HCFR - Open source projector and display calibration software - Page 129 - AVS Forum
Forum Jump: 
 29Likes
Reply
 
Thread Tools
post #3841 of 4085 Old 07-04-2014, 05:52 AM
Advanced Member
 
vega509's Avatar
 
Join Date: Apr 2003
Location: The Northeast
Posts: 793
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 17 Post(s)
Liked: 34
thanks Zoyd, I did not notice that tick box.
vega509 is online now  
Sponsored Links
Advertisement
 
post #3842 of 4085 Old 07-04-2014, 09:46 AM
Advanced Member
 
crakarjax's Avatar
 
Join Date: Apr 2007
Location: Raleigh, NC
Posts: 896
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 37 Post(s)
Liked: 18
Quote:
Originally Posted by zoyd View Post
-web updates is disabled as it's the old 2.x version code, any volunteers?
-generator initialization error means you tried to run an internal/madVR pattern set with the DVD generator selection
-to view the Y targets, click the little down arrow next to "editable data" to expand the window
-resizable page elements would need a complete UI overhaul, too much work atm.
The web updater isn't disabled, it just doesn't work -- perhaps disabling it would be a good option though so people don't click it. I'd be happy to take a stab at fixing it but haven't gotten the code to compile (yet).

I used the "HCFR generator" option and got the error, not DVD. "View Images" just quickly runs through the images (without meter attached, maybe behavior is different without meter attached).

I see the editable data arrow and that does work -- perhaps the default should be one tick down?

As far as a UI overhaul, how modular is the UI vs backend processing? I think it would be really cool to turn the project into a backend/frontend project where a service runs in the background with a web UI frontend. This would make the app cross-platform and open the possibility of even running on a remote computer. I'm not sure how the meter data collection would be handled in this case though, as I've never interacted with hardware from a web interface. Thoughts? Due to MFC being used, I think wxWidgets would be the quickest way to bring cross-platform capabilities though.

Edit: so, HCFR seems to work perfectly well in Linux using Wine, the only problem is that the Meters don't show up because Wine doesn't support usb devices well. I know that ArgyllCMS exists for Linux; would it be possible to use the Linux version of this instead of Windows for Wine-specific builds?

Last edited by crakarjax; 07-04-2014 at 10:52 AM.
crakarjax is offline  
post #3843 of 4085 Old 07-04-2014, 11:09 AM - Thread Starter
AVS Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 4,431
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 78 Post(s)
Liked: 299
I've uploaded version 3.1.5 as the latest stable version with the final tweaks to custom colors. I sped up loading and display of the colors compared to the test version.
zoyd is offline  
post #3844 of 4085 Old 07-04-2014, 11:33 AM
AVS Special Member
 
LastButNotLeast's Avatar
 
Join Date: Feb 2007
Location: 08077
Posts: 4,350
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 70 Post(s)
Liked: 210
Quote:
Originally Posted by zoyd View Post
-to view the Y targets, click the little down arrow next to "editable data" to expand the window
-resizable page elements would need a complete UI overhaul, too much work atm.
Worth the price of admission! Thank you!
Solves the other problem, too.

Downloadable FREE demo discs:
http://www.avsforum.com/t/1475769/de...ently-authored 

Did you really need to quote that entire post in your reply?
LastButNotLeast is offline  
post #3845 of 4085 Old 07-04-2014, 11:46 AM - Thread Starter
AVS Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 4,431
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 78 Post(s)
Liked: 299
Quote:
Originally Posted by crakarjax View Post
The web updater isn't disabled, it just doesn't work -- perhaps disabling it would be a good option though so people don't click it. I'd be happy to take a stab at fixing it but haven't gotten the code to compile (yet).

I used the "HCFR generator" option and got the error, not DVD. "View Images" just quickly runs through the images (without meter attached, maybe behavior is different without meter attached).

I see the editable data arrow and that does work -- perhaps the default should be one tick down?

