OTOY Forums

Hello ,


first of all sorry for my bad english

This is my first work with this software and i'm not really satisfied (even if octane is a wonderful software obviously) .
As you can see , theses pictures were rendered with Directlighting Kernel , too long with PMC or PT but before all , too many fireflies . I'v read that roughness value of material should not be lower than 40 .. but the value cannot be out the range of 0 and 1 . What 80 mean ? 0.80 or 0.080 ?
i didn't find a post where people let there most efficient preset for each kernel render .
this work is also a try in using livedb material . thanks for contributor .

of course my scene is bad, i'm a beginner and there is something wrong that i can't explain (except ugly materials but i didn't find a tutorial that really explain how to make material node .)

feel free to kill my work just explain why before i know that the scene is dark sorry .

Yes, a bit dark, but rather nice !

Reducing the reflection and changing the roughness will help. Also where is the light around the ceiling being emitted from is from the wall itself via a mix material.

MaTtY631990 wrote:Reducing the reflection and changing the roughness will help. Also where is the light around the ceiling being emitted from is from the wall itself via a mix material.

Thanks for comment.

i'm not really sure that is a question , but il will try to answer it . the light around the ceiling is simply a diffuse material with blackbody emission property .

concerning the rougness . my cursor can go from 1 to 0, what is the minimum value in order to avoid fireflies ?

Usually the higher the roughness the less fireflies. Try this to see what you get. Also what settings did you use on pmc kernel.

how about change the gamma setting in current imager?

well I'm guessing you were expecting PMC and PT to be as fast as DL, but you have to understand that DL is biased and light is not calculated "properly" hence its speed, but for finals (at least in my opinion) it has a "plastic" look that doesn't cut it. PT and PMC in the other hand calculate the light in the "hard" way, giving a warm feeling that creates a real life illusion that we love so much. Yet it takes more time to "cook", try both, it does depend much on the light type and position to choose one kernel or the other, do some testing and let them cook for about 3K samples and play with the hotpixel removal, usually a value between 0.05-0.15 for 3K is fine to get a clean render. There are no "standard" setup 'cause it has a lot to do with your scene and emmitters, with a single powerfull emmiter, as the sun, the scene gets solved really quick, when you start to add lots of them, the things change. Also try to keep a balance surface for the emitters, huge surface difference between emmiter create too much noise. Also huge bump values create unrelistic bumps and noise, usually I keep them about 0.1-.2 or lower.
It's a nice first job btw.

thanks for the explanation . it will help me a lot . i'm looking to expand my cuda processor .

By the way i'm not sure to understand this sentence : " Also try to keep a balance surface for the emitters, huge surface difference between emmiter create too much noise" . Will be kind enough to explain me again what you mean ?

Did you mean , that if i have an emitter about 100 cm² , i should'nt use a ceilling emiter about 4 m² ? it will give a difference between emitter that will provide extra noise ? am i right ?

Thanks a lot .

yeap that's right, at least for now, I'm sure this is something they'll address in future releases. It's not quite easy to predict how much difference produces the noise, but it's something you can clearly see, if there is an emmitter than instead of shedding an even light it creates a noisy one that's more likely the cause. Usually I've had problems when using regular downlights which are more or less correct in size and adding a "fake" one, say a hidden ceiling one, that to save me some modeling was a huge surface instead of smaller emmiters (like it would be in real life).

thanks for this !!!!
i understand better .
i think a trick is possible using texture emitter and well fitted ies ligting .
we can use large surface emitter (in order to save time in modelling) and using the scale propriety of the ies to define an effective emission surface with the same surface than other ligth source in the scene ...

what do you think of this ?