Bug: Difference between small and high Solution

Started by Irmisato, December 22, 2006, 02:43:36 AM

Previous topic - Next topic

Irmisato

Is it a Bug or a Feature?  ;D Or did I make a Mistake?

The Problem: I rendered this Picture, linked below with "Quick Render". When I was satisfied with everything, I made the Rendering-Attitudes for the highest Quality, that is possible and rendered again in the "Quick Render" with a solution of 320x240. There was still a bit visible structure on the surface, that was, what I wanted.

So I rendered the Picture in the "Full Render" with exactly the same Attitudes as in the "Quick Render", but a solution of 800x600.
What happened? The middle hill was black. Thats not, what I wanted.

So I thaught, ok, the "Quick Render" works different from the "Full Render" and I rendered the File again in the "Quick Render", but with a solution of 800x600. But the middle hill was black, too.
Now I thaught, maybe I have changed something, and tested again in a solution of 320x240. But everything was ok, the middle hill had a visible structure.

Here's the small picture:
http://www.ariallah.de/picts/Bergseekl2.jpg

And here you can see the big one:
http://www.ariallah.de/picts/Bergsee2.jpg

Why do I think it is a Bug and what disturbes me?

When I work several days on a Picture, I want to know, how it will look like - from the small preview. When I start a big Rendering, that will last maybe 6 hours, and after rendering I can see, it ist not, what I wanted, then it is a bit frustrating to me. I think, it is important, that a small view should show the same as the big view will, as far as the attitudes are the same.
A test with the "Full Render" showed the same mistake. (Same attitudes - different results in the two different solutions)

(Sorry for my unprofessional English! I hope, you understand, what I mean.)

Matt

This is probably a difference in the Global Illumination / Enviro Light. In your low resolultion renders there would have been less GI samples and it would have been a less accurate GI render. The final render is more accurate. You can see how many GI samples are computed when it is rendering the "pre pass" - the GI samples are computed at each of the dots during the pre pass.

I don't think there is a solution right now, but I want to add a "GI cache file" feature which you should be able to use to match the lighting between different renders.

That is a very nice render, by the way :-)

Matt
Just because milk is white doesn't mean that clouds are made of milk.

Matt

One more thing.. you may be able to get similar lighting in your final render if you reduce the GI relative detail so that there is a similar number of GI samples in both renders. It won't look exactly the same but you may get lighting that is more "averaged out" if you reduce the GI relative detail.
Just because milk is white doesn't mean that clouds are made of milk.

Irmisato

So it is not really a bug of the program, more a special problem with this Picteure because of very many rays?

I thank you for answering so quickly and try out to reduce the GI relative detail.
Thank you very much  :)

michigan_1112

I agree with  Matt, that is an excellent image.

Thanks Matt, I had a similar situation the other day. This clears it up, no pun intended ;)

Oshyan

Fantastic image! Yes, the Global Illumination can affect lighting of the terrain a lot, so if the detail/number of samples for the GI calculations is different between renderers you can get different results. In this case as Matt said you can probably just reduce the GI accuracy for the higher resolution version.

- Oshyan