As far as a UI overhaul, how modular is the UI vs backend processing? I think it would be really cool to turn the project into a backend/frontend project where a service runs in the background with a web UI frontend. This would make the app cross-platform and open the possibility of even running on a remote computer. I'm not sure how the meter data collection would be handled in this case though, as I've never interacted with hardware from a web interface. Thoughts? Due to MFC being used, I think wxWidgets would be the quickest way to bring cross-platform capabilities though.

Edit: so, HCFR seems to work perfectly well in Linux using Wine, the only problem is that the Meters don't show up because Wine doesn't support usb devices well. I know that ArgyllCMS exists for Linux; would it be possible to use the Linux version of this instead of Windows for Wine-specific builds?
I don't have any expertise in cross-platform development so I don't know how to answer your questions. Previous discussions on this topic have indicated that due to the heavy use of MFC that any such effort would be very time consuming.
zoyd is offline  
post #3846 of 4085 Old 07-04-2014, 01:51 PM
Advanced Member
 
crakarjax's Avatar
 
Join Date: Apr 2007
Location: Raleigh, NC
Posts: 896
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 37 Post(s)
Liked: 18
Quote:
Originally Posted by zoyd View Post
I don't have any expertise in cross-platform development so I don't know how to answer your questions. Previous discussions on this topic have indicated that due to the heavy use of MFC that any such effort would be very time consuming.
Is that all?
crakarjax is offline  
post #3847 of 4085 Old 07-04-2014, 02:23 PM
AVS Special Member
 
Brian Hampton's Avatar
 
Join Date: Apr 2000
Posts: 6,679
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 41 Post(s)
Liked: 59
Quick question....

I never measured saturation values before and when I did today all the xy data was pretty good but all the luminance values were about +80% (for every color and every value of saturation) So I think I did something wrong.

Maybe it's because when I measured primary and secondary colors I used 75% brightness patterns and some aspect of those measurements carries over to the saturation plots which must be based on 100% brightness patterns.

Sound like the problem?... ??

-Brian
Brian Hampton is offline  
post #3848 of 4085 Old 07-04-2014, 02:36 PM - Thread Starter
AVS Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 4,431
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 78 Post(s)
Liked: 299
Quote:
Originally Posted by Brian Hampton View Post
Quick question....

Maybe it's because when I measured primary and secondary colors I used 75% brightness patterns and some aspect of those measurements carries over to the saturation plots which must be based on 100% brightness patterns.

Sound like the problem?... ??

-Brian
yes, measure primaries at 100% level prior to the RGBCYM saturations.
zoyd is offline  
post #3849 of 4085 Old 07-04-2014, 03:28 PM
Member
 
harlekin's Avatar
 
Join Date: May 2007
Posts: 86
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 31 Post(s)
Liked: 12
@zoyd , today had a bug with the realtime measurements for primary/secondary colors (test build and stable could reproduce even after a restart of the program). Havent tried to repeat it on other systems. edit: used R - G - B mode (caibrated a Samsung)

When taking continuos readings from all prim/sec colors - green wouldn't pop up a field window. The entire screen would go and stay black and thats it. Other colors would pop up the right pattern windows, no problem. If you arent able to reproduce it, maybe its in relation to the german translation setting?

Last edited by harlekin; 07-05-2014 at 06:40 PM.
harlekin is offline  
post #3850 of 4085 Old 07-04-2014, 08:41 PM
AVS Special Member
 
HDTVChallenged's Avatar
 
Join Date: Jan 2002
Posts: 8,312
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 3 Post(s)
Liked: 109
Quote:
Originally Posted by crakarjax View Post
Is that all?
I haven't looked at any of the source code, but I'm guessing that it would probably be faster to start from scratch and build a native Linux app.

