Maxwell Render | Official public forum

Hi all

I designed a light object some time ago which I want to stage in a nice scene. Once again I cant set the scene up in a way that will render even tiny pictures without noise in reasonable time.

When looking at other WIPs (and the time in which re-dos are presented) I get the impression that I somehow always do it the wrong way. Of course I read the available info in the manual and on the forums on what to do and what not to do in setting up materials and scenes.
I tested a dozen methods like 3-times as big rendering and downscaling, hide from GI, turn off indirect refractions/reflections etc. But nothing really helped...

This picture rendered 8h on a k2600 and reached SL22. It´s way from beeing cleared.


This ist a test without the difficult parts of light object which of course is the reason for all the noise. again 8h on k2600 with SL23. Perfect (might have rendered even shorter)


I´d also provide the scene but I am unsure cause there are arroway textures and other licensed stuff in it. How do you guys proceed in such situation? Render the texture unusuable e.g. by heavy blurring?

Any help greatly appreciated
Rogurt

What is that stuff hanging from the light fixture? I mean what is it "supposed" to be, and what does the MXM look like(could you share that one)?

Best,
Jason.

Hi Jason

those are pieces of synthetic cloth like with lampshades or those draperies with lamellas seen often inn offices (hope you understand my crappy English).
If there´s some correct way to do it I´d provide the mxs. Maybe like I mentioned a strong blur on the licensed textures? At least the overall color and lightness of the texture would be kept (for to check on failures here). Or a huge red cross over the texture?

The mxm of these pieces uses thin sss. Since it uses a texture from the gallery I suppose I can share it here?
http://www.william-associates.com/tmp/lamella_mxm.zip

Best
Rogurt

Try this edit of the material -- ideally what you would want to do is use a true SSS material and model each strip of fabric to real world scale... however I know that is not a very practical thing.

Best,
Jason.

Using that mxm the render was perfectly clear after one night (stopped it after 7,5h). most likely it could have rendered even shorter.
I worked my way through your tweaked version of the mxm. Am I right that the main failure was that I forgot to clamp the high rgb values in the base refl 0 channel? You left the texture in the ghost sss (which I tried to set up like advised in the "mxm tip of the day" topic) at full brightness. I clamped the whites severly (according to the "dont use 255 values" tip). Did you do that because in the end the opacity map of the entire ghost sss layer leads to a darker output?



Thanks a lot so far!
Rogurt

P.S. Could I possibly ask you to have a look at the entire scene and give me some more thoughts? I know "give someone an inch..."

Rogurt wrote:Using that mxm the render was perfectly clear after one night (stopped it after 7,5h). most likely it could have rendered even shorter.
I worked my way through your tweaked version of the mxm. Am I right that the main failure was that I forgot to clamp the high rgb values in the base refl 0 channel? You left the texture in the ghost sss (which I tried to set up like advised in the "mxm tip of the day" topic) at full brightness. I clamped the whites severly (according to the "dont use 255 values" tip). Did you do that because in the end the opacity map of the entire ghost sss layer leads to a darker output?

A high Saturation or Value (or both) in Reflectance 0 is always the first thing I look for when a material is noisy (or blown-out) -- and since we are using additive blending mode here that is doubly true.

The opacity map mitigates a bit of the additive mode issues since the entire surface is not uniform -- a uniformly additive material near white would be a very bad idea and will lead to excessive noise. But more than that it also enhances the results of the thickness map to indicate there are areas of the cloth where light can pass through and areas where it cannot.

Mostly I had two priorities with the edits I did here:

1. Reduce the noise
2. Enhance the feeling of a cloth based material

Also having a preview scene (like the one I used here) to torture test such materials can save you alot of time in the final renders.

Rogurt wrote: P.S. Could I possibly ask you to have a look at the entire scene and give me some more thoughts? I know "give someone an inch..."

Sure -- do a pack and go and zip the whole thing up... you can PM me with a link.

Best,
Jason.

Jason, why would I want to use real sss? I assume that rendertimes would get astronomic in this case. The geometry is all solids with thickness so I could use sss right away...

Cheers
Rogurt

ThinSSS is a hack of the material system and is (IMO) not really physically correct -- the virtual thickness parameter tries to make up for this but since it does not create any real thickness it is only a poor visual approximation of what an object with true SSS (with correctly modeled details) would give.

If you have the choice of using real SSS materials I would take it over thinSSS 99 times out of 100.

IMO the only real advantage thinSSS has is you can map the scattering color -- which is useful in things like human skin and leaves, but for what you are doing here is of negligible value.

It's true the render time will go up, but the end result should look/be much more correct in relation to the "real world".

For something like a fabric you might even want to do some small displacement to the faces of the geometry to mimic the additional variation in weave thickness for the highest amount of realism... but I would only bother with these things if the material is going to be the "centerpiece" of the render -- otherwise it's just not worth the extra work/render time.

Best,
Jason.

BTW it does not matter for this render, since you are not likely to see the surface of the fabric much here -- but if you were to do a render with the light turned off (like a daylight scene) then I would use this version instead.
This adds another layer to complete the anisotropy effect for the areas cut out by the opacity mask of the ghost layer.

Best,
Jason.

mille gracias. I´ll explore that some more in the evening...

Rogurt

Rogurt ist happy ... i like

This one rendered 2h with only little noise left. I think this is as low as I can go...


@djflod I am not quite sure if I am happy. The rendertime ist acceptable now. But I find once more that it´s really easy to mess up the scene so it wouldn´t clear until the last judgement...

Cutting the render to 25% of the original is still a good accomplishment, considering the previous wasn't really clear at 8 hours.

Best,
Jason.

you bet! And I am quite sure 8h wasn´t nearly the end of it all!
I just wish I´ll get to the point soon where I don´t have to bother you each time I want a render to get finished within a week...

Until then thanks again for your help
Rogurt

P.S. I havent tried out your other material tweaks yet. Quite curious about what I´ll get.

A few of your original MXMs (which looked like downloads from the Maxwell Resources page) were definite noise-makers... so even if you only change out those few, you should see a substantial increase in quality/speed ratio.

Best,
Jason.