Closed GoogleCodeExporter closed 8 years ago
I imagine this can only be a problem with scripting?
Otherwise it is quite useful to be able to turn a gray color into non-gray
without having to switch to RGB mode.
Original comment by ilija.melentijevic
on 4 Dec 2010 at 11:15
No, it has nothing to do with scripts (I have custom algorithms for HSL etc.
though so my scripting wouldn't be affected even if it did). This is an error
and not how a HSL-model should work. You simply can't increase the saturation
of an image that contains grays without getting erratic results (not that you
should...I have a superior saturation-script for that ;) ). And I have no idea
what you mean with usefulness of this bug...but if you're looking for a certain
effect/feature - I could probably write a script to mimic it, hehe! :D
Original comment by annas...@hotmail.com
on 4 Dec 2010 at 11:36
Well, I don't know... These sliders have simple additive effect, so if you
perform S+=25 on color HSL=0/0/172 (light grey) you obtain HSL=0/25/172
(lightly red-teinted grey), it's not wrong.
If I handled special case that greys have an unchangeable S of zero, it would
be permanent... you wouldn't be able to "colorize" a grey color while using HSL
sliders.
Original comment by yrizoud
on 8 Dec 2010 at 12:38
Grey is a special case; It has no saturation and no hue, they cannot be
increased/changed. You simply cannot add color to greys, they don't have any
color. Grey 172 is not equal to HSL 0/0/172 - it only has lightness 172,
nothing else.
If you now have an image with colors and some greys and you increase the
saturation - the greys go red and the image is corrupted! It's not how the
HSL-model works.
Original comment by annas...@hotmail.com
on 8 Dec 2010 at 1:19
For HSL 'mode' and when several colors are selected, I could change the
behavior of S and L sliders so they are multiplicative, but then:
- greys would be totally impossible to colorize. That's what you want, but for
some other usages it's a hindrance. With Photoshop color selector, you CAN
tweak S from zero (greyscale) to max (pure color gradient), why not in Grafx2 ?
- black would stay black, no matter what you do with L (500% of zero is zero).
Original comment by yrizoud
on 8 Dec 2010 at 11:41
Hm, in Photoshop grayscales in images are not affected when adjusting
H&S...However I do see that the single color-selector works like Grafx2, I'd
still call this a flawed behaviour.
Let's make this clear: Hue of 0 degrees is RED. Greys have an UNDEFINED hue. A
color with S=0 and Hue=0 is a gray version of red...NOT TRUE GRAY.
The problem is that a program's interface cannot easily(?) display 'undefined',
so the program would likely have to keep track of that it's dealing with a grey
but still displaying the (incorrect) 0 for Hue, as in RED.
Now, take a look at Brilliance and how they solved this:
Adjusting S for a grayscale does nothing...but as soon as you have touched the
H-slider (even if it still displays 0) you assign a numeric value to H, thus
it's no longer a grayscale - adjusting S now will have an effect.
Now if this could be implemented it would give us the best of both worlds,
right?!
Original comment by annas...@hotmail.com
on 9 Dec 2010 at 3:23
I've made a test with the following rule:
When several colors are selected, and at least one of them is non-grey (S>0):
The S slider doesn't affect the grey colors.
This solves the issue when you want to tweak all palette (doesn't redden greys:
what you asked for), and it also allows you to select an existing greyscale
range and "colorize" it.
By the way, what I said about additive system for S and L wasn't true: I had
implemented a weighted average : +255 sets maximum, +(255/2) sets the average
of 255 and original value, etc. I found in the code that I commented "This
makes a better use of the value range 0-255".
Original comment by yrizoud
on 9 Dec 2010 at 1:46
Attached here, please test
Original comment by yrizoud
on 10 Dec 2010 at 12:59
Attachments:
Seems to work fine. All things considered this should be a very acceptable
solution. :)
Original comment by annas...@hotmail.com
on 10 Dec 2010 at 6:56
Adopted in r1658
Original comment by yrizoud
on 12 Dec 2010 at 6:09
Original issue reported on code.google.com by
annas...@hotmail.com
on 30 Oct 2010 at 4:18