To be honest, you could probably do 98% of what (traditional) calibration software does with just a spreadsheet or two... minus the actual meter interface code. 3D LUT generation would be more difficult, of course.
HDTVChallenged is offline  
post #3851 of 4085 Old 07-04-2014, 09:11 PM
Advanced Member
 
gwgill's Avatar
 
Join Date: Jan 2013
Location: Melbourne, Australia
Posts: 561
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 24 Post(s)
Liked: 70
Quote:
Originally Posted by crakarjax View Post
I'm not sure how the meter data collection would be handled in this case though, as I've never interacted with hardware from a web interface. Thoughts?
There are some projects to add more support for local facilities such as USB via Javascript on certain platforms, but I have my doubts about whether they will even be comprehensive enough to support accessing color measurement instruments. I certainly have no interest in translating instlib into Javascript ...
Quote:
Originally Posted by crakarjax
Due to MFC being used, I think wxWidgets would be the quickest way to bring cross-platform capabilities though.
wxWidgets is a great project, but if you are starting from scratch, I'd recommend Qt instead. It's a bit nicer to use, and has more support.
Quote:
Originally Posted by crakarjax
Edit: so, HCFR seems to work perfectly well in Linux using Wine, the only problem is that the Meters don't show up because Wine doesn't support usb devices well. I know that ArgyllCMS exists for Linux; would it be possible to use the Linux version of this instead of Windows for Wine-specific builds?
It's no surprise that many emulated USB interfaces such as those implemented by Wine or Virtual Environments don't work properly for some of the instruments - some of the instruments are not robust in the way they handle USB, and need various workarounds on different platforms, and Wine doesn't seem to allow installing system USB drivers. It's not clear if Wine allows access to the native Linux system, if it did then it would be possible to compile Argyll's instlib for Linux, with some hope of linking it into HCFR. An alternative would be to see if you could compile HCFR using Winelib.

Last edited by gwgill; 07-05-2014 at 07:12 PM.
gwgill is offline  
post #3852 of 4085 Old 07-05-2014, 06:36 PM
Member
 
harlekin's Avatar
 
Join Date: May 2007
Posts: 86
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 31 Post(s)
Liked: 12
Two questions.

1. When comparing color checker SG results between HCFR and Calmans Color Checker it is apparent that HCFR uses the 0-1023 range (10bit RGB) as default, while Calman uses the video range at 64-940. What confuses me is that this is separate from Calmans "expand colors to PC niveau" setting (235>255 (8bit?)).

Question. When the whole device chain (including the TV) is set to provide/expect a Full RGB Signal - but lets say, the source is a Bluray Disc - would the Bluray be displayed at 0-1023 levels, or at 64-940? (It should be mastered at 64-940)

What I'm actually asking is the following. When I set Calman to "expand to PC colors" (and the whole test chain is FULL RGB) and do a color checker run at 0-1023 and one at 64-940 (SMPTE) I get significantly different dE's. If the main purpose of the device (although in a FULL RGB chain) was to watch Blurays, should I optimize for low dE's in 64-940, even on a FULL RGB chain system? Or will the player software output 0-1023 levels automatically, if I've verified that it outputs at "PC levels" (in my head this is all grouped together under the 255 label)?

I'm asking in here, because HCFRs Pattern generator always generates within the 100% Stimulus 0-1023 range, when set to 0-255.

(Calman calls that Option "Luminance Levels" as if it would be different from the color space - is it?)

2. Concerning the bt1886 curve. Even when set to absolute in HCFR (=0), the majority of the curve will lay at about 2.3 gamma. Wasnt bt1886 supposed to bend from 2.0 to 2.4 with the majority of the curve touching 2.4 gamma? Calmans Color checker also produces the lowest dEs (for bt1886) when I set gamma to predominantly touch 2.3, so I assume that the implementation in HCFR is correct (or the other way around.. ), but I've definitely seen people optimize for a 2.0 to 2.4 curve and call it bt1886. Why?

Thank you for any explanations in advance. And then once more after youve given them.

