HCFR - Open source projector and display calibration software - Page 125 - AVS Forum
First ... 123  124  125 126  127  ... Last
Display Calibration > HCFR - Open source projector and display calibration software
thegladiator75's Avatar thegladiator75 11:25 AM 06-07-2014
Hi Zoyd, I own an i1dpro and a VT60 and I would like to understand why my readings of 0 IRE shows something like the red at 912%, blue at 0% and green at -47%. And it's weird because Calman reading of 0 IRE shows a perfectly balanced black...

HDTVChallenged's Avatar HDTVChallenged 12:17 PM 06-07-2014
Quote:
Originally Posted by thegladiator75 View Post

Hi Zoyd, I own an i1dpro and a VT60 and I would like to understand why my readings of 0 IRE shows something like the red at 912%, blue at 0% and green at -47%.

That's normal. Your black level is darker than the meter can read.
Quote:
And it's weird because Calman reading of 0 IRE shows a perfectly balanced black...

I suspect Calman recognizes that the reading is "bogus" and doesn't bother to display it. Possibly to avoid messing up the gamma calculations, etc.

What happens with a 5% (or 10%) target?
zoyd's Avatar zoyd 02:39 PM 06-07-2014
Version 3.1.2 has been uploaded to sourceforge.

  • Updated APL patterns from fixed background video level to average per channel video level.
  • Added single decimal precision to RGB/LCH bar text.
  • Disabled RGB/LCH Bars and target graphs when not in grayscale, colorspace, or freemeasures pages.
  • Added BT.1886 relative option [see ArgyllCMS collink -Ib:x.x option].
  • All dE and graph targets now use actual BT.1886 gamma values instead of user measured average.
  • Pure power law targets updated from fixed 2.22 to user specified via the Reference value.
  • Fixed bug when number of saved color checker measures conflicted with CCSG selection.



Set the BT.1886 gamma value to 0 for normal behavior. Set it to X.X to shift the curve up/down such that the point gamma at 50% is X.X. I like calibrating this to 2.4 on my display for dark room viewing.
Quote:
Originally Posted by HDTVChallenged View Post

The only issue here is whether or not HCFR is adjusting it's expected target values based on the measured (average) gamma vs the disk encode gamma, which it does ... apparently, as long as you don't bring BT1886 into the mix. (It does for BT1886 too, but the adjustments can't be trusted since bt1886 <> power-law gamma.)

This update takes care of this issue, all targets are now calculated based on actual BT.1886 transfer characteristics if you have selected that option as your reference. Same goes for the power law option and "use measured" as an average is still available when using a power law reference but it's no longer very useful unless you are targeting something other than power law or BT.1886.
HDTVChallenged's Avatar HDTVChallenged 03:55 AM 06-08-2014
Quote:
Originally Posted by zoyd View Post

Version 3.1.2 has been uploaded to sourceforge.

This update takes care of this issue, all targets are now calculated based on actual BT.1886 transfer characteristics if you have selected that option as your reference. Same goes for the power law option and "use measured" as an average is still available when using a power law reference but it's no longer very useful unless you are targeting something other than power law or BT.1886.

Sounds good. I just spent a couple of days learning *all* about Luv, Lab, the LCH representations thereof and their relationships to the base CIEXYZ and xyY colorspace. My head's about to explode. wink.gif

During the marathon, I discovered that HCFR has always based the LCH values on the Lab colorspace ... which explains why the "H" values never made sense to me. I could never figure out why they changed dramatically as the saturation level changed. Now it does ... different scales for a* and b* gets ya every time. wink.gif I made a spreadsheet to compare all of the various possible color spaces and error deltas, so far it doesn't look like the Lab base would have a big effect on the LCH target bars.
zoyd's Avatar zoyd 05:27 AM 06-08-2014
Quote:
Originally Posted by HDTVChallenged View Post

Sounds good. I just spent a couple of days learning *all* about Luv, Lab, the LCH representations thereof and their relationships to the base CIEXYZ and xyY colorspace. My head's about to explode. wink.gif

At some point I'll probably add more choices to the bar indicators, including the components of dE.
Holiday121's Avatar Holiday121 07:44 AM 06-08-2014
Thanks for the updates zoyd.

