These images relate to a gamma discussion over at
http://news.povray.org/povray.general/thread/%3Cweb.439a1b692d1f46002a1c213f0%40news.povray.org%3E/

Please post any comments there.

The images were rendered with different assumed gamma settings...1.0 vs.
2.0.  My own computer system is set up with 2.0 gamma.

I replaced the cube in my original scene with a sphere.

The gray-band backdrop, made from equally-stepped gray values, is
"self-illuminating"....finish {ambient 1 diffuse 0}.  The sphere is lit
from a single white point-light source placed far away, and has finish
{ambient 0 diffuse 1}. The sphere and light source are inside a lightgroup,
so the background is unaffected by the light.

Which image looks more realistic?

Ken

Post a reply to this message

Attachments:
Download 'assumed_gamma_tests.jpg' (82 KB)

Preview of image 'assumed_gamma_tests.jpg'

"Kenneth" <kdw### [at] earthlinknet> schreef in bericht
news:web.43b38ea7db5693254a6a0d0@news.povray.org...
> These images relate to a gamma discussion over at
>
http://news.povray.org/povray.general/thread/%3Cweb.439a1b692d1f46002a1c213f
0%40news.povray.org%3E/
>
>
> Which image looks more realistic?
>

Without any understanding of the physical properties behind the whole
discussion (and I want to keep it so!), I would definitely say 1.0

For me, 2.0 is too dark; the terminator line is at the wrong place (Too far
to the right) for a very distant light source.

Thomas

Post a reply to this message

> >
> > Which image looks more realistic?
> >
>
> Without any understanding of the physical properties behind the whole
> discussion (and I want to keep it so!), I would definitely say 1.0
>
> For me, 2.0 is too dark; the terminator line is at the wrong place (Too far
> to the right) for a very distant light source.
>
> Thomas

Thanks for posting.

How about the gray bands in the background?

Ken

Post a reply to this message

Well, of course, the trick is what *our* monitors are working with. When I
play with my gamma to get it to 2.0 (which is a wierd value to pick, btw),
the bands on the bottom fall much more smoothly from white to black, where
as the bands for gamma 1.0 are heavy towards the light end. 1.0 is not, in
fact, a gamma I can get, so I have no way of comparing - but as I
understand it, if I could reach such a gamma, the top image would look like
2.0 did at 2.0.

It might make more sense to use gammas such as 1.8 and 2.2, so you can get
opinions from people with the proper settings.

-s
5TF!

Post a reply to this message

as i followed the discussion here and in the previous thread, there are 
some points which seem to get obvious.

There is a linear color space, which is pure logical and the way a 
computer can deal with, there is the viewing device (monitor, printer, 
etc.) which interprets these numbers and reproduce something more or 
less eye-pleasing and there is the way the human eye interprets the 
intensity of these colors (like in acoustics in no way linear; neither 
in amplitude nor in wavelength).

If your are going to render an image, which will only be shown on one 
viewing device (video card + driver + monitor) you may indeed use 
gamma-correction in forehand. In this way it made no sense to send a 
sample image to let people deside (everyone on a different setup) what 
would be the better, beside you invite us all to visit you at home to 
see it on your monitor.

If you plan to use your renders to view it, to join it, or just simply 
to print it, you have to do color correction for each device you 
reproduce your image on. As these color corrections include always loss 
of information through number crunching, the best way IMO is to render 
it with NO correction at all but with high color resolution (16 bpc, or 
better floating point like hdr) and post-process it for each device 
according to there ability to reproduce colors and intensities.

.... dave

Post a reply to this message

"stm31415" <sam### [at] cscom> wrote:

> ...1.0 is not, in
> fact, a gamma I can get, so I have no way of comparing - but as I
> understand it, if I could reach such a gamma, the top image would look like
> 2.0 did at 2.0.

Hmm.  I didn't mean for anyone to have to switch his or her monitor/OS gamma
to equal 1.0 (or 2.0) just to look at my image. (1 makes things look pretty
horrible, IMHO.) The
test image I posted is just a ,jpg file, combined (in Photoshop) from two
..bmp POV renders of my scene, with no embedded gamma info. It should show
up in any typical image viewer *more or less* as I intended. My original
question...and I apologize if it wasn't clear...should have been "Which
rendered image looks more realistic on your own system?" Any differences in
the OVERALL image quality from system to system, gamma-wise, would be
"smaller" than the
differences between my two renders (assuming folks are looking at my image
on a system with a gamma somewhere between 1.8 and 2.2.)

>
> Well, of course, the trick is what *our* monitors are working with. When I
> play with my gamma to get it to 2.0 (which is a wierd value to pick, btw)...
> It might make more sense to use gammas such as 1.8 and 2.2, so you can get
> opinions from people with the proper settings.

Weird?  Nope. As I mentioned in my gamma discussion, my own PC's monitor/OS
system gamma  is set at 2.0, as a "compromise" between the Mac's 1.8 and
the PC's "normal" 2.2 (though of course, *most* of the world's computer
systems ARE set at 2.2.) This makes perfect sense, if I want to create
image files (though NOT .png files) that will look *just about* the same on
others' systems. Before deciding on 2.0, I did MANY tests, comparing lots
of different images on a system gamma of 1.8 vs. 2.2. (I have both a Mac
and a PC, each with its own monitor, so that was relatively easy.) My
choice of 2.0 seems to work quite well.