Last edited by harlekin; 07-05-2014 at 06:47 PM.
harlekin is offline  
post #3853 of 4085 Old 07-05-2014, 06:51 PM - Thread Starter
AVS Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 4,431
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 78 Post(s)
Liked: 299
Quote:
Originally Posted by harlekin View Post
Two questions.

1. When comparing color checker SG results between HCFR and Calmans Color Checker it is apparent that HCFR uses the 0-1023 range (10bit RGB) as default, while Calman uses the video range at 64-940. What confuses me is that this is separate from Calmans "expand colors to PC niveau" setting (235>255 (8bit?)).

Question. When the whole device chain (including the TV) is set to provide/expect a Full RGB Signal - but lets say, the source is a Bluray Disc - would the Bluray be displayed at 0-1023 levels, or at 64-940? (It should be mastered at 64-940)

What I'm actually asking is the following. When I set Calman to "expand to PC colors" (and the whole test chain is FULL RGB) and do a color checker run at 0-1023 and one at 64-940 (SMPTE) I get significantly different dE's. If the main purpose of the device (although in a FULL RGB chain) was to watch Blurays, should I optimize for low dE's in 64-940, even on a FULL RGB chain system? Or will the player software output 0-1023 levels automatically, if I've verified that it outputs at "PC levels" (in my head this is all grouped together under the 255 label)?

I'm asking in here, because HCFRs Pattern generator always generates within the 100% Stimulus 0-1023 range, when set to 0-255.
I'm not following you here, both CalMAN and HCFR address the video card frame buffer in 8 bits. Both programs generate the same colorchecker 8 bit triplets and when set to video levels (16-235) both program map 0% to 16,16,16 and 100% to 235,235,235. Expand to PC levels in CalMAN is the same as "view images" 0-255 in HCFR.

Regarding DVD playback etc. I recommend calibrating the entire chain in video levels (not full levels), either RGB or YCbCr. Although it really shouldn't matter as long as you keep display levels matched to playback levels.

Quote:
Originally Posted by harlekin View Post
2. Concerning the bt1886 curve. Even when set to absolute in HCFR (=0), the majority of the curve will lay at about 2.3 gamma. Wasnt bt1886 supposed to bend from 2.0 to 2.4 with the majority of the curve touching 2.4 gamma? Calmans Color checker also produces the lowest dEs when I set gamma to predominantly touch 2.3, so I assume that the implementation in HCFR is correct (or the other way around.. ), but I've definitely seen people optimize for a 2.0 to 2.4 curve and call it bt1886. Why?
If you plot the formula for various black levels you'll find that it deviates from 2.4 over the entire range unless you have very deep blacks, well under 0.01 cd/m^2

Last edited by zoyd; 07-05-2014 at 07:00 PM.
zoyd is offline  
post #3854 of 4085 Old 07-06-2014, 12:24 AM
Member
 
harlekin's Avatar
 
Join Date: May 2007
Posts: 86
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 31 Post(s)
Liked: 12
One more try.

Back to the "Luminance Level" setting in Calman. Video chain is FULL RGB, expand to PC levels in Calman is enabled.

When set to 0-1023 (Full) Calmans color checker produces about the dE's I see in HCFRs color checker.

When set to 64-940 (SMPTE Levels) avg dE increases about 0,4 on the same color checker measurements. The increase is uniform - so almost all colors are effected by about the same amount.

So this setting has an impact.

Now - what does that mean regarding the rec 709 recording standard.

(Also when the whole chain is set to limited, and PC levels is disabled - the non default 0-1023 for me produces lower dE's - that then are in sync with HCFR readings.)
-

Or simply put - I followed your explanation effortlessly, because so far that was was my understanding about how this should work. But then Calman has this extra "Luminance Level" setting - which effects results. And defaults to something (64-940) that is a "limited range" (?) and responsible for producing different dEs than I get with HCFR.

