Photon tracing 6k-spp

PostPosted: Sat Feb 03, 2024 4:20 pm
by nuno1980
Hello, guys! :D

I tested a GeForce RTX 4090 with GRD 546.33WHQL driver on Windows 10 22H2. After 12M-spp at path tracing, this GPU got no any GPU fails.

--- Ice stone on glossy indoor with the roughness of 0.06 at our sun scene ---

-> Full-image at Photon Tracing kernel with 500 spp during alnost 2-min at GI clamp 1.0, at radius-0.09 and at 49-photon depth but no any variable caustics*(1):
full-image - 500 spp during alnost 2-min at GI clamp-1, at radius-0.09 and at 49-photon depth.png


-> 1st blue region picked at Path Tracing with 1M-spp during alnost 5h13min at full GI clamp, at 33-diffuse+49-specular depthes and no "caustics blur":
1st blue region picked_1M spp - 5h13min.jpg

- 5 cycles of number of spp from 1M to 12M (or 12 laps*(2) of 1M-spp):
5 cycles until 12M-spp at path tracing.png


---> the difference of 2 small images between Photon Tracing with 6k-spp and Path Tracing w/ 12M-spp:
photon tracing vs path tracing.jpg


At 1920x1080 for all images rendered without denoiser.

I confirm no any issues at more than 2M-spp at path tracing! :D

*(1) - Learn - HERE
*(2) - How to increase to more than 1M spp for region picked only -> HERE

Re: Path tracing 12M-spp vs Photon tracing 6k-spp

PostPosted: Fri Feb 23, 2024 9:06 pm
by nuno1980
I tested 7680x4320 for caustics at path tracing...

12M-spp@1920x1080 is as better caustics' quality as approximately 1.5M-spp@4320p.

New screenshot - this region picked on a RTX 4090:
4320p test -  2Mspp during ~1h.png
4320p test - 2Mspp during ~1h.png (163.8 KiB) Viewed 464 times

:D <3