On 2022-02-27 13:09 (-4), Cousin Ricky wrote:
> On 2022-02-22 09:36 (-4), Cousin Ricky wrote:
>>
>> One question that must be answered is should we assume that the colors
>> are gamma pre-encoded?  Image stock_metal_gamma-srgb.jpg assumes that
>> they are, and uses the srgbft keyword to decode them.  Image
>> stock_metal_gamma-linear.jpg assumes they are not, and just uses the
>> colors as-is.  Comparing them, it seems to me that the colors were not
>> pre-encoded, unlike those in colors.inc.
> 
> I discovered that the demo scenes from POV-Ray 3.0 explicitly set
> assumed_gamma to 2.2, suggesting that the pigment colors were gamma
> pre-encoded.  However, as the first OP image shows, this results in
> metals that are too dark when used with a realistic finish.  It appears
> that the old finishes exaggerated the luminances of the colors while
> reducing their saturations, so it seems best to leave the colors as-is,
> as if they were not pre-encoded.

In POV-Ray 3.7, the assumed_gamma in the demo scenes was changed to 1,
and the results look horrible, further confirming that the colors are
gamma encoded.  As assumed_gamma 1 is the recommended setting going
forward, one would think these colors should be decoded for the new
include file; but decoding the colors is more problematic than not.

> Not gamma-decoding the colors does result in hue drift, but I think the
> drift is in a better direction.

On second thought...

I massaged the colors in an attempt to eliminate the hue drift while
keeping them suitable for linear gamma and metallic reflection.

The top row of each image uses the original F_MetalC, but with the
colors decoded to retain the gamma 2.2 look when using assumed_gamma 1.

The second row uses the 2.2 encoded colors to maintain somewhat high
luminance with an updated metallic finish.  A drift in hue away from the
top row is evident.

The third row maintains the original hues, while using the 2.2 encoded
saturations and values.

The bottom row maintains the original hues, while using the decoded
saturations and 2.2 encoded values.

I believe the third row best matches the original colors as presumably
intended by the POV-Team, although, for the brasses and golds, the
second row seems closer to colors I see in real life.  What do you think?

Post a reply to this message

Attachments:
Download 'stock_metal-massaged-bs.jpg' (197 KB) Download 'stock_metal-massaged-cr.jpg' (172 KB) Download 'stock_metal-massaged-cu.jpg' (205 KB) Download 'stock_metal-massaged-au.jpg' (207 KB) Download 'stock_metal-massaged-ag.jpg' (176 KB)

Preview of image 'stock_metal-massaged-bs.jpg'

Preview of image 'stock_metal-massaged-cr.jpg'

Preview of image 'stock_metal-massaged-cu.jpg'

Preview of image 'stock_metal-massaged-au.jpg'

Preview of image 'stock_metal-massaged-ag.jpg'