Ad
Ad
Ad
Pages: « 1 [2]   Bottom of Page
Print
Author Topic: (8 bit internal LUTs) LCD whitepoint adjustment  (Read 33788 times)
Serge Cashman
Full Member
***
Offline Offline

Posts: 200


« Reply #20 on: June 05, 2006, 02:38:59 PM »
ReplyReply

Thank you so much for taking your time asking Karl Lang. That was a very authoritative response.

Although I don't understand the 3D-1D part I can run the greyscale gradient test and see what the results are. What's the best way to look at the greyscale gradient? Colormanaged application set to monitor profile as a working space?

Unfortunately in my software (and most other solutions on the market) not using TRC (tone response curve) gamma correction  is impossible since Native Gamma target is not available. Do I understand this point correctly?
Logged
digitaldog
Sr. Member
****
Offline Offline

Posts: 9225



WWW
« Reply #21 on: June 05, 2006, 05:10:39 PM »
ReplyReply

According to Karl, you'd make an R/G/B/C/M/Y and black gradient in say Photoshop. What you want to do is view this with the numbers going directly to the display to view smoothness as you alter the various on board display settings. However to do this, you need to either view them in a non ICC aware application (like most web browsers) OR you need to create a profile that is linear. On the Mac, you could do this using Apple's Calibrator WIHTOUT altering any settings on the display. Don't move any of the sliders in the software and set it all for native. In theory that would create a profile that would send the data to the screen without any adjustments then you could Assign it to the ramp you made while viewing this all in Photoshop. Oh, when you make the ramps, make sure all dither settings in Photoshop are OFF (color settings and gradient).

Seems a bit easier to view this in a non ICC aware application like a browser.
Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
Serge Cashman
Full Member
***
Offline Offline

Posts: 200


« Reply #22 on: June 05, 2006, 06:12:46 PM »
ReplyReply

Quote
On the Mac, you could do this using Apple's Calibrator WIHTOUT altering any settings on the display. Don't move any of the sliders in the software and set it all for native. ...

On a PC that could be done using Adobe Gamma I believe. A "no adjustments" profile... I'll look into that.
Logged
61Dynamic
Sr. Member
****
Offline Offline

Posts: 1442


WWW
« Reply #23 on: June 08, 2006, 09:53:54 AM »
ReplyReply

Quote
Just had to ping Dr Karl on all this. His response:
---

[...]

The simple answer is you can't really know. Manufacturers don't give 
you enough information in most cases.

[...][a href=\"index.php?act=findpost&pid=67460\"][{POST_SNAPBACK}][/a]
That's a big problem with displays these days. The LCD market is still very young and little info is given making it difficult to find decent displays for our use. And, as unnecessarily confusing as people let things get in calibrating LCDs today, just wait until the next technology SED (Surface-conduction Electron-emitter Display) spills into the monitor market in the next year or two. Then after that, OLED (Organic Light-Emitting Diode).

The future promises even better color reproduction and added confusion!
« Last Edit: June 08, 2006, 09:55:02 AM by 61Dynamic » Logged
Serge Cashman
Full Member
***
Offline Offline

Posts: 200


« Reply #24 on: June 09, 2006, 07:26:41 PM »
ReplyReply

Well, I'm still trying to run the test correctly (without any LUT adjustments). I just got a Display2, hopefully it'll help.

Meanwhile I came across a very convincing post by Sergey Oboguev (at Pro Photo - I don't feel like paying to join) arguing against using RGB buttons adjustments on 8 bit LUT monitors. One of the points is that by adjusting in two places (buttons for WP, LUTs for gamma) you may drop more than one adjacsent RGB values...

http://www.prophotocommunity.com/ubbthread...ge/0#Post422964
« Last Edit: June 09, 2006, 07:28:10 PM by Serge Cashman » Logged
Stephen Best
Guest
« Reply #25 on: June 09, 2006, 08:04:46 PM »
ReplyReply

