|
|
TC schrieb:
> Thank you for the explanation - it surely beats wikipedia.
Thanks :-)
> I was referring to luminance, not to the reflection in the sphere. The
> brightness of the sky is very different, too. I meant: what is more correct
> (related to the sky) - luminence on the left or right?
That depends on how you prefer to interpret the color values in the
scene file.
I'd say both are wrong in this respect, because the original scene
author tried to achieve what can be seen on the left, but due to not
paying attention to gamma issues he ended up with parameters that result
in the image on the right when the computations are done properly.
> Of course reflection on the left makes more sense than on the right. Is this
> a gamma-related problem or is this an error in povray?
Whoops - are you mixing up left & right here?
It actually is a gamma-related problem, which in this case manifests
itself at the antialiasing step.
You could, with some legitimacy, also call this a design error in the
old POV-Ray versions (in this case 3.6 used with default parameters). It
is /not/ possible to produce the effect seen on the left-hand side with
POV-Ray 3.7, unless deliberately forcing it to enter 3.6-compatibility mode.
> I am having lots of problems with gamma correction on my system. At least
> there are a lot of iffy things.
... which doesn't seem to be too uncommon, unfortunately.
> Contrary to what has been posted here, my video card's driver seems to do
> some gamma correction (Nvidea). At least I can do a some gamma correcting in
> the setup. However, it is a lot of voodoo: I don't know for sure what is
> corrected and when. I really don't see any difference.
Yes, I guess I know that tool. I used it, too, to calibrate my display.
This should actually "patch" some lookup table in the graphics hardware,
which maps 8-bit framebuffer values to values supplied to the ADC (to
generate the analog signal for the VGA interface) or transmitted on the
DVI interface.
> Now, I am using Corel quite a lot. Corel has a lot of colour correcting,
> too. I downloaded the proper colour-management system files for my monitor
> and my printer, respectively, but the printed result did not look much like
> the screen. The PDF-file I sent to my printer (I mean the real one, the guy
> who doesn't bother printing anything below 1000 copies ;-) did look like
> neither my printout, nor the screen.
I guess there's some caveat here: The monitor vendor doesn't know what
graphics card you are using, and how you set your display gamma.
> So I did trash all colour-correcting. All set to "ignore" - from video card
> driver to Corel to printer profile and printer driver. And behold: now I get
> on my printer what I see on my TFT and the guys at the printing press
> produce something that comes close enough to what I want.
I guess you just have to know exactly what you're doing in order to
benefit from color management. I for one don't, so I stick to plain old
gamma correction for now.
> So I am a bit distrustful of colour correction, especially when it is stored
> as a kind of optional parameter in the image-file, like in PNG. Depending on
> the respective viewer / editor, gamma correction in PNG is either ignored or
> applied. If it is applied, one only can hope it is applied correctly. This
> really is a problem. Do you know which browsers support PNG
> gamma-correction?
You can try yourself: Just render a POV-Ray scene to PNG using the 3.7
beta, once with File_Gamma=2.2 and once with File_Gamma=1.0.
Any software displaying both files identically (except maybe for a bit
of color banding in the File_Gamma=1.0 version) is doing proper gamma
handling.
Internet Explorer 8 does it right. The IC image viewer does, too, and
Windows Explorer's image preview likewise.
Photoshop 6.0 does it wrong (but I guess current versions do a better
job), and so does Microsoft Paint.
Post a reply to this message
|
|