I will try to look into it further (google fu), but this isn't something that should be glanced over imho. Either setting must be defined as "wrong" when optimizing the video chain. Which means, that by proxy also HCFR is effected.
-

As for configuring the entire chain to limited. Yes, that is the cop out everyone seems to fall back to - because "less video information conversion". You can see from my graphs that I don't have black crush (because both devices output/expect FULL RGB) - but the player software I use most often (XBMC) outputs PC levels (0-255) so I have an incentive to optimize for that. Also, at least when using my i5/Sony TV combo - the FULL RGB chain has lower dEs than the limited chain on the same systems (settings on both devices set to limited) - not by a significant margin, but reproduceable, so that lets me make assumptions about the TVs internal signal processing...
-

edit: And thanks for the response on the gamma curve, I follow that one (dependened on the measured black level the curve flattens).

Last edited by harlekin; 07-06-2014 at 12:36 AM.
harlekin is offline  
post #3855 of 4085 Old 07-06-2014, 12:49 AM
Member
 
harlekin's Avatar
 
Join Date: May 2007
Posts: 86
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 31 Post(s)
Liked: 12
The more I read about it, the more it seams that it is "just" the difference between 8 and 10 bit output.

0-255 equals 0-1023 (10 bit)
16-235 equals 64-940 (10 bit)

So why on earth would Calman have a "output PC levels" option, which even changes the color codes on the pattern window (expands to 255) if they then in return still output (or read?) a "limited range" if this other setting isnt touched.

zoyd - you used Calman to verify LUT results, you must have an opinion on it. (no pressure.. )
harlekin is offline  
post #3856 of 4085 Old 07-06-2014, 11:27 AM - Thread Starter
AVS Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 4,431
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 78 Post(s)
Liked: 299
Quote:
Originally Posted by harlekin View Post
The more I read about it, the more it seams that it is "just" the difference between 8 and 10 bit output.

0-255 equals 0-1023 (10 bit)
16-235 equals 64-940 (10 bit)

So why on earth would Calman have a "output PC levels" option, which even changes the color codes on the pattern window (expands to 255) if they then in return still output (or read?) a "limited range" if this other setting isnt touched.

zoyd - you used Calman to verify LUT results, you must have an opinion on it. (no pressure.. )
CalMAN's Luminance levels = PC produces exactly the same 8-bit triplets out the video card as Luminance levels = video + "expand internal patterns". The difference between the two configurations is the luminance scale that is used internally for the production of graphs etc. When luminance levels = video, 100% stimulus = reference white and 109% = peak white. When luminance levels = PC, 100% stimulus = both peak white and reference white. In the first case there are 219 levels between black and reference white and in the second case there are 255 levels between black and reference white. If you set your display to full range they should measure identically because the same full range triplets are being sent through the video chain. I just tried this to verify and measured no difference between the two within my set-up error, average dE's of the color checker were within 0.05.

You should also obtain the same average dE errors if you set CalMAN to video levels and your display is set to accept video levels.
If you find a difference between these two settings and you are sure you have set your display to process the correct level spacing, then it is your display that treats the two inputs differently.

HCFR does not have a luminance scale switch, everything done internally is on the video level scale (219 codes over the black to reference white range). When you check 0-255 in the generator set-up box this is exactly the same as CalMAN's video level + "expand internal patterns" configuration. I have also measured HCFR against itself using these two settings as well as HCFR against CalMAN and found no discrepancies in average dE of the color checker in my set-up.
zoyd is offline  
post #3857 of 4085 Old 07-06-2014, 11:54 AM
Member
 
harlekin's Avatar
 
Join Date: May 2007
Posts: 86
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 31 Post(s)
Liked: 12
Thank you for the explanation - will look into it further (must take some time to digest the information in detail). Definitely had a avg dE difference of 0.4 between measurement runs, both on a full RGB chain, with 0-1023 > 64-940 as the only parameter changed in between. I actually am sure about it. Both taken at the 75% luminance level Calman defaults to for color measurements.

