SSS lighting: Indigo vs. Famous Commercial Unbiased Renderer
SSS lighting: Indigo vs. Famous Commercial Unbiased Renderer
deleted by user (sorry)
Last edited by ave on Fri Feb 20, 2009 11:12 pm, edited 2 times in total.
interesting results you got here.... anyway the commercial software result seems to be to wrong to be the way it works, maybe check out on thiar forums for explanation!
BTW: since Indigo 1.1.15 there is a slightly change in SSS, maybe you just can start a test with the actual 1.1.16 and post the result!
BTW: since Indigo 1.1.15 there is a slightly change in SSS, maybe you just can start a test with the actual 1.1.16 and post the result!
polygonmanufaktur.de
I think it would be interesting to submit the same test to an experimented M. (:)) user, it's difficult to compare unless you are experienced with both softwares.
And, I can hardly believe that there is no "workaround" for the "issue" of the other render. If you can confirm that you are experienced with the culprit of such a bad rendering performance, I'll remove the quotes
And, I can hardly believe that there is no "workaround" for the "issue" of the other render. If you can confirm that you are experienced with the culprit of such a bad rendering performance, I'll remove the quotes
obsolete asset
Re: SSS lighting: Indigo vs. Famous Commercial Unbiased Rend
Huh?ave wrote:deleted by user (sorry)
-
- Posts: 1828
- Joined: Mon Sep 04, 2006 3:33 pm
test reloaded
ok, after messing around a bit here we are with this test, now in (almost) full respect of the law .
The test scene is about light emitted from a small source, passing through a SSS volume (a kind of window) and lighting a floor and a wall. You can see the attached scheme. In this test I am interested in these two things:
a) how fast and accurately is direct light rendered. By direct light I mean light which has passed through SSS without significant scattering. In the scene it is the bright square spot on the floor.
b)how fast and accrately is INDIRECT light rendererd, where inidrect light is light which HAS been scattered around and it is lighting the scene all around.
The test features Indigo 1.16 (MLT transport) vs. Maxwell renderer 1.7.1 (Demo ). Each render has been computed for 12 hours on my MacBook Pro. Indigo runs under WINE, while Maxwell is in the native OSX version.
These are my remarks: Indigo wins by far on the direct light. In a few minutes the spot on the floor is completely cleared, while Mxl is significantly noisy even after 12h.
On indirect light Mxl is apparently much faster. Seems that somehow it is forcing the algorithm to choose paths related to scattered light first (but I am no expert of this stuff). What I don't like in Indigo is that scattered light (which should be white since scattering is uniform and equal for RGB) has some tint, which however is gradually disappearing.
Overall Indigo performs better to me here.
Final note: I am no Mxl expert, but I don't see which rendering parameters could be changed there to speed things up.
The test scene is about light emitted from a small source, passing through a SSS volume (a kind of window) and lighting a floor and a wall. You can see the attached scheme. In this test I am interested in these two things:
a) how fast and accurately is direct light rendered. By direct light I mean light which has passed through SSS without significant scattering. In the scene it is the bright square spot on the floor.
b)how fast and accrately is INDIRECT light rendererd, where inidrect light is light which HAS been scattered around and it is lighting the scene all around.
The test features Indigo 1.16 (MLT transport) vs. Maxwell renderer 1.7.1 (Demo ). Each render has been computed for 12 hours on my MacBook Pro. Indigo runs under WINE, while Maxwell is in the native OSX version.
These are my remarks: Indigo wins by far on the direct light. In a few minutes the spot on the floor is completely cleared, while Mxl is significantly noisy even after 12h.
On indirect light Mxl is apparently much faster. Seems that somehow it is forcing the algorithm to choose paths related to scattered light first (but I am no expert of this stuff). What I don't like in Indigo is that scattered light (which should be white since scattering is uniform and equal for RGB) has some tint, which however is gradually disappearing.
Overall Indigo performs better to me here.
Final note: I am no Mxl expert, but I don't see which rendering parameters could be changed there to speed things up.
- Attachments
-
- indigo render
- im1235252371.png (344.15 KiB) Viewed 9417 times
-
- maxwell render
- scat_compar.png (288.85 KiB) Viewed 9416 times
-
- scheme of the test scene
- scheme.jpg (26.53 KiB) Viewed 9420 times
Re: test reloaded
sorry, but to me this is, if this is really the best what one can get out of Maxwell, a perfect win for indigo - with a bad ass fatality!ave wrote:Overall Indigo performs better to me here.
that the indirect light on the wall is nowhere near to sufficiently sampled is a confession of failure.
(sorry Maxwell im sure you have other strengths )
Yes... I totally agree with you. And really I can't imagine what I could have done wrong or not optimized in Maxwell...That the indirect light on the wall is nowhere near to sufficiently sampled is a confession of failure.
BTW, there is another point which I forgot: with Indigo I set the emitter to 500 W, 14% efficency and it worked properly.
With Mxl I had to increase efficency to 100% to see some significant light (and it's still too low). Bearing in mind that the whole setup is a couple of meters large, I am sure Indigo is right and Mxl wrong (try to put a 500W light in a small room if you don't agree )
Who is online
Users browsing this forum: No registered users and 37 guests