Re: Tone mapping and color space conversions
- From: Hans-Bernhard Broeker <broeker@xxxxxxxxxxxxxxxxxxxxx>
- Date: 31 Oct 2005 18:26:32 GMT
giganut <giganut@xxxxxxxxxx> wrote:
> Because tone mapping I'm implementing is defined as a technique of scaling
> world luminances to display luminances, s.t. a just-noticible-difference in
> world luminance maps to a JND in display luminance. There is no altering of
> chromaticity. The tone mapping operator described in Reinhard, Stark et. al
> (http://www.cs.ucf.edu/~reinhard/cdrom/tonemap.pdf) deals just with the
> luminance in an image.
That's a good thing to do, but "tone mapping" is almost certainly the
wrong name for it, because it's completely misleading. Different
"tones" of a colour are usually understood to mean slight variations
in hue, not in brightness.
> Lworld //world luminance of the pixel from the hdr image
> Ldisplay // display luminance of the pixel computed using the tone map
> operator in the paper
> double sf = Ldisplay/Lworld;
> display_pixel_yiq = sf * world_pixel_yiq; //in short i'm scaling all 3
> cmpnts of the pixel by the scalefactor
Scaling Y'IQ components as if they were RGB ones is almost guaranteed
to be wrong. But that's the least of your worries, since Y'IQ is
probably the wrong colourspace to use in the first place.
> Right. But you're not scaling along just *one* dimension - you're scaling
> across 3 dimensions.
.... which, in the case of (linear!) RGB, happens to be exactly the
thing that needs doing.
--
Hans-Bernhard Broeker (broeker@xxxxxxxxxxxxxxxxxxxxx)
Even if all the snow were burnt, ashes would remain.
.
- References:
- Tone mapping and color space conversions
- From: giganut
- Re: Tone mapping and color space conversions
- From: Hans-Bernhard Broeker
- Re: Tone mapping and color space conversions
- From: giganut
- Tone mapping and color space conversions
- Prev by Date: Re: Tone mapping and color space conversions
- Previous by thread: Re: Tone mapping and color space conversions
- Index(es):
Relevant Pages
|