Btw I _love_ the implemented color checker SG and USER in HCFR - probably most, because it turns out I missed a certain greyscale setting and my TV could be optimized a bit further. Its so much easier to have the calibration and evaluation parts both in the same program, it is invaluable. It also helps that I'm navigating about five times faster within HCFR than in Calman. It also helps that you have the individual results up front, and the graphs can be pulled up when needed. God I love this program. Thank you so much for maintaining the project.

Enough with the sweet talk, lets start the bragging once more..



(Configuration is suitable measured against both 2.2 and bt1886 reference gamma (gamma slider 0>-1), with 2.2 producing a tiny amount less dE (so tiny it is almost insignificant). Configuration is suitable for 92-MAX contrast and 100 to 150 cd/m2 without any of its dE characteristics changing significantly.)

Quote:
If you find a difference between these two settings and you are sure you have set your display to process the correct level spacing, then it is your display that treats the two inputs differently.
Have to double check. After your last posting, today I checked the performance of a different source device on a limited color chain and found it to be exactly within the parameters I expected from my RGB full configurations.

The majority of the uncertainty potential in my case comes from me having used Calmans Color Checker before to evaluate configurations - with it measuring at 75% stimulus and these configurations not adhering to the old (GDC at 100% stimulus) color checker "vectors" (no unified trend on "how the display performs").

Last edited by harlekin; 07-06-2014 at 01:08 PM.
harlekin is offline  
post #3858 of 4085 Old 07-06-2014, 12:02 PM
AVS Special Member
 
Holiday121's Avatar
 
Join Date: Feb 2009
Posts: 1,646
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 35 Post(s)
Liked: 35
Nice graphs was that with a lumagen or any processing
Holiday121 is offline  
post #3859 of 4085 Old 07-06-2014, 01:04 PM
Member
 
harlekin's Avatar
 
Join Date: May 2007
Posts: 86
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 31 Post(s)
Liked: 12
Quote:
Originally Posted by Holiday121 View Post
Nice graphs was that with a lumagen or any processing
Nope, just internal CMS - and the unprocessed output of my HTPC (Haswell i5 driving the graphics). (edit: Output confirmed neutral, using several other source devices.)

TV is a Sony KDL42W65* (KDL42W650A) with the VA-type LCD panel from AUO - which was a 500 Euro purchase at the end of its lifespan.

Curiously enough, the thing doesnt even have 10 pt grayscale (best greyscale configurations are possible in w2 and w1 mode (cinema or gaming preset (same characteristics))), or a CMS for that matter. Color set to 49, hue to R1, the rest was getting the greyscale "just right". hdtvtest.co.uk tested the model and commented that setting clear white to "low" would be a better calibration base setting - which cost me about 10 hours calibrating (while still learning at the same time), that I wont get back, because it turns out - doing so worsened its greyscale tracking on my model. *hrm*

Last edited by harlekin; 07-06-2014 at 01:12 PM.
harlekin is offline  
post #3860 of 4085 Old 07-06-2014, 01:21 PM - Thread Starter
AVS Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 4,431
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 78 Post(s)
Liked: 299
Quote:
Originally Posted by harlekin View Post

Have to double check. After your last posting, today I checked the performance of a different source device on a limited color chain and found it to be exactly within the parameters I expected from my RGB full configurations.

The majority of the uncertainty potential in my case comes from me having used Calmans Color Checker before to evaluate configurations - with it measuring at 75% stimulus and these configurations not adhering to the old (GDC at 100% stimulus) color checker "vectors" (no unified trend on "how the display performs").
The color checker color difference errors should not depend on how you have configured the reference white stimulus value (i.e. 75%), I think that is only used for the colorspace dE errors. The color checker always uses 100% stimulus as reference white. Here is a summary of some comparisons I did with the CCSG between the two programs. All four configurations tell the same story.

zoyd is offline  
post #3861 of 4085 Old 07-06-2014, 02:48 PM
Member
 
harlekin's Avatar
 
Join Date: May 2007
Posts: 86
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 31 Post(s)
Liked: 12
Thx. Will put up some data as well after I've done a few measurements. Also I will make sure that my version of Calman is up to date. The 0.4 avg deviation was a recent result I got though - after I've had the other factors like getting to know the programs (settings), source inputs - and configurations pretty much ironed out..

