My Renderings in KT2008 Echo
-
These three were done last night in the new KT.
The first was done using preset 06.PM-Medium+AA0.3. Render time was 2 min 17 seconds.
Next at Preset 07.PM High+AA0.3 in 6 minutes 39 seconds.
And the third was done with Preset 20.MLT (BPT) in 7 hours, 50 minutes and 43 seconds. I did this one rather than shutting the computer down when I went to bed.
The first two were done while I was checking e-mail and otherwise occupying the computer with stuff. In both cases I was amazed at how rapidly the renders were completed. the major difference I see between them is some jaggies along the floor near the wall and some graininess near the back on the upper face of the seat in the PM versions. My guess is going to a higher PM preset would take care of that. MLT (BPT) did a little nicer job on the floor and shadows.
Over all from my non expert point of view, the first two would be acceptable renders to show most people. They wouldn't notice the little differences.
-
@dave said:
Over all from my non expert point of view, the first two would be acceptable renders to show most people. They wouldn't notice the little differences.
To tell the truth even after you explained what I should see as difference, I cannot really see them. I know MLT is supposed to be much better for interior renderings but maybe that's just true with more "complex" scenes?
I don't know. Maybe I'm just guessing here (and they all are nice renders for a test!)
-
Those look great Dave. I was playing around with the new version last night as well. It's really a credit to the KT team that I was able to go in and do some pretty decent renderings without much knowledge of the program or previous experience with it.
-
It's interesting how the handling of shadows is different between PM and MLT (BLT). The foreground shadows in the PM renders is almost absent.
-
@Gaeius,
I have to agree that it looks as if a MLT (unbiased) rendering in Kerkythea converges to the exact same output one would get by using the default biased settings (PM+FG).
The only difference I see (when rendered with MLT for a loooooong time) is better AA (antialiasing).Somehow I would expect better quality from an unbiased solution than a biased one, but in my tests it is (almost) exactly the same.
-
@kwistenbiebel said:
Somehow I would expect better quality from an unbiased solution than a biased one, but in my tests it is (almost) exactly the same.
Funny thing. It just means that with Photon Mapping and proper setup you can achieve almost same result as with unbiased method. It just proves the quality of PM solution in KT.
In most of simple scenes PM will be sufficient and it is not worth to invest time in MLT.Tomasz
-
It's really much up to the scene how results are if one compare PM to MLT (PPT or BPT or MLT/BPT). One will spot differences with unbiased methods, if you made the scene a bit more challenging. These stools, do not give imho any advantage to unbiased rendering, materials seems to be set fairly well, so those wont produce artifacts in PM, lighting is direct and simple, no aluminum or glass, no caustics. I would not even bother to use anything else than PM with a scene like this - here it is a clear winner. Fast and clean result, KT PM & FG is simply good.
Try to render something like Tadao Ando's Church of the Light with real windows and good materials - then you will find greater differences.
Anyhow - nice stools, a bit more work with floor....
-
Vray and MentalRay are also, under normal circumstances and use, 'biased'. But I do not expect them to result in lower quality than other more popular 'unbiased' engines.
In fact, Mray and Vray are so popular in the industry because of their amazing speed and quality... kinda why I like them and KT too.
'Unbiased' does not mean 'better' it means 'unbiased'. In fact, Giannis (developer of Kerkythea) stated
@unknownuser said:here's a small lesson about unbiased rendering:
-
Unbiased rendering was first introduced back in 1986. Yes!
It was not introduced by a commercial renderer. Just marketed
nicely (but this is just my opinion). -
Writing an unbiased renderer is simpler than writing a biased
one! Yes! Developing a path tracer or even Kelemen's MLT is easier
than writing photon mapping. -
There are many more unbiased renderers that you can think off!
Virtually all ray tracers can turn to unbiased rendering relatively
easily. Sunflow for example can operate in path tracing mode.
Overall! I see users reproducing the same information found in
commercial leaflets but never digging a bit more in theory or.. history! Winkbest regards.
The idea of a 'biased' render is that it makes some assumptions, and tries to give you what it thinks you want... it's, you know, biased... in a good way.
The idea of an 'un-biased' render is that it calculates bounces of light until they stop boucing... so if you have bad materials, this will 'bias' an 'unbiased' render engine in a bad way... but with a 'biased' render like KT's PM+FG it will be more 'forgiving'... especially with the higher quality presets. This is because Giannis and the team have put a lot of work in getting them to give a less-render-experienced user quick and easy presets.
The presets are not, of course, meant to be a silver bullet, but they are meant for you to load and learn from. Nor are the lower or medium settings meant for client presentation (though I do it all the time because, as Dave pointed out, they don't know or probably even care about the difference because KT's PM+FG is an amazing machine).
BiPT and MLT and MLT/BiPT and PTP are meant for a scene with less direct light...
MLT cranks very well on scenes lit completely by indirect light... but of course, so does PM+FG.. and with the right tonemapping, well... it's great...(Vray and Mray, as far as I can tell, have the right tonemapping, etc. already built in and transparent... KT is working on this...)
BiPT handles caustics (like sparkles onto a wall from an adjacent water surface) well, depending on the light source, in my experience, and PTP is much faster with exteriors lit with only sun and sky... so each have their advantages.
You can't expect to learn a sophisticated program like this in a day or even a year. You can get great results in a day, and better in a year... but I still learn new things all the time. that's part of the fun.
May want to read more at comparison thread between many engines...
-
Advertisement