Indigo Renderer 2.4.10 Released

[Prof4D]

Excuse but I do not have time to be engaged in it. I have simply compared two identical scenes and have specified you concrete result. And all manipulations of a scene in the Photoshop the identical. The scene difficult enough that that has flown down all matte. Even in 11 hours there is a noise, I think it long. I wait when render it will be possible to keep normally in a file and then to continue рендер as it is made in Makswelle. In the scenes I use only MLT with BiDir, as it the most realistic but it purely my opinion.
I would like to have concrete scenes where concrete improvement is shown.

[Zom-B]

Post some other model to show the BiDir bug then :P

I never told there is a bug (!) only speed doesn't seemed for me to be as fast as I thought it should be! But I'll try to create a simple but (light-wise) complex scene to demonstrate and benchmark!

Excuse but I do not have time to be engaged in it.

Totally understandable!
User != beta tester! :)

I have simply compared two identical scenes and have specified you concrete result. And all manipulations of a scene in the Photoshop the identical.

Well, yes but we are interested in pure Indigo output, because only there we can see issues best! (just like in the image you posted at last!)

The scene difficult enough that that has flown down all matte. Even in 11 hours there is a noise, I think it long.

Yes, I have to admit that after 5700 samples your scene is still quite noisy :/
If I should guess why this is so, I would point my finger @ light Strenght of a/some emitter in your scene!

Indigo weights importance of light Emitters for casting rays by the streanght of light they emit!
For example: A lightplane on the left that iluminates 100 stronger then a lightplane on the right.
The left light plane get 100times more priority then the right one!
Basically this is less a problem, because the right Light is also a lot weaker...
But if you use Rainhard tonemapping or some IES lights, this may become a issue!

How to check this (quite) easy: start a render with lightlayers for all your lightources (by material!).
After 30min of rendering you already should see (by cecking each Lightlayer for itself) if one of them is the evil basterd that fills the room with noise.

[Zom-B]

I'll try to create a simple but (light-wise) complex scene to demonstrate and benchmark!

Ok, here it is: 6 cubes of glass inside each other on a plane lit by a EnvMap.

Pure MLT has here 61% more samples and image looks cleaner (PNG filesze 724kb vs 678kb)

**EDIT**
I added also some PT & PT+BiDir test with same tendency regarding speed & noise for this scene

[Prof4D]

Thanks Zom - b that help me to understand with a problem.
I have found in what a problem, which I have applied Material Ljuminositi not indigo and is pure
Material of Cinema4d, after material replacement all has settled into shape. But then it is strange whose then this error. And concerning tests I with pleasure would receive the working indigo version in exchange for works for Indigo advancement. I am registered on Evermotion a site and too participated in competitions but a problem in the licence. I cannot expose anything from the works as the size of the image does not allow to make it.

Xeon 5506
Workstation Board S5520SC
6GB RAM

[Zom-B]

Yo Prof,

The way Cindigo converts Materials into Indigo materials is quite robust, but has problems with tricky material solutions. I really would only use the material converter to start off with own & adjusted Indigo materials, because only this way you have full control, and can create great blended Materials with Indigo related special values and functions.

The point about the low res of the Demo & Evermotion Competitions... well... its a shame.

But maybe you are someone that can apply for a cheap educational version.
Another great possibility is to win the actual Indigo Competition :)
Here you can show your skills!

Last but not least there are plenty of users with Licensed Indigo version who maybe would render your final scene for you!
Glare for example could be interested to offer you this and therefore keep your scenefile for internal testing & optimizations... why not ask :)

Since we seem to hijack this thread I just added two images on my last post: Pt & PT+BiDir ;-)

[Prof4D]

About competition of thanks I will try.
And who such Glare if the help as to me to communicate is required to me.
In advance Thanks.

[lycium]

About competition of thanks I will try.
And who such Glare if the help as to me to communicate is required to me.
In advance Thanks.

We would indeed be glad to check out that scene, it's quite unexpectedly noisy.

[fused]

I'll try to create a simple but (light-wise) complex scene to demonstrate and benchmark!

Ok, here it is: 6 cubes of glass inside each other on a plane lit by a EnvMap.

Pure MLT has here 61% more samples and image looks cleaner (PNG filesze 724kb vs 678kb)

**EDIT**
I added also some PT & PT+BiDir test with same tendency regarding speed & noise for this scene

yo dawg i herd you like cubes...

[Godzilla]

yo dawg i herd you like cubes...

so we put a cube inside a cube inside a cube inside a cube inside a cube inside another cube so you can render a cube while you're rendering your cubes.

[Borgleader]

yo dawg i herd you like cubes...

so we put a cube inside a cube inside a cube inside a cube inside a cube inside another cube so you can render a cube while you're rendering your cubes.

LOOOOOOOOOL

[Zom-B]

Yes, hollow cubes with 2.5mm thick glass walls inside each other... I thought this could be some somehow "complex" ^^

Just tested today the difference between RGB & uniform for glass in this scene today, and my result is that uniform is faster here for 1,9%.
Not much, but I take it... for free!

[PureSpider]

Post some other model to show the BiDir bug then :P

I never told there is a bug (!) only speed doesn't seemed for me to be as fast as I thought it should be! But I'll try to create a simple but (light-wise) complex scene to demonstrate and benchmark!

I was referring to the massive "black edges on glass" bug seen before Indigo 2.4.10!
And that indeed was a bug, so I asked for a comparison between "old" and "new" BiDir :D

[Whaat]

Ono, thanks for finally squashing that evil black edge glass bug!