Luxcore vs CyclesX Benchmark

Our forum gallery for displaying your finished renders.
User avatar
Sharlybg
Donor
Donor
Posts: 3076
Joined: Mon Dec 04, 2017 10:11 pm
Location: Ivory Coast

Luxcore vs CyclesX Benchmark

Post by Sharlybg »

Finally done !

Many artists just wonder wich one to pick between Cycles and Luxcore.
With the release of Blender 3.0 and Cycles X debut many just made bench
and compare previous Cycles version to the new CyclesX but Not only that.
We also witness the Blender Guru Benchmark between Corona Cycles X Luxcore
And Arnold Vray. Strangelly enought the result of blenderguru come out with lot of
controversies.
So after some encouragement from users in the Lux community I just gave it a shot
with an actual real world good size Archiviz project. and result are quite suprising
even for myself.
Make sure to watch the Video here on:
Youtube https://youtu.be/Qr9t8VaV8z4

DOWNLOAD THE SCENE HERE:

https://www.cgtrader.com/3d-models/arch ... nd-luxcore

Blenderartits Thread : https://blenderartists.org/t/luxcore-vs ... rk/1358059

Final 3K Render with Luxcore Gi cache:


Big_Lux_02_Denoised.jpg
Big_Lux_Denoised.jpg
Big_Lux_03_Denoised.jpg
Big_Lux_01_Denoised.jpg

A cropped View comparison of all engines mode
AAA_SUPER_CROPPED.jpg
Support LuxCoreRender project with salts and bounties

Portfolio : https://www.behance.net/DRAVIA
JulianoLisboa
Posts: 146
Joined: Sat Feb 22, 2020 3:29 am

Re: Luxcore vs CyclesX Benchmark

Post by JulianoLisboa »

Very good my friend, I have already posted in all the 3d forums and groups that I participate. Great job.
User avatar
Dez!
Posts: 364
Joined: Sun Apr 08, 2018 1:09 am
Location: Ekaterinburg
Contact:

Re: Luxcore vs CyclesX Benchmark

Post by Dez! »

Cool.
I asked Cycles users for their opinion on this, and they told me that you have to use Scrambling Distance for such scenes. This will speed up the Cycles so much that it becomes much faster than Luxcore.
Linux Plasma | Ryzen 5, 32Gb, SSD M2, GT 590 RX | BenQ 27 | Wacom One | Microsoft Ergo | Tie Guan Yin tea
http://dezigner.tilda.ws/
sarmath
Posts: 128
Joined: Wed Feb 07, 2018 1:22 pm

Re: Luxcore vs CyclesX Benchmark

Post by sarmath »

Dez! wrote: Mon Jan 31, 2022 1:12 am Cool.
I asked Cycles users for their opinion on this, and they told me that you have to use Scrambling Distance for such scenes. This will speed up the Cycles so much that it becomes much faster than Luxcore.
I think Scrambling Distance is not available in vanilla Blender... for that you need Bone Studio build or E-cycles.
It may produce artifacts if you put wrong values as well.
Manjaro Linux Ryzen 9 3900x GTX 1070ti 32 GB RAM
User avatar
Dez!
Posts: 364
Joined: Sun Apr 08, 2018 1:09 am
Location: Ekaterinburg
Contact:

Re: Luxcore vs CyclesX Benchmark

Post by Dez! »

Scrambling is available starting with Blender 3.0
Linux Plasma | Ryzen 5, 32Gb, SSD M2, GT 590 RX | BenQ 27 | Wacom One | Microsoft Ergo | Tie Guan Yin tea
http://dezigner.tilda.ws/
kintuX
Posts: 772
Joined: Wed Jan 10, 2018 2:37 am

Re: Luxcore vs CyclesX Benchmark

Post by kintuX »

sarmath wrote: Mon Jan 31, 2022 12:11 pm I think Scrambling Distance is not available in vanilla Blender... for that you need Bone Studio build or E-cycles.
....
It's in Blender v.3.0x (vanilla) || Render Settings > Sampling > Advanced > Scrambling Distance
V17
Posts: 4
Joined: Tue Feb 01, 2022 12:45 pm

Re: Luxcore vs CyclesX Benchmark

Post by V17 »

You generally do not need to set scrambling distance to get the best rendering times. You do however need to leave the "noise threshold" checkbox enabled (it's enabled by default), which disables the scrambling distance options altogether anyway. Noise threshold enables the new implementation of adaptive sampling, meaning that once a section of the image converges to a result with sufficiently low noise (the default setting is sane) it stops rendering.

This is ime the biggest optimization in Cycles X which reduced rendering times by up to 70% with some of my renders because only the dark and noisy parts would get rendered with a lot of samples, while large parts of the image would be done after a couple hundred.

Since this setting is enabled by default, the results of the benchmark should be accurate unless Sharlybg turned it off. Without it we would have a similar situation like benchmarking LuxCore without PhotonGI I think.
User avatar
Sharlybg
Donor
Donor
Posts: 3076
Joined: Mon Dec 04, 2017 10:11 pm
Location: Ivory Coast

Re: Luxcore vs CyclesX Benchmark

Post by Sharlybg »

I understand that people like the tools they are using everyday but at some point we need to be serious.

1) What Scrambling distance have to do with Cycles / E-Cycles / Corona in Blenderguru bench
be so down in Gi quality while other render engine correctly compute GI ?

2) Is Scrambling Distance responssible of the Fact that Photon Gi cache wasn't used in luxcore for This Blguru Bench ?

3) It only work on GPU and can cause artefact so doesn't apply To CPU

4) And even with that scrambling enabled if you just follow the explanation it isn't the reason of cycles bad perf here :
Scrambling.jpg
Now I have the noise plus the artefact thanks you. And even without the furniture and trees plus displacement geometry.
And this took 26 min with the simplified scene and noisy output.
Scramb.jpg
I have nothing against Cycles and actually use the engine for what it as been buid for.
You just have to accept also that opensource CGI solution can exist also outside of Blender
And it isn't a bad thing. Actually how much opensource production render engine project are still alive
Cycles aside ?
By the way Let see what happen if we manage to make Luxcore 3.0 a reality ?
Support LuxCoreRender project with salts and bounties

Portfolio : https://www.behance.net/DRAVIA
V17
Posts: 4
Joined: Tue Feb 01, 2022 12:45 pm

Re: Luxcore vs CyclesX Benchmark

Post by V17 »

I don't think anybody here is talking about the quality of LuxCore vs Cycles, the point only is to not do the same mistakes as Blender Guru did and use Cycles X with the correct settings, as other settings may needlessly slow it down. Which in this case, as far as I know, means leaving it in the default settings, in other words having "noise threshold" enabled and not touching the scramble distance at all. So if you left the settings in Cycles in default, I believe your results should be correct.
User avatar
Sharlybg
Donor
Donor
Posts: 3076
Joined: Mon Dec 04, 2017 10:11 pm
Location: Ivory Coast

Re: Luxcore vs CyclesX Benchmark

Post by Sharlybg »

Exactly the things that people forget in comparison is that quality and speed are quite interconnected.
And why should I enable scrambling if it is giving thoses
Artefact without removing the numerous fireflies.

I think the video was explanatory enought for anyone willing to get the point. The file is also available to anyone interested.

At the end it doesn't change the story.
Support LuxCoreRender project with salts and bounties

Portfolio : https://www.behance.net/DRAVIA
Post Reply