Its on me now to try to close in on the problem, you've provided me with all the background information I should need. I will give a report as soon as I've done so.
-

With my previous configuration (with a greyscale deviation (blue) at IRE 10 and 20) I noticed somewhat differing dE's measuring with 100% and 75% stimulus, even in the (100% saturation) primaries (blue, red) - just using HCFRs pattern generator.

As I now have a configuration that is pretty much optimal in adhering to the standard in theory (in that there are almost no breakout values), I can look at two comparatively as well.

Last edited by harlekin; 07-07-2014 at 12:23 AM.
harlekin is offline  
post #3862 of 4085 Old 07-06-2014, 03:17 PM
AVS Special Member
 
Holiday121's Avatar
 
Join Date: Feb 2009
Posts: 1,646
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 35 Post(s)
Liked: 35
Very nice then. Wish my results were similar haha. Mine keep getting worse.

Question
Adjusting the 10 point greyscale in my projector gamma does that mess up colors?

Because the only way I get my dE so low is adjusting each point in the 10 point gamma
Holiday121 is offline  
post #3863 of 4085 Old 07-06-2014, 04:51 PM
AVS Special Member
 
Holiday121's Avatar
 
Join Date: Feb 2009
Posts: 1,646
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 35 Post(s)
Liked: 35
Question in my 10 point gamma adjustment I have a white as well as rgb what does white do
Holiday121 is offline  
post #3864 of 4085 Old 07-06-2014, 11:56 PM
Member
 
harlekin's Avatar
 
Join Date: May 2007
Posts: 86
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 31 Post(s)
Liked: 12
Yes, greyscale impacts colors. You can think of it as the "exact color of paper" on which you then draw upon with colored pens. Of course the analogy falls apart even before the start, because RGB is an additive color system and therefore by definition there can be no "physical medium" equivalent.

If you are looking for a more factual explanation ( ): http://www.techterms.com/definition/grayscale


No clue what white does in greyscale settings (luminance? beamer specific?). Also calibrating greyscale AFTER youve applied a 3D LUT is a no go.
-

Also, I tried _many_ greyscale configurations until I hit those setting. I had other ones where greyscale was even more perfect (as you can see the RGB lines arent aligned perfectly), but they produced higher dE's in color checker.

BUT. You always have to keep the 6500K graph (white line) in mind. From my experience it is detrimental that it kisses the D65 line (at least from 30 IRE to possibly 100 IRE) and does not deviate in any direction (up/down), not even for about 100K (if your display is capable of such a configuration), because it will negatively impact color performance.

Or in short - I did "compensate" (measured again, and again, and again, ...) display characteristics with very specific greyscale settings, but NEVER to the point where the D65 graph would visibly suffer.

(f.e. on warm 2 (2pt greyscale) -1,-4,0;0,0,0(/-1) would deliver more accurate greyscale, but -2,-5,0;-1,-1,-2 more accurate color performance (with the greyscale dE's you see in the screengrabs I posted).

(btw. (might be useful to some) Irfanview (Freeware img editor) can be used to very easily create a "panorama image" of several screengrabs in a column (workflow is print(take screengrab)/ctrl+v into Irfanview/ctrl+H to original size/draw rectangle(mouse) and ctrl+Y to crop/S to save as -- once you've all individual screengrabs: image/create panorama)

Last edited by harlekin; 07-07-2014 at 12:58 AM.
harlekin is offline  
post #3865 of 4085 Old 07-07-2014, 02:27 AM
AVS Special Member
 
Holiday121's Avatar
 
Join Date: Feb 2009
Posts: 1,646
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 35 Post(s)
Liked: 35
Hmmmm. After some google searching still can't find what the white does. It's a 10 point gamma just like the r g b.