Ken

Post a reply to this message

David El Tom <dav### [at] t-onlinede> wrote:

>
> If your are going to render an image, which will only be shown on one
> viewing device (video card + driver + monitor) you may indeed use
> gamma-correction in forehand. In this way it made no sense to send a
> sample image to let people deside (everyone on a different setup) what
> would be the better, beside you invite us all to visit you at home to
> see it on your monitor.
>

--Not necessary. ;-) see reply to stm3145 above--

But thinking about it, it has occured to me that a fair number of POV users
might be working on a system with a gamma of 1.7 (Unix users?) If so, then
my image might in fact appear TOO far out of its intended viewing range,
gamma-wise. Sorry about that; it does muddle things. But that would be a
problem with ANY "typical" image created on a 2.2-gamma system (or even
one like my own, set at 2.0) How do 1.7 systems reproduce any such image
"correctly"?  I'm guessing it has to be in the .png format, with a proper
embedded gamma. Personally, I don't like creating .png files, for two
reasons: Practically no app on either of my own computers recognizes the
embedded gamma info (not even my two *older* versions of Photoshop), and 2)
from what I've been reading in other posts here, some of the apps that DO
handle it seem to do so oddly.  POV itself is the exception, of course. But
I'd rather not take the chance, when posting images to the web.

Post a reply to this message

Here is how I understand it - and I may be wrong, it has, astoundingly
enough, been known to happen ;)
The gamma info that is included in some images (not this one, as you said,
it being jpeg) is there specifically to allow for apps to correct the image
to make it appear as it did on the monitor it was originally made on, i.e.
yours. IF there is no such information, then the image is seen in an
altered state, because ht egamma of the viewing monitor is different then
yours. If I change my monitor's gamma to match yours, that is, I make the
correction instead of the software, it works the same way. If you do not
plan on my changing my viewing system, then you might as well not do gamma
correction, make the image look good on your monitor, and accept that it
will be different on mine.
So unless you use a filetype that includes gamma, and a program that adjusts
for it, or make us all switch to gamma of 2.0 (which may be logical but is
still wierd) then there is no point in using assumed gamma.
That is how I understand it. PLease correct me if I am wrong - I want to
learn.

-s
5TF!

Post a reply to this message

"stm31415" <sam### [at] cscom> wrote:
> Here is how I understand it - and I may be wrong, it has, astoundingly
> enough, been known to happen ;)
> The gamma info that is included in some images (not this one, as you said,
> it being jpeg) is there specifically to allow for apps to correct the image
> to make it appear as it did on the monitor it was originally made on, i.e.
> yours.

Yes, that's how I understand it as well (though as mentioned, I never render
..png's to send elsewhere, so I haven't seen that "with my own eyes," so to
speak.)  BTW, as you may already know, a .png image (with an embedded gamma
of 1.0) works wonderfully when used as an image_map within POV.  It will
render accurately no matter what the assumed_gamma value of the scene is.
Alas, not so with standard file types like .jpeg or .bmp; then
assumed_gamma must be set to [whatever overall system gamma the image was
created in.] 2.0 in my case.  But a most *incorrect* thing to do, though
(as has been drilled into me by just about everyone.) But I do it anyway...
;-)

> IF there is no such information, then the image is seen in an
> altered state, because the gamma of the viewing monitor is different than
> yours. If I change my monitor's gamma to match yours, that is, I make the
> correction instead of the software, it works the same way.

Right again. But from the many tests I've done, I can honsetly say that the
"altered state" is *minimized* by my use of a 2.0-gamma setup. Of course,
others may disagree.  Mine was simply  practical consideration. But as you
say, to see my image exactly as I had intended, another system would have
to be changed to 2.0 as well.

>If you do not
> plan on my changing my viewing system, then you might as well not do gamma
> correction, make the image look good on your monitor, and accept that it
> will be different on mine.

True, very true.


> So unless you use a filetype that includes gamma, and a program that adjusts
> for it, or make us all switch to gamma of 2.0 (which may be logical but is
> still wierd) then there is no point in using assumed gamma.

Do you mean assumed_gamma in POV?  If so, then...uh...I don't know how to
accurately answer that.  Mainly because I don't yet FULLY understand what
happens when leaving it out. Call me slow, I dunno. I would humbly suggest
taking a look at my original gamma discussion, and the follow-up comments
that have been posted...but swallow a strong headache pill beforehand...
:-)

Ken

Post a reply to this message

"Thomas de Groot" <t.d### [at] internlnet> wrote:

>
> Without any understanding of the physical properties behind the whole
> discussion (and I want to keep it so!), I would definitely say 1.0
>
> For me, 2.0 is too dark; the terminator line is at the wrong place (Too far
> to the right) for a very distant light source.

Thanks for the input.
I guess I haven't said so yet here--my vote is for 2.0. Especially as to the
grey bands.

BTW, one of the fellows over at the main gamma discussion posted a url to a
school Professor's site describing why a half moon in a dark sky actually
looks like the 1.0 image. As the Professor describes it,  that's
misleading; lots of odd things contribute to that.

In and of itself, it's an interesting read...
http://www.asterism.org/tutorials/tut26-1.htm

But also posted there is a CGI image of an "ideal lambertian" sphere (I
hope I've said that right), lit from the side, very similar to my sphere.
(He compares its apperance with that of the moon.)  AFAIK, POV uses the
same, rather simple "lambertian" diffuse lighting formula for
creating light/object interaction...and his image looks like my 2.0 image.

Ken Walker

Post a reply to this message