|
As this forum is rarely used anymore, we've locked it. Feel free to browse and read. Questions? Please reach out to us directly. Cheers! |
|
 |
|
|
| Author |
Message |
perisoft
Joined: 29 Aug 2007
Posts: 2920
Location: Ithaca, NY
|
| Posted: Wed Feb 25, 2009 3:01 pm Post subject: gamma - gray ramp appearance |
|
|
So, I've got my low-end gamma pretty good - nice shadow detail, good black levels, etc.
But I'm using 'normal' gamma - just the stuff that comes with the ATI drivers.
My question is this: When I look at a gray ramp - say, 10% steps - it's kind of odd. I can just barely see 7% (which I've heard is about right), but the difference between 10, 20, and 30% is, to my eyes, much much greater than the difference between 80, 90, and 100%. The differences are visible at the top end, but far less than at the low end.
Thing is, this is the case on every display I've got - from crappy to great CRT monitors, crappy to good LCD monitors... it was the case on the plasma I have, the RPTV I have... I did some experiments taking screenshots with my Nikon D80, and according to its CCD, the difference between 80% and 90% was far less than the difference between 20 and 30, or 30 and 40.
Is that normal, and just 'the way things are supposed to be', or should this stuff be linear, and everything out there is screwed up?
_________________
|
|
| Back to top |
|
 |
AFryia
Joined: 09 Mar 2006
Posts: 965
Location: S.E. Michigan VPH-G70Q
|
| Posted: Wed Feb 25, 2009 11:12 pm Post subject: |
|
|
My under standing is "Gamma" is a power function therefore can't be perfectly linear.
Try this read.
http://www.poynton.com/PDFs/GammaFAQ.pdf
There is a whole history on gamma and why it is what it is.
|
|
| Back to top |
|
|
perisoft
Joined: 29 Aug 2007
Posts: 2920
Location: Ithaca, NY
|
| Posted: Wed Feb 25, 2009 11:20 pm Post subject: |
|
|
So, from the doc:
| Quote: | For example, if you want to produce a numerical simulation of a lens
performing a Fourier transform, you should use linear coding. If you
want to compare your model with the transformed image captured from
a real lens by a video camera, you will have to “remove” the nonlinear
gamma correction that was imposed by the camera, to convert the image
data back into its linear-light representation. |
So, if I'm reading this correctly, it means that the visual difference between 80% and 90% literal RGB output, say, really IS lower than the visual difference between 20% and 30% - but if I'm displaying a film source, the original 'came through inverted' itself, so that when it gets put back through that transform, it looks correct: ie, if the director had a shot of a linear grade between black and white light, it would look linear when run through the nonlinear gamma curve on my projector?
Kind of an odd way to do things... am I right, or am I misreading something?
_________________
|
|
| Back to top |
|
|
AFryia
Joined: 09 Mar 2006
Posts: 965
Location: S.E. Michigan VPH-G70Q
|
| Posted: Wed Feb 25, 2009 11:36 pm Post subject: |
|
|
Now you are only as confused as I am! 
Go to page 8 of 12
Basiclly the camera records in a non-linear function (gamma). You need to apply kind of an inverse function to display the image properly.
|
|
| Back to top |
|
|
perisoft
Joined: 29 Aug 2007
Posts: 2920
Location: Ithaca, NY
|
| Posted: Wed Feb 25, 2009 11:49 pm Post subject: |
|
|
| AFryia wrote: | Now you are only as confused as I am!
Go to page 8 of 12
Basiclly the camera records in a non-linear function (gamma). You need to apply kind of an inverse function to display the image properly. |
Soo... instead of fixing the recording devices, we break the display devices? 
OK, so that means that I really should 'see' compression toward the top. Even though it says 80% and 20%, visually it's more like 30% and 95%... (or whatever)
Makes sense. 
Now the only question is - why hasn't someone written a grayscale calibration guide for people who can't afford colorimeters? I mean, I know you can't get it RIGHT, but there's gotta be something better than watching R/G/B ramps and adjusting gamma separately for each of them!
