Alternative biased method
Posted: Mon Jul 11, 2005 5:33 pm
It would be great if we have both choices. Biased and unbiased, with the same innovative spectral method of light calculation.
yeeeess please
like vray
Maxwell Render | Official public forum
The official Maxwell Render public forum. A place to share your latest projects and be inspired by others, or get help and feedback from the Maxwell community
https://forum.maxwellrender.com/
Alternative biased method
Page 1 of 1
Posted: Wed Jul 13, 2005 10:15 pm
Posted: Thu Jul 14, 2005 12:06 am
We don't like the word biased 
Posted: Thu Jul 14, 2005 1:45 am
Sirs,
Just interested in knowing what advantagous there are with "biased" rendering techniques over "unbiased". (I admit that this is still terminology that I'm not entirely familar with yet.)
rzick
Just interested in knowing what advantagous there are with "biased" rendering techniques over "unbiased". (I admit that this is still terminology that I'm not entirely familar with yet.)
rzick
Posted: Thu Jul 14, 2005 2:59 am
me eather...can you describe what is biased unibiased
Posted: Thu Jul 14, 2005 10:09 am
Just leave this render engine unbiased... please. 
However, a biased, blotchy and really fast mode would be great to help scene setups. Just use the current engine, to calculate some samples, and then interpolate them over the surfaces. Strictly only for light tweaking purpuses. But, a "really fast preview" has been promised, so, I'll shut up. I'm quite happy with the current Maxwell engine, I really don't need "another good biased renderer".
An unbiased rendering is an exact method to calculate the image, and introduces no artifacts during the rendering, and it always converges to the perfect solution of a light distribution equation. Sometimes called "brute force", and usually needs a lot time to be noiseless.
Biased methods use approximation, and thus, even with really high settings and low thresholds, the solution is never perfect, this is why its called "biased".
Also, I don't think there is much difference in a real good approximation, and a really long unbiased render. A real good approx takes almost as long as the unbiased rendering. There is no way we can currently solve the rendering equation in finite time with infinite precision (except for very, very simple cases), so, from this point of view, an unbiased render still remains biased, if we consider the noise an artifact.
However there are more aspects to consider. For example hardver usage, and felxibility.
Btw, morbid angel: Just read what Vlado wrote bout the GI methods in the VRay beta help under "Classification of GI methods", its a good reference.
However, a biased, blotchy and really fast mode would be great to help scene setups. Just use the current engine, to calculate some samples, and then interpolate them over the surfaces. Strictly only for light tweaking purpuses. But, a "really fast preview" has been promised, so, I'll shut up. I'm quite happy with the current Maxwell engine, I really don't need "another good biased renderer". An unbiased rendering is an exact method to calculate the image, and introduces no artifacts during the rendering, and it always converges to the perfect solution of a light distribution equation. Sometimes called "brute force", and usually needs a lot time to be noiseless.
Biased methods use approximation, and thus, even with really high settings and low thresholds, the solution is never perfect, this is why its called "biased".
Also, I don't think there is much difference in a real good approximation, and a really long unbiased render. A real good approx takes almost as long as the unbiased rendering. There is no way we can currently solve the rendering equation in finite time with infinite precision (except for very, very simple cases), so, from this point of view, an unbiased render still remains biased, if we consider the noise an artifact.
However there are more aspects to consider. For example hardver usage, and felxibility.
Btw, morbid angel: Just read what Vlado wrote bout the GI methods in the VRay beta help under "Classification of GI methods", its a good reference.
Posted: Thu Jul 14, 2005 4:32 pm
gotcha...I have faith in NL team that they can produce a great biased engine and thus making maxwell so powerfull that no other renderer can compare
Posted: Thu Jul 14, 2005 8:54 pm
i think victor said it would always stay unbiased
Posted: Thu Jul 14, 2005 8:56 pm
well one engine can be unbiased, another biased....doesnt hurt to have both