Cycles-X

Did a quick test on cycles 2.92, Kcycles 2.92, and Cycles-X 3.0

720p , 2000 samples, no denoising

Cycles-X

Kcycles

Cycles

Cycles-X lighting looks a bit different than vanilla but overall for a barebone prototype demo this looks really promising. Coming from Octane, I really miss progressive rendering. Can’t wait to see what Cycles-X ends up being by end of 2021.

7 Likes

Wow you got big time savings!
Is there anything special about your scene/materials setup?

Not really, I think all the materials use the Principled BSDF, except for glass, and the lighting is made of an hdri for the exterior and a couple of area lights for the interior.

1 Like

So I finally got around to some testing, and used three of my own scenes for comparison.

With CPU only rendering, I noticed little difference in time.

GPU on the other hand, does show in some cases some improvement in a same hardware environment (I have dual GPUs, but for comparison on two of the scenes, rendered using classic rendering on both one and two GPUs since cycles-x doesn’t yet make use of multi-GPU rendering.

On one of my simpler scenes, cycles-x outperformed single GPU classic cycles, but fell behind dual GPU. ON a scene of a character headshot, with hair and SSS, cycles-x outperformed the dual GPU classic render by around 1/3 time.

Scene three, however, is a complex indoor scene, with complex materials. It originally had volumetrics, but I dumped those a few weeks ago for adding rays in the compositor. I deliberately chose this as a test because it is a long rendering scene.

Dual GPU classic renders the scene in 01:19:30.87.

Cycles-X renders the scene in 00:26:06.82. (I put part of the improvement down to lack of caustics in Cycles-X).

Nevertheless, when Cycles-X gets multiple device support, it will blow classic out of the water with no way back.

Yes, there are some differences in lighting when doing a direct side by side comparison, but again, this may be down to caustics.

However, as Cycles-X matures, and particularly the missing features in the roadmap, it will be awesome.

At the moment, Optix render time denoising with Cycles-X is less satisfactory than classic, though OIDN (again, render time rather than post) works just fine, but I expect that will improve, too.

7 Likes

I can’t get this to work on my 1080Ti.

I use Cuda in my system settings as optix doesn’t work. I just get no render result. Scene goes through whole process of loading bvh , calculating normals and samples are processed but no render result in the end.

Render does work in view port but cpu only. When I select gpu the image is cut off…

Hi, I too have a scene it does not work, does default cube render?
Please add your OS, version.

Cheers, mib

I use windows 10 …I got the default cube to render and I think I found the problem…in render properties panel under final render… I deselect save buffer. Thanks for the advice to just render the default cube.

Will try a more complex scene now.

just an update its working now…just did 50 samples and seems quite a bit faster

1 Like

Yeah, this is like forums should work, I can render my complex scene, too.
Save buffer was the issue.

Cheers, mib

2 Likes

Just wow.

Cycles 2.92, 256x256 tile size - 16:32
Cycles X - 8:41

Optix, adaptive sampling is off, rtx 2060 super, 2048 spp

12 Likes

OMG!!

It’s Super faster than before

12 Likes

I’ve been testing it for a bit as well, and it’s kinda annoying that it is so much faster :slight_smile: I don’t want to go back to even the regular 3.0 alpha now because of the speed. But the fact that ambient occlusion for optix is missing, and also volumetrics makes the choice kinda hard. Really looking forward to having this in the main branch. Keep up the good work Brecht and Serge!

14 Likes

I rendered several scenes in 3.0 CyclesX and now I’m going back to render them in 2.92 Cycles and they all “feel” faster as I watch the paint dry even though it’s 2 times slower. There’ just something about watching tiles render that satisfies me. As a test I rendered with cpu and it felt significantly more satisfying watching the tiles snake around (although I had to drop samples to get a similar render time).

5 Likes

I know that feeling bro :wink:

2 Likes

I love watching tiles too, especially the little 8 px :rofl:
I am a little disappointed here because I am restricted to CPU (GPU is no go)
At least there is a bit of difference though :slight_smile:
1000 samples, open image de noise.

Edit: 3.0 alpha 19m 06s
2.92 20m 48s (automatic tile size)

I rounded the seconds.


I can not see any difference in the renders.

Edit:
Out of interest has anyone compared to E Cycles?

Edit 2: I just noticed the difference in memory and peak, maybe due to the tiles.

2 Likes

And what is the time you got with Cycles X build?

Just in case you got confused, at the end of the first post in this thread is the link to download cycles-x build (labelled as “Blender 3.0.0 cycles-x branch”)

3 Likes

Don’t you mean “Cycles-E”?

2 Likes

Yes Sorry it was 3.00 I wrote it wrong and have corrected.

1 Like

I meant this one

I know just joking.

4 Likes

Ha ha, I just checked that thread and they are already comparing.