Jump to content


Color bleeding


5 replies to this topic

#1 Alienizer

    Member

  • Members
  • PipPipPipPip
  • 435 posts

Posted 27 April 2012 - 06:35 PM

I'm puzzled as to why my white shed on the green grass outside my house is not greenich, but in a render, it is!!!?

#2 Reedbeta

    DevMaster Staff

  • Administrators
  • 5310 posts
  • LocationSanta Clara, CA

Posted 27 April 2012 - 07:15 PM

Perhaps the grass in your render is too saturated? I've noticed that people tend to make their colors way more intense in renders than they are in real life. It makes the image look "better" to some extent, but it can produce side effects like this. :)

Or, saturation aside, it could be that your materials are just too bright in general and reflect too much light. You might experiment with darker materials and brighter light sources; this will reduce the relative intensity of bounce light. This is a very common mistake. It's a good idea to get a photographer's color card that has a set of gray levels on it, like this one. If you take this outside and hold it up against the wall of your shed to compare the colors, you might be surprised at how far from white the shed actually is!
reedbeta.com - developer blog, OpenGL demos, and other projects

#3 geon

    Senior Member

  • Members
  • PipPipPipPip
  • 939 posts

Posted 27 April 2012 - 08:42 PM

Just like Reedbeta said, it's easy to make the materials to bright.

I did some research in this when I wrote my bidirectional pathtracer (a fancier kind of raytracer), and it turns out white paper has a reflectivity of somewhere around 50%.

You'd expect that to be a medium grey, but it's not. It's all relative to how bright your lightsources are.

What I ended up doing was to specify all my colors like red = (1, 0.01, 0.01) * standardReflectiity, where "standardReflectivity" is 0.5. Notice that even my "red" has a tiny bit of green and blue, since no real material is 100% black or pure in the color.

This way I can divide the light power by the standard reflectivity to get the same exposure, in case I need to change it.

#4 Reedbeta

    DevMaster Staff

  • Administrators
  • 5310 posts
  • LocationSanta Clara, CA

Posted 27 April 2012 - 08:53 PM

geon, do you happen to have the source for that 50% figure for white paper? I'm skeptical. In my experience white paper seems to have an albedo closer to 85-90%. Materials are often less bright than you'd think, but white paper is still pretty bright. :)
reedbeta.com - developer blog, OpenGL demos, and other projects

#5 geon

    Senior Member

  • Members
  • PipPipPipPip
  • 939 posts

Posted 27 April 2012 - 10:22 PM

Sorry, no. The little data I could find was not complete or consistent at all.

Edit:

This says 60%, but it's just used as an example:
http://www.astro.was...orials/Spectra/

This says 87-93%, but doesn't explain how it was meassured:
http://amper.ped.mun...ance/albedo.htm

Some values:
http://www.ukscience...dia/Albedos.doc
http://en.wikipedia....Albedo-e_hg.svg



#6 Alienizer

    Member

  • Members
  • PipPipPipPip
  • 435 posts

Posted 28 April 2012 - 02:54 PM

Right on Reedbeta. You're right, my shed isn't white like I thought. So by mult all my materials by 0.5 and my lights by 2, it looks much more natural and my shed is not green now, just a little bit, but barely. Thank you all :)





1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users