I haven't got around to run my test again on my htpc been busy at work will get to it Monday or Tuesday
HDTVChallenged's Avatar HDTVChallenged 10:53 AM 06-08-2014
Quote:
Originally Posted by zoyd View Post

At some point I'll probably add more choices to the bar indicators, including the components of dE.

I haven't gotten any farther than the basic cartesian dE formulas .... and the deltaH*(uv) component. .... dE94 and especially dE2000, OTOH .... eek.gif I need a few days of rest and recovery first. smile.gif
HDTVChallenged's Avatar HDTVChallenged 12:48 AM 06-09-2014
Quote:
Originally Posted by zoyd View Post

At some point I'll probably add more choices to the bar indicators, including the components of dE.

Ok ... I did some crosschecking between my latest craptacular spreadsheet and HCFR 3.1.2 with data from my last calibration run. The good news is that all of the LCh (ab) values line up. The not so good news is that delta-hue differences between LCh (ab) and LCh (u*v*) might be big enough to cause a click or two worth of "tint" adjustment difference on red and cyan. FWIW, the RGB bar method (@75% sat) lines up more with the L*u*v* delta-hue values. I don't know if this is a big issue or a small issue ... just thought I'd mention it. I confess that I don't yet understand the L*a*b* choice for display measurement, other than perhaps it allows for "legal" L* values >100 ... otherwise it just confuses me ... especially the effect on hue values. smile.gif

In the further good news department, it looks like the BT1886 target adjustment is working ... that pesky "light skin" color-checker target finally falls in the box with BT1886. A nice refinement there. smile.gif
Holiday121's Avatar Holiday121 03:21 AM 06-09-2014
Quote:
Originally Posted by gwgill View Post

No. spotread -D8 is going to give all the information that is available, short of running it under a debugger, and single stepping it.
Something is very strange on your system.

Ok finally got around to doing the spotread -D8 here it is

spotread-D8.txt 4k .txt file
Attached: spotread-D8.txt (4.4 KB) 
zoyd's Avatar zoyd 07:58 AM 06-09-2014
Quote:
Originally Posted by HDTVChallenged View Post

Ok ... I did some crosschecking between my latest craptacular spreadsheet and HCFR 3.1.2 with data from my last calibration run. The good news is that all of the LCh (ab) values line up. The not so good news is that delta-hue differences between LCh (ab) and LCh (u*v*) might be big enough to cause a click or two worth of "tint" adjustment difference on red and cyan. FWIW, the RGB bar method (@75% sat) lines up more with the L*u*v* delta-hue values. I don't know if this is a big issue or a small issue ... just thought I'd mention it. I confess that I don't yet understand the L*a*b* choice for display measurement, other than perhaps it allows for "legal" L* values >100 ... otherwise it just confuses me ... especially the effect on hue values. smile.gif

In the further good news department, it looks like the BT1886 target adjustment is working ... that pesky "light skin" color-checker target finally falls in the box with BT1886. A nice refinement there. smile.gif

Thanks for the feedback, I think it will make the most sense for knob tweaking to make the bars selectable between RGB and HSL coordinates as those are most likely to linearly proportional to CMS system adjustments. And then add a separate dE indicator to look at trade-offs.
HDTVChallenged's Avatar HDTVChallenged 11:45 AM 06-09-2014
Quote:
Originally Posted by zoyd View Post

Thanks for the feedback, I think it will make the most sense for knob tweaking to make the bars selectable between RGB and HSL coordinates as those are most likely to linearly proportional to CMS system adjustments. And then add a separate dE indicator to look at trade-offs.

Sounds reasonable.

It did occur to me that the Lab based delta-h might be enhancing the differences between power-law and bt1886. My craptacular spreadsheet doesn't account for that, and I was using data from a bt1886 run. I should probably recheck this during a "live" session to make sure it's not a power-law vs. bt1886 artifact.
HDTVChallenged's Avatar HDTVChallenged 11:12 PM 06-09-2014
Zoyd,

On third thought, perhaps the choice of RGB, HSL or LCh(u*v*) ... or Lsh(u*v*) bars? Lsh(u*,v*) seems to match the way CMS (HSL) controls usually work, while retaining the references to CIE color spaces and user defined white-points. HSL(rgb) loses those REC709/601 anchor points. I'm not sure what "HSL" version you meant ... I assumed Lsh(u*,v*) above, then I did more research.
gwgill's Avatar gwgill 12:02 AM 06-10-2014
Quote:
Originally Posted by Holiday121 View Post


Ok finally got around to doing the spotread -D8 here it is

spotread-D8.txt 4k .txt file
Sorry, that doesn't add anything. Spotread just dies or is killed without a message sometime around trying to open the device.