That's where I think my problem is coming along. I am tweaking my greyscale way to much to perfect it and it's throwing everything else off.

I am going to factory reset my projector Tuesday morning and try to start over. Most say my projector is pretty close straight out the box.

Hopefully I can get it a little better then it was.

My 3dlut i created so far is no good so I think it might be something wrong badly in my original calibration before it.

So should I concentrate on the color temp 6500k bar being straight before I start playing with gamma 10 point or rgb
Holiday121 is offline  
post #3866 of 4085 Old 07-07-2014, 06:10 AM
AVS Special Member
 
Holiday121's Avatar
 
Join Date: Feb 2009
Posts: 1,646
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 35 Post(s)
Liked: 35
Also in hcfr running the primary and secondary color checker in madtpg what do I want to shoot for in dEs for the colors all under 2?

And does color effect primary and hue tint effect secondary
Holiday121 is offline  
post #3867 of 4085 Old 07-07-2014, 06:30 AM
Member
 
harlekin's Avatar
 
Join Date: May 2007
Posts: 86
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 31 Post(s)
Liked: 12
As low as you can possibly get them. They are your baselines for what color performance will be in each "sector". As you can only configure the 100% saturation point in any (most?) CMS this is the part where "it all should line up", or it doesnt.

And some of it can be undefeatable device characteristics (as the color slider just positions ALL color points in relation to the center of the CIE triangle (closer, farther) and the hue slider just rotates them on an axis).

For dialing in color and hue, look at the primaries and secondaries table (just the colors at 100% saturations), as most tweaks should be especially visible there. Also in the process adjust for avg dE as low as possible (not necessarily for lowest max dE). Write down possible configurations options (also those with low max dE) - you will check them against the color checker SG values later to determine their effect on the broader color spectrum.

The tool most useful for that step (beside the actual dE values in color checker SG) then is the dE saturation graph (the bottom one of the saturation graphs). Try not to have any outliers, while keeping dE as low as you can.

edit: When looking at the CIE triangle also keep two things in mind. You wont see luminance errors (the "third dimention" and factor for calculating dE) - also (if Ive understood that right) the dE formulas allow for more pronounced errors in certain "orientations" - while the orientation (to the left, to the right, more upleft, ...) changes depending on the position of the color point in the triangle > meaning - when it comes to color checker values dont give too much about the positioning within the CIE triangle, go by the dE values of that color point.

Last edited by harlekin; 07-07-2014 at 06:56 AM.
harlekin is offline  
post #3868 of 4085 Old 07-07-2014, 06:52 AM
AVS Special Member
 
Holiday121's Avatar
 
Join Date: Feb 2009
Posts: 1,646
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 35 Post(s)
Liked: 35
Hmmm can you post a example
Of the graph? I'm at work trying to understand right now.
I'm off Tuesday and wed so I am going to Attempt the again from scratch.
Holiday121 is offline  
post #3869 of 4085 Old 07-07-2014, 06:54 AM
AVS Special Member
 
Holiday121's Avatar
 
Join Date: Feb 2009
Posts: 1,646
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 35 Post(s)
Liked: 35
So basically for the color.

Measure primary and secondary and adjust the color and tint to achieve lowest dE possible for them.

At what point is it right to adjust my 10point gamma.
Holiday121 is offline  
post #3870 of 4085 Old 07-07-2014, 06:58 AM
AVS Special Member
 
Holiday121's Avatar
 
Join Date: Feb 2009
Posts: 1,646
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 35 Post(s)
Liked: 35
And also if I am
Correct your highest dE was 2.7 for blue correct?

Did you adjust color to bring down red and green and blue so you favored red?
Holiday121 is offline  
Reply Display Calibration

User Tag List

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off