|
![](/i/fill.gif) |
Chris Huff wrote:
>
> Or are you talking about an automatic thing that goes through the image
> looking for the darkest and brightest pixels, and adjusts the image so
> those pixels are black and white? That could be done...
Almost. The Photoshop filter is a bit more complex but I think all those
features would be cumbersome in POV. In Photoshop you can adjust the settings
according to a colour histogram while seeing the preview image.
"Auto Levels" in Photoshop adjusts each channel so that the reddest gets a 255
red value, the greenest a 255 green, etc. This is not very reliable and more
often than not produces an unwanted tint.
I suggest the following syntax:
levels { <dark_tint>, <bright_tint> }
If the brightest was <.9,.95,.93> the new value would be:
<.9,.95,.93>/.95*bright_tint
For most "real world" scenes, levels{1,1} would then give good results. I have
to do this manually in Photoshop, adjusting the endpoints to the histogram
endpoints.
The in-between values could be changed as a linear function between the two new
endpoints. In Photoshop there's a value which controls this function, but I'm
not sure how it works. It seems to use a power function taking values from 0 to
10 where "1" is linear, 1-10 sort of moves the pure red/green/blue to pixels
which previously were dark. 1-0 moves the palettes against the light area
making the whole image darker. Aw, I'm hopeless at explaining. I could post
some demo images if you're interested.
sig.
Post a reply to this message
|
![](/i/fill.gif) |