OTOY Forums

In a discussion over at DAZ, a person who maintains she and her geek friends are right, say that if you stop an unbiased render, the result is a biased render, because you took a shortcut in stopping the render. I say it's not. I've explained what unbiased is and what biased is and yet she maintains I'm wrong.

I wrote:

"Stopping it does not make it biased and is not taking a shortcut. Shortcuts, as often referred to in these discussions, are how a biased render engine gets to its results, meaning it cheats on figuring the computations for light, shadow, etc. A biased render engine will often ad blur, will dither or do some other shortcut to get to its result. An unbiased render just improves on what it has already computed with no bias. If you stop the render, it's still an unbiased render. Like I said before, the expected value of the unbiased estimator will always be the correct value, for any number of samples, which means you can end the render early and it will still be unbiased. When you stop an unbiased render you are stopping the sampling, but everything that made it unbiased in the first place has not changed."

What say you folks?

I would say, you are correct. She doesn't appear to understand the difference. Have you tried pointing her to some on-line references? Even if it's just Wikipedia?

http://en.wikipedia.org/wiki/Unbiased_rendering

Kevin Sanderson wrote: say that if you stop an unbiased render, the result is a biased render, because you took a shortcut in stopping the render.

Well, according to this statement, the only unbiased render is that which runs forever.

Point her at this interesting site:

http://www.cgafaq.info/wiki/Bias_in_rendering

An image path traced with only one sample per pixel is still an unbiased estimate of the correct image.

grimm wrote:I would say, you are correct. She doesn't appear to understand the difference. Have you tried pointing her to some on-line references? Even if it's just Wikipedia?

http://en.wikipedia.org/wiki/Unbiased_rendering

Thanks, I've tried and others have, too, even with that link once, but she has these so-called experts she works with who tell her otherwise.

grimm wrote:Point here at this interesting site:

http://www.cgafaq.info/wiki/Bias_in_rendering

An image path traced with only one sample per pixel is still an unbiased estimate of the correct image.

Thanks!

matej wrote:

Kevin Sanderson wrote: say that if you stop an unbiased render, the result is a biased render, because you took a shortcut in stopping the render.

Well, according to this statement, the only unbiased render is that which runs forever.

Isn't that silly!

Well it is pretty easy to find definitions for biased and unbiased rendering. Your explanation is pretty accurate and it sounds like this other person is just being stubborn. Sometimes people just wont listen to reason. Not sure what "experts" she is talking to, but if they don't know the mathematical and inherent differences between biased and unbiased then they aren't much of experts. Best of just to state your case and ignore the rest and be done with it.

Perhaps she is just being Biased

tell her to fix you a sandwich

Kevin, as long as there is people on this earth there will be sites like theflatearthsociety and those who can magically fabricate hard evidence that the earth is aproximatly 10.000 year old and deny everything else.

Take her with a smile and put her into the box called "internet" where everything random goes.