Actually, that wasn't the only question. I lied. The OTHER question is: When people say 'gamma bump' for CRT PJs, do they mean normal gamma adjustment, or an additional 'bump' on the low end?
OK, those are the only questions. For now. Really. I think.
_________________
|
|
| Back to top |
|
|
stefuel
Joined: 07 Mar 2006
Posts: 3353
Location: Green Harbor MA USA
|
| Posted: Thu Feb 26, 2009 12:05 am Post subject: |
|
|
Make sure you are not over driving the tubes to the point of clipping or blumming. You can see this with just one color on. Try the green with a 10 step grayscale. Try turning the sub contrast up and down while looking at the 80-100 ire steps. If you are over driving, the highest ire steps start to blend together. Reduce the green drive until you have distinct even steps then adjust the other colors around it.
_________________
Chip
A Barco is only a AmPro with training wheels
Card carrying member of the AVS chain gang.
|
|
| Back to top |
|
|
perisoft
Joined: 29 Aug 2007
Posts: 2920
Location: Ithaca, NY
|
| Posted: Thu Feb 26, 2009 12:52 am Post subject: |
|
|
| stefuel wrote: | | Make sure you are not over driving the tubes to the point of clipping or blumming. You can see this with just one color on. Try the green with a 10 step grayscale. Try turning the sub contrast up and down while looking at the 80-100 ire steps. If you are over driving, the highest ire steps start to blend together. Reduce the green drive until you have distinct even steps then adjust the other colors around it. |
Yeah, I've been checking that during my woeful color setup procedure I've developed:
-Start with horrid default color setup on the PJ and no gamma on the HTPC
-Turn out all the lights and set brightness just below the threshold of visibility
-Set contrast to a comfortable margin below blooming and where I can clearly differentiate at the top level
Now I set whitepoint... I throw up an all white screen with ntest, and set up my D80 at 6700k (it only has 6300 and 6700, frustratingly) to aim, pretty well zoomed in, at the center of the screen.
I set the D80 to display curves for the captured image. I can see where the R/G/B peaks are - 6700k should be white, and thus should have all peaks at the same point. So I go into picture tuning in the barco, and adjust gains so everything matches.
-Now a final check on overall contrast as before.
-Now I put up a gray ramp with ntest.
-I use color select, and for each green, red, and blue:
I go into the ATI color control menu, and set gamma ramps for each color separately. Using color select, I can see just the color I'm working with. So I set that color's gamma so I get a visually-mostly-correct image - since this is pretty rough anyway, I basically set things so I can barely discern 7IRE.
-And that's it!
The results are relatively crude compared to a real grayscale calibration, but it's a hell of a lot better than trying to do a mush of whitepoint/black level/overall gamma/individual gamma/color temperature, all by eye and at once!
_________________
|
|
| Back to top |
|
|
AFryia
Joined: 09 Mar 2006
Posts: 965
Location: S.E. Michigan VPH-G70Q
|
| Posted: Thu Feb 26, 2009 3:04 am Post subject: |
|
|
HCFR and an EYE-One or similar make all this a lot easier.
I'm actually playing around with a LUT modifying program at the moment. From what I'm reading you want to keep the gamma adjustments at the LUT/hardware level. If the adjustments are 8bit it supposedly cause banding.
|
|
| Back to top |
|
|
perisoft
Joined: 29 Aug 2007
Posts: 2920
Location: Ithaca, NY
|
| Posted: Thu Feb 26, 2009 4:02 am Post subject: |
|
|
Yeah, I'd like HCFR, Eye-1, etc... but I can't justify or afford the couple hundred (?) to do it. My whole HT only cost $600! So, I try to do the best with what I have.
_________________
|
|
| Back to top |
|
|
David_Web
Joined: 02 May 2007
Posts: 418
Location: Sweden
|
|
| Back to top |
|
|
|
|
|
|
|
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
You cannot attach files in this forum
You can download files in this forum
|
Forum powered by phpBB © phpBB Group
|
|
