[DELETED]

DELETED

[Nuno Faria]

The emitter is at the end of the hall, and it has a value of 300,000 W/m^2 (!). The building has a reasonable scale (ceilings are 1.5 M). Here's the RC3 render after 5 minutes:

hi adam,

what values did you use for shutter speed and aperture?

[Nuno Faria]

maybe 1/10 and f/2.8 !?

btw: i couldnt download your file... could you please email it to me, so i can give it a try.

thanks in advance,
nuno

[Frances]

Adam,

I've set up a scene similar to yours, approximately the same dimensions in meters, with a custom emitter at 300k W/m^2 (Exitence in Rhinoll). Here are my findings:



There is something seriously wrong with the new engine, and I noticed it from the start. I was hoping that RC3 would have a fix. What concerns me more is that someone might think there is nothing wrong here.

[michaelplogue]

Adam -

Just curious, but how does the preview look with your setup?

.

[rivoli]

there sure is something wrong, almost every rc render i've seen looks like a poorly sampled, splotchy and dark qmc solution (or as if it was a poorly calculated photon map with artifacts and dark corners). oskit said they are aware of this though, and they already sorted out similar problems when the first betas came out pretty quickly. i'm confident they'll do the same this time.

[Frances]

Adam -

Just curious, but how does the preview look with your setup?

.

Not Adam, but I get a result similar to my other RC3 examples in preview.

[3dtrialpractice]

Adam,

I've set up a scene similar to yours, approximately the same dimensions in meters, with a custom emitter at 300k W/m^2 (Exitence in Rhinoll). Here are my findings:


There is something seriously wrong with the new engine, and I noticed it from the start. I was hoping that RC3 would have a fix. What concerns me more is that someone might think there is nothing wrong here.

Wow fran, thanks for that "arond the corner" shot

This totaly go with the theories that ppl talk about not enogh bounces in the scene....
WOW.. I mean thats stil just crazy that the ball renders so dark and the halway is so bright...

If you ask me I would love, and have Longed for the Bounces settings we had in ALPHA

plus this rc engine keeps giveing thos terrible "sparkles" all over too..
But your test is a big eye opener!

Adam... An chance you could do a quick render like fran
s from around the corner, I wanna see if there are any inconsistencies between the two plugins(rhino and c4d)- since the rhino plugin has a date of DEC05 and the C4d is DEC07 date...

-L

[Tyrone Marshall]

I have not run into this one yet, but I will play closer attention. I did notice one of my scenes not being bright as it should. I closed studio, reloaded it and it was fine again. I also remade a diffuse material for some reason that along with something else was causing dark scene.

I will be more observant on this one as I think I did encounter it, but the combination of new diffuse material and restarting studio took care of it.

[SPUTNIK3D]

A test with a 6 years old software, still champion of speed,stability and hight quality images. The fastest OpenGL real time visualization system, fully texturized.
Please NL, take a look in the lightscape openGL engine.
http://paginas.terra.com.br/negocios/sintese/

[SPUTNIK3D]

Those are certainly the crispest, sharpest, clean looking red X's I have ever seen

I was able to see the pics from this link:

http://paginas.terra.com.br/negocios/sintese/

Thanks maxbish.

[iandavis]

As a photographer I obviously work with light a lot and the real world (assuming that maxwell 'simulates' reality)
I would expect to find that kind of behaviour if the hall were several hundred feet high. Light is reduced exponentially, that is imagine it being four times darker for every unit... It would terminate as abruptly if the scale were way off.

There is most likely a scale issue here. Even if it's a numbers problem, I am willing to bet that if you reduce the scale in maxwell by a factor of at least 10.. you should see more realistic results from exactly the same power emitter.

Also, another possibility is the texture. It's not beyond the realm of possibility there is some problem with the way the numbers for texture colors are translated into the spectral space. If the material were unnaturally light absorbant, imagine absorbing as much light as the color red, but LOOKING like white. After only a few bounces all of the light would be absorbed... where in the other renderer, the colour/bounce% ratio would be correct, thus realistic results.
On this front, try using a 190,190,190 grey bitmap on the walls, though this may not bypass the bug, assuming there is a bug.

cheers

[SPUTNIK3D]

Lighstcape really was a thing of wonder. What a shame Discreet had to ruin it. Imagine how good it could be if it had been under development the last four years.

It's true Adam. But Lightscape was cowardly murdered by autodesk thugs!
Can you imagine where are the Cornell guys (the creators)?
NL, find them!

[Tyrone Marshall]

I am can at the moment confirm the dark renders. I am rendering an old scene from Cinema 4D and it is very dark. Cannot figure it out. I will try changing scale.

Okay, I have retested this scene to just see what the initial results are like, and it looks fine, I think the dark renders have something to do with using older scenes (beta materials) - somehow all of that is not yet connected. It looks like it may be best for Cinema 4D to still do renders from Studio after mxs export.

Newly configured scenes (material) seem to be fine under RC3. Its the ones from the plugin that are triggered from the mxcl engine.

[Frances]

I have not run into this one yet, but I will play closer attention. I did notice one of my scenes not being bright as it should. I closed studio, reloaded it and it was fine again. I also remade a diffuse material for some reason that along with something else was causing dark scene.

I will be more observant on this one as I think I did encounter it, but the combination of new diffuse material and restarting studio took care of it.

These were not rendered with the Studio. Taking the scene into Studio and applying new textures to the surfaces made no difference.

Ian, I also tried rendering this scene in the Studio and using a scale of 0.1. It makes no difference.

The GI algorithm is broken. Either that, or it lacks any meaningful communication with the user input controls. When MW passes this test, I will consider it to be a viable engine again. I'm amazed that anyone could actually put in a readme doc that the render engine is much improved .

I'm not going on about this just to be mean. But I think Adam and I have proven that the engine is broken in a major way.