Quote
Meanwhile I came across a very convincing post by Sergey Oboguev (at Pro Photo - I don't feel like paying to join) arguing against using RGB buttons adjustments on 8 bit LUT monitors. One of the points is that by adjusting in two places (buttons for WP, LUTs for gamma) you may drop more than one adjacsent RGB values...

http://www.prophotocommunity.com/ubbthread...ge/0#Post422964
[a href=\"index.php?act=findpost&pid=67814\"][{POST_SNAPBACK}][/a]

As I understand it, the average delta-E from the verification phase tells you how well the generated profile models the actual displayed values ... but nothing about gamut, accuracy, smoothness or otherwise. The eyeball test above I think would be more useful.
Logged
Serge Cashman
Full Member
***
Offline Offline

Posts: 200


« Reply #26 on: June 09, 2006, 09:17:25 PM »
ReplyReply

Oh, that Delta E is about higher bit monitors... In the first couple of paragraphs he was talking about 8 bit ("dumb" in his words) monitors.

Since I don't know how his software works (I think it's Monaco Pro or whatever highend Xrite is called) I can't tell what color difference he's measuring over there. I think it measures a bunch of patches and gives a list of Delta Es for each one (compared to the target)...
« Last Edit: June 09, 2006, 09:27:46 PM by Serge Cashman » Logged
Stephen Best
Guest
« Reply #27 on: June 09, 2006, 10:12:23 PM »
ReplyReply

Quote
Oh, that Delta E is about higher bit monitors... In the first couple of paragraphs he was talking about 8 bit ("dumb" in his words) monitors.
[a href=\"index.php?act=findpost&pid=67821\"][{POST_SNAPBACK}][/a]

I think the moral of all this is that if you're stuck with an 8-bit monitor, just use it as is. Most today seem to be modeled around 6500/2.2 which is a good match for general use, web browsers etc. I've got an Apple Cinema Display 20" with no knobs at all and it does the job ... either profiled natively or tuned to exactly 6500/2.2 in the card. But for much the same price these days you can buy a monitor with smarter electronics and (maybe) better gamut. I've got an NEC 2090 on order as a second monitor (replacing an old Sony CRT) and am keen to see how it compares. Having two smaller monitors means you can upgrade each independently as technology/price-performance improves. If your current monitor isn't meeting your requirements, you may want to consider a similar arrangement. You don't need high-end for palettes etc.
« Last Edit: June 09, 2006, 10:13:30 PM by Stephen Best » Logged
Serge Cashman
Full Member
***
Offline Offline

Posts: 200


« Reply #28 on: June 09, 2006, 10:31:36 PM »
ReplyReply

I use dual monitors that have different Native white points. So I do need to adjust one of them. They are not highend monitors so I do see artifacts like banding and non-uniform color temperature on the greyscale ramp.

But the question is also important because most software solutions bundled with colorimeters guide you to go ahead and adjust LCD buttons if you have them.  From what I learned so far it only makes sense if you have a 10 or more bit internal LUTs monitor that for some reason does not work with your software via DDC.
« Last Edit: June 11, 2006, 03:58:36 PM by Serge Cashman » Logged
Stephen Best
Guest
« Reply #29 on: June 15, 2006, 12:53:50 AM »
ReplyReply

Quote
I use dual monitors that have different Native white points. So I do need to adjust one of them. They are not highend monitors so I do see artifacts like banding and non-uniform color temperature on the greyscale ramp.

But the question is also important because most software solutions bundled with colorimeters guide you to go ahead and adjust LCD buttons if you have them.  From what I learned so far it only makes sense if you have a 10 or more bit internal LUTs monitor that for some reason does not work with your software via DDC.
[a href=\"index.php?act=findpost&pid=67827\"][{POST_SNAPBACK}][/a]

My NEC LCD2090UXi arrived today. I initially profiled it native (40% brightness) and didn't really see much difference to my Apple Cinema Display 20" (also native). I got pretty well identical results with Match 3.6 and basICColor display 4. Gray ramps for both displays were very clean and with no banding. The NEC had more contrast but if anything the gamut was slightly smaller than my older Apple. I was a bit disappointed at this stage and wondered whether I should have bought a cheaper Samsung etc.

Anyway, then I bumped the brightness to 65%, set temperature/gamma to 5000/2.4 and profiled it with display 4 in software LUT mode for D50/L*. The video LUT curves came out nearly flat with just a bump in the shadows. I went through image after image from a set of exhibition prints I did recently for a client and they all looked very close. The gray ramp is clean and with no banding. In fact I've only noticed a single case of banding (in a graduated sky area) where the working space gamma didn't match the display. I converted the image to Lab and it went away. I'm doing more of my work in Lab nowadays anyway. So this is where the extra bits in the monitor come in handy. Maybe if you're printing on resin-coated with lots of OBAs you can get away with 6500 ... but 5000 is a much better match to the rag paper I use.

I was still hopeful that the gamut for the new monitor would be slightly better. I haven't seen a lack on screen but compared it with the Apple using the ColorSync utility. I guess the story is if you want better gamut you'll have to pay for it. I was very impressed with the Eizo monitors I saw recently but the closest model to the 2090 would be the L997 at twice the price (at least here). Note that I haven't compared the 2090 side-by-side with the 2180/2190 (same panel) so maybe the larger monitor is better in this department. But again, more dollars. The 2090 will do me for a while.
Logged
Stephen Best
Guest
« Reply #30 on: June 15, 2006, 04:20:10 AM »
ReplyReply

More on the 2090 gamut.

[attachment=699:attachment]
[attachment=701:attachment]

Above are plots (as profiled) comparing it to sRGB (white mesh). It's better in cyan, but not quite as good as my Apple in the reds/yellows (but close to sRGB). Note that different settings for the display result in a different gamut.

All in all images look great on the 2090 and have real depth. It's a definite step up from my old Sony CRT.
Logged
Serge Cashman
Full Member
***
Offline Offline

Posts: 200


« Reply #31 on: June 15, 2006, 12:44:59 PM »
ReplyReply

I'm not familiar with Basiccolor. What's software LUT mode? Do Match 3 and Basiccolor have DDC control over the monitor?

If "LUT mode" means adjusting videocard LUTs, are the resulting curves (which I assume are correction curves like in Match 3) close to 1 (run straight down the center)?
Logged
Stephen Best
Guest
« Reply #32 on: June 15, 2006, 06:11:51 PM »
ReplyReply

Quote
I'm not familiar with Basiccolor. What's software LUT mode? Do Match 3 and Basiccolor have DDC control over the monitor?

If "LUT mode" means adjusting videocard LUTs, are the resulting curves (which I assume are correction curves like in Match 3) close to 1 (run straight down the center)?
[{POST_SNAPBACK}][/a]

More on basICColor display 4 here:

[a href=\"http://www.basiccolor.de/english/Datenblaetter_E/display_E/display_E.htm]http://www.basiccolor.de/english/Datenblae...E/display_E.htm[/url]

Match 3.6 has closed the gap considerably but I think display 4 still has an edge.

In "Software LUT mode" it creates a LUT for the video card. You can of course just profile it as is (namely with a linear LUT).

The monitor supports hardware calibration but there's currently no software to support it ... AFAIK. An updated version of NEC's SpectraView is on the way. I don't think there's much (if anything) that hardware mode supports that you can't do with the buttons on the monitor, but you do miss out on closed-loop calibration. From what I can currently see, I don't think I'll bother with SpectraView.

Here's the video LUT generated for D50/L*:

[attachment=703:attachment]

The green and blue curves are similar. Most of this is because a gamma of 2.4 isn't exactly L*. If all I wanted was a gamma of 2.2 (or whatever) I'd just dial it (and the colour temperature) in on the monitor and profile it as is. The monitor supports a gamma from 0.5 to 4.0 (in 0.1 increments) and a colour temperature of 3000-9600K internally, though obviously you don't want to move too far away from the backlight temperature. This flexibility is what you're paying for.
Logged
Serge Cashman
Full Member
***
Offline Offline

Posts: 200


« Reply #33 on: June 15, 2006, 07:48:10 PM »
ReplyReply

Stephen, thanks so much for such a detailed reply.

So, the curves do in fact display LUT adjustments of the vcgt tag.

Even if Spectraview does not work for your monitor it should come with some kind of utility allowing you to adjust RGB sliders and whatnot  from your computer. Also, theoretically even if you do adjust RGB buttons on that monitor the resulting quality should be better than LUT adjustments.

So, I suppose with the  L* gamma target and White point adjusted via either monitor software or monitor buttons (as opposed to calibration software adjusting LUTs) the curves are supposed to be almost perfectly linear. I suppose the ultimate goal is to achieve a target significantly different from Native without vc LUTs adjustments...

I don't say that to nitpick -  it appears to follow from what more than 8 bit LUT monitors were designed for.

I still don't know how to thoroughly test a display to check the negative results of either videocard or internal LUTs adjustments unfortunately - I'd really like to learn more about it.
« Last Edit: June 15, 2006, 07:51:32 PM by Serge Cashman » Logged
Stephen Best
Guest
« Reply #34 on: June 15, 2006, 08:14:54 PM »
ReplyReply

Quote
Even if Spectraview does not work for your monitor it should come with some kind of utility allowing you to adjust RGB sliders and whatnot  from your computer. Also, theoretically even if you do adjust RGB buttons on that monitor the resulting quality should be better than LUT adjustments.
[a href=\"index.php?act=findpost&pid=68277\"][{POST_SNAPBACK}][/a]

My 2090 came with a copy of NaviSet but this is for Windows (blehhh) only. There's nothing to stop you adjusting the colour temperature with the buttons at the same time as you're measuring what you're getting with a puck. There's also plenty of other monitor settings I haven't looked at yet. The aim should simply be to get close to what you want for whitepoint/gamma and not see any banding from the combination of video card and monitor LUT ... without agonizing over the last ounce of linearity. At some point you have to move on from the curve shapes, gamut plots etc. and just trust what your eyes are telling you.
Logged
Serge Cashman
Full Member
***
Offline Offline

Posts: 200


« Reply #35 on: June 15, 2006, 08:23:47 PM »
ReplyReply

Sure. I'm just being geeky about it.
Logged
jani
Sr. Member
****
Offline Offline

Posts: 1604



WWW
« Reply #36 on: June 18, 2006, 06:45:30 PM »
ReplyReply

Quote
That's a big problem with displays these days. The LCD market is still very young and little info is given making it difficult to find decent displays for our use. And, as unnecessarily confusing as people let things get in calibrating LCDs today, just wait until the next technology SED (Surface-conduction Electron-emitter Display) spills into the monitor market in the next year or two. Then after that, OLED (Organic Light-Emitting Diode).

The future promises even better color reproduction and added confusion!
Or perhaps we're simply looking in the wrong places?

A friend of mine who works with dentists pointed out that they were purchasing some nice 3 Mpx displays from Eizo, with 10-bit grayscale from a 13-bit palette. Precision is paramount in medical imaging, so I decided to check out Eizo's pages on medical imaging products, and lo and behold:

EIZO RadiForce R31
3 Mpx
10 bpc
Price: ouch

IIRC, the grayscale GS310 sells for around USD 3,000 plus taxes.  I suppose the colour version is a bit more expensive.

I'd presume that these displays would be useful for photographers, too, but I have no practical way of verifying that.  
Logged

Jan
Serge Cashman
Full Member
***
Offline Offline

Posts: 200


« Reply #37 on: June 18, 2006, 08:25:15 PM »
ReplyReply

<edit> Deleted by author.
« Last Edit: June 18, 2006, 08:38:31 PM by Serge Cashman » Logged
Pages: « 1 [2]   Top of Page
Print
Jump to:  

Ad
Ad
Ad