Luminous Landscape Forum

Raw & Post Processing, Printing => Digital Image Processing => Topic started by: ato on July 06, 2006, 12:49:51 am

Title: delta E result
Post by: ato on July 06, 2006, 12:49:51 am

(http://www.atoyu.com/image/de.gif)

anyone know why only gray color so high?

Title: delta E result
Post by: Stephen Best on July 06, 2006, 02:30:07 am

What this means is that basICColor display has done a great job of mapping where the primaries are (not too hard with LUT profiles) but not so good on neutrality. I guess it depends on how neutral your monitor is uncalibrated. After a lot of playing around with display 4, I went back to Match 3.6 with whitepoint/gamma set to native (namely a linear LUT for the video card). These figures look pretty good overall though.

BTW, did you set your monitor to a gamma of 2.4 first? Does your monitor allow you to do so losslessly?

Title: delta E result
Post by: Serge Cashman on July 06, 2006, 02:49:57 am

Quote

...I went back to Match 3.6 with whitepoint/gamma set to native (namely a linear LUT for the video card).

Namely "leave your monitor as it is and measure the output".

Although those delta E differences are indistinguishable to  a human observer it is interesting why greyscale differences are higher.

I bet L* gamma has something to do with it.

Stephen, what do you mean by setting the 2.4 gamma first?

Title: delta E result
Post by: Stephen Best on July 06, 2006, 02:58:41 am

Quote

Stephen, what do you mean by setting the 2.4 gamma first?
[a href=\"index.php?act=findpost&pid=69895\"][{POST_SNAPBACK}][/a]

A gamma of 2.4 is as close as you can get to an L* curve. If the monitor doesn't have a 10-bit (or greater) LUT I wouldn't even attempt L* calibration.

Title: delta E result
Post by: Hermie on July 06, 2006, 03:21:48 am

Quote

A gamma of 2.4 is as close as you can get to an L* curve. If the monitor doesn't have a 10-bit (or greater) LUT I wouldn't even attempt L* calibration.
[{POST_SNAPBACK}][/a] (http://index.php?act=findpost&pid=69896\")

Serge,

See graph on page 8 (text is in German language):
[a href=\"http://homepage.mac.com/hanspeterharpf/LStar-RGB/LStar-Dokumentation.pdf]http://homepage.mac.com/hanspeterharpf/LSt...kumentation.pdf[/url]

Hermie

Title: delta E result
Post by: Serge Cashman on July 14, 2006, 12:07:39 am

Hmm... I have an even more telling result from Basiccolor. It's all LUT-based on a very average LCD (Dell 2001FP). Stephen Best is correct I believe - it doesn't do such a good job on neutrality, at least with LUTs. Since all those differences may be in different directions from the targets it may be even worse than it looks on the chart.


I'll keep looking at different settings though - it's the first time I use Basiccolor.

[attachment=817:attachment]

Title: delta E result
Post by: Serge Cashman on July 15, 2006, 02:45:46 am

Quote

A gamma of 2.4 is as close as you can get to an L* curve. If the monitor doesn't have a 10-bit (or greater) LUT I wouldn't even attempt L* calibration.
[{POST_SNAPBACK}][/a] (http://index.php?act=findpost&pid=69896\")

OK, I'm trying to make sence of this. Can't find much in English...

[a href=\"http://lists.apple.com/archives/colorsync-users/2005/May/msg00344.html]http://lists.apple.com/archives/colorsync-...y/msg00344.html[/url]


Of course "L* is a function which is defined like most of us know:
http://www.brucelindbloom.com/Eqn_XYZ_to_Lab.html (http://www.brucelindbloom.com/Eqn_XYZ_to_Lab.html)"
is a very discouraging statement for the rest of us.

However Peter's statement at the end of the post (to the extent of "screw L*, just look at colormanaged applications")  does make sence and corresponds to my understanding of the process. But I think I do get the L* idea.

Arn't most displays (and sRGB gamma for that matter) close to 2.4 with some offset though?

Title: delta E result
Post by: Dinarius on August 29, 2006, 07:19:16 am

I have calibrated my screen using both basICColor Display 4 and Eye-One Match 3.

Both have been done using the Eye-One Match Display 2 profiling instrument.

The 'Validate' feature at the end of the basICColor program allows any profile to be measured. You simply choose File/Select system profile..... from the Display 4 window and then select the profile of your choice.

I have compared both calibrations using this feature and the difference is astonishing.

While I too have greyscale readings that are a little higher, overall the bar chart for the 23 scales is pretty even.

The bar chart for my Eye-Match 3 profile, on the other hand, is all over the place!

As in the chart below, the bar chart for the Display 4 profile never extends beyond the green zone. In the Eye-One Match 3 bar chart, some of the bars go into the red zone.

I have been told to profile with the System Profile set to LStar when using Profile 4. Do others do this?

Having done this, presumably it's OK to then have CS2 set to Adobe 1998?

Thanks.

D.

Title: delta E result
Post by: Dinarius on August 29, 2006, 07:25:37 am

One other thing........

Regarding the higher greyscale values....

If you look at the example in the Display 4 manual (p.35), you will see that they are higher there too! ;-)

D.