Am 04.01.2015 um 09:39 schrieb jmichae3:

>                  color rgb <251.0/255*BRIGHTNESS, 230.0/255*BRIGHTNESS,
> 1.0/255*BRIGHTNESS>//nice bright gold color I picked using ps's color picker
>                  //color rgb <233.0/255, 157.0/255,211.0/255>

Use the "srgb" keyword whenever you pick some colour from a different 
application (unless you know exactly what you're doing).

> I am still experimenting to get what I want.
> is the dithering because of:
> - the viewer which is doing the dithering on my 32-bit display?

Maybe.

> - the monitor, which is a low-end monitor and can't display 10-bit color?

Maybe.

> - roughness .1? (beginning to suspect lack of high resolution in this area or
> specificity of resolution for roughness should that be necessary - really it
> should be derived from the image size).

Nope. Roughness doesn't introduce noise.

I don't see your image, so it's difficult to diagnose. It might also be 
due to the presence of noise in whatever the golden object reflects.


> is the posterization because of:
> - the colored lights I am using (Like the gels I learned about in theater stage
> lighting in high school), and this is a shadow (I think so)?

Maybe.

> maybe I should
> soften the shadows! I was looking at area lights, but I don't understand the
> direction of the vectors, the manual shows 2 of them them going in 2 directions
> at once. there are vectors that do that sort of thing, but how would you write
> that with the syntax given? this is not explained very well.

Normally you'll be fine with just choosing two vectors of equal length, 
each along one coordinate axis (e.g. "32*x, 32*y") and specifying the 
"circular orient" keywords.

> - the Bits_Per_Color=16 and Output_File_Type=N (but not sure if PNG can handle
> more than 8bpc)? but the posterizing should not look like 2bpc like it does now
> on the shadows. still suspect the previous item.

Nope. PNG is fine with Bits_Per_Colour=16, and should indeed not cause 
any colour banding.

> also, the manual section on bits per color does not explain the necessary detail
> that there are some bit resolutions it doesn't support (OK, what DOES it
> support? what are the boundaries?),

The limits depend on the file format chosen.

> and that it means bits per primary color, so
> that if you choose 16, the total number of bits is RGB 3colors*16bitsPerColor=48
> bits total.

This is implicit in the parameter's name: It's not Bits_Per_Pixel, but 
Bits_Per_Colour (meaning red, green and blue respectively).

Also, the colour banding might be due to the lack of something for the 
golden object to reflect. If you're too lazy to build a full-fledged 
scene around it, try a light probe.


> what I really wanted was a special effect I was able to get from POV-RAY once a
> long time ago where a certain configuration caused (oh what's that effect called
> where you get the rainbow-diamonds lens effect that looks so cool in some
> photos? I think it's diffraction grating (in cameras, a diffraction filter) that
> does that. too bad I can't seem to find a keyword for that.

Look for dispersion.

> I don't like the way the colors look like brightness was cranked up too high. I
> think I fixed this before using roughness 0.5. I could turn down the lights, but
> the surrounding sphere (not listed here) would seem darker, maybe I need to
> bring up the light levels in that area or something.

Might be due to using "rgb" instead of "srgb" to specify the colour.


> I read in the manual I think that increasing roughness is supposed to
> be useful for shiny metals to help make them look shinier (I could not see this
> logic, some explanation needed).

Other way round: Reduce roughness to make stuff look more shiny. 
Actually your current value of 0.5 is close to pathologic, and may 
contribute to the banding effect.

Post a reply to this message