You could try this version, and see if it behaves any differently
Holiday121's Avatar Holiday121 08:41 AM 06-10-2014
Will do and post back thanks
Synnove's Avatar Synnove 05:17 PM 06-10-2014

Thanks to the generosity of Klein Instruments, I was able to procure readings from their CS2000 of my Kuro's red, blue, green, and white levels in xyY format. I'm trying to create conversion matrices for DispcalGUI as well as HCFR. I've read up on the process one uses for entering the corrections in to HCFR, however I cannot seem to get a proper result. Does anyone have some suitable advise for addressing this?


Stereodude's Avatar Stereodude 05:52 PM 06-10-2014
Isn't there a spreadsheet floating around somewhere that you enter the data into and it generates the corrections?
gwgill's Avatar gwgill 05:56 PM 06-10-2014
Quote:
Originally Posted by Synnove View Post

Thanks to the generosity of Klein Instruments, I was able to procure readings from their CS2000 of my Kuro's red, blue, green, and white levels in xyY format. I'm trying to create conversion matrices for DispcalGUI as well as HCFR. I've read up on the process one uses for entering the corrections in to HCFR, however I cannot seem to get a proper result. Does anyone have some suitable advise for addressing this?
Without some details of what steps you have tried, and where you got stuck, it's hard to help.
Synnove's Avatar Synnove 06:27 PM 06-10-2014

I shall look for the spreadsheet, thank you for the suggestion!

 

gwgill: sorry, let me start with the actual measurements:

 

White
0.3152
0.3285
100.8

 

Red

0.6681
0.3239
23.59

 

Blue
0.1468
0.0631
7.5

 

Green
0.2755
0.6502
72.79

 

From my understanding, if I wish to use these measurements with HCFR, I need to the meter property page and click the "sensor matrix" tab. I assume I'm supposed to convert my primary color measurements in to XYZ and paste them in the grid as RGB from top to bottom?

 

With DispcalGUI, I need to do more research on the format of the file as the instructions for it are somewhat confusing for a layperson like myself.


gwgill's Avatar gwgill 08:34 PM 06-10-2014
Quote:
Originally Posted by Synnove View Post

From my understanding, if I wish to use these measurements with HCFR, I need to the meter property page and click the "sensor matrix" tab. I assume I'm supposed to convert my primary color measurements in to XYZ and paste them in the grid as RGB from top to bottom?
Sorry, I can't help you with HCFR.

A thread on creating an ArgyllCMS .ccmx is here. I'd imagine that a .ccmx is usable both in HCFR and ArgyllCMS + DispcalGUI.
HDTVChallenged's Avatar HDTVChallenged 02:03 AM 06-11-2014
Quote:
Originally Posted by Synnove View Post

I shall look for the spreadsheet, thank you for the suggestion!

From my understanding, if I wish to use these measurements with HCFR, I need to the meter property page and click the "sensor matrix" tab. I assume I'm supposed to convert my primary color measurements in to XYZ and paste them in the grid as RGB from top to bottom?

I can't tell what you're trying to do here. Are you trying to profile another meter with this data? If so, it sounds like you're making the process harder than it needs to be (in HCFR.)
zoyd's Avatar zoyd 04:30 AM 06-11-2014
Quote:
Originally Posted by Synnove View Post

I shall look for the spreadsheet, thank you for the suggestion!


From my understanding, if I wish to use these measurements with HCFR, I need to the meter property page and click the "sensor matrix" tab. I assume I'm supposed to convert my primary color measurements in to XYZ and paste them in the grid as RGB from top to bottom?

With DispcalGUI, I need to do more research on the format of the file as the instructions for it are somewhat confusing for a layperson like myself.

Enter those measurements in a synthetic probe page and check the reference box. Then you need the corresponding measurements for those patches from the probe you want to profile (either entered manually in the primaries page or measured). Then select advanced->meter correction file->create from existing measurements.
zoyd's Avatar zoyd 05:07 AM 06-12-2014
Sorry for the rapid updates but I tied up some loose ends related to the last update and added a coordinate option for the levels indicators as well as a dE bar.

3.1.3
-------
  • Fixed creation of matrix correction using manual reference entry
  • Fixed display of near black/near white selected colors
  • Updated dE of near black to use gray world weighted white reference
  • Updated special colorspace targets to BT.1886 gamma if selected
  • Added dE bar to real-time levels indicator
  • Added option for HSV coordinates to real-time levels indicator
  • Fixed luminance and satshift graphs when using BT.1886
  • Fixed delta luminance for saturation sweeps when using BT.1886

The added option is under user measures (same place as "detect primary colors...") as a checkbox for "Use HSV Level Bars". If left unchecked RGB coordinates will be used. These coordinate indicators should provide better correlation with HSV(L) CMS display controls for real-time adjustments. I've set the "V" to be identical to Luminance level, H is hue angle with values > 100% a rotation counter clockwise from target. S is a measure of the distance to the white point.

The RGB bars should also be more usable for RGB based CMS systems as they have been changed to actual coordinate differences instead of white balance values. Normal white balance indicators are maintained when you are measuring a gray/white patches.

Also put in a 4th bar with dE level. [Note: dE updating in continuous mode for the gray scale uses the same absolute dE formula as the RGB measures info window and not the user selected formula]
Attached: hcfr313.PNG (166.2 KB) 
rickardl's Avatar rickardl 10:22 AM 06-12-2014
Quote:
Originally Posted by zoyd View Post
The RGB bars should also be more usable for RGB based CMS systems as they have been changed to actual coordinate differences instead of white balance values. Normal white balance indicators are maintained when you are measuring a gray/white patches.
Outstanding...
HDTVChallenged's Avatar HDTVChallenged 11:11 AM 06-12-2014
Quote:
Originally Posted by zoyd View Post
The RGB bars should also be more usable for RGB based CMS systems as they have been changed to actual coordinate differences instead of white balance values. Normal white balance indicators are maintained when you are measuring a gray/white patches.
Humm ... I guess I'll have to wait and see how well the HSL bars work. But I think the white point reference with the RGB bars was one of the reasons they were so effective for setting "Tint/Hue."

PS: I haven't even had a chance to run 3.1.2 through all the paces yet. Gettin' dizzy here. ... I know ... nag, nag, nag.
zoyd's Avatar zoyd 11:40 AM 06-12-2014
Quote:
Originally Posted by HDTVChallenged View Post
Humm ... I guess I'll have to wait and see how well the HSL bars work. But I think the white point reference with the RGB bars was one of the reasons they were so effective for setting "Tint/Hue."

PS: I haven't even had a chance to run 3.1.2 through all the paces yet. Gettin' dizzy here. ... I know ... nag, nag, nag.
It should be straight forward, set the right hue angle first (H=100%), then get the point in plane by setting luminance (V=100%) and save saturation for last.
HDTVChallenged's Avatar HDTVChallenged 11:51 AM 06-12-2014
Quote:
Originally Posted by zoyd View Post
It should be straight forward, set the right hue angle first (H=100%), then get the point in plane by setting luminance (V=100%) and save saturation for last.
Yeah ... you're probably on the right track with the way the UI *should* work. And the new RGB treatment is probably better for the poor souls that don't have "proper" HSL CMS controls. As always, the devil's in the details.
xvfx's Avatar xvfx 02:37 PM 06-12-2014
Will there be an option for Mascior's disc for the saturation sweeps to get rid of 0%?
Synnove's Avatar Synnove 01:45 AM 06-13-2014
Quote:
Originally Posted by zoyd View Post
Enter those measurements in a synthetic probe page and check the reference box. Then you need the corresponding measurements for those patches from the probe you want to profile (either entered manually in the primaries page or measured). Then select advanced->meter correction file->create from existing measurements.
Thank you for your assistance, the instructions you described worked great! The only oddity was a 1DE difference in the blue primary (the rest was spot on after the corrections).
zoyd's Avatar zoyd 03:43 AM 06-13-2014
HCFR 3.1.3 is back up, let me know how it goes.


Quote:
Originally Posted by xvfx View Post
Will there be an option for Mascior's disc for the saturation sweeps to get rid of 0%?
no

Quote:
Originally Posted by Synnove View Post
Thank you for your assistance, the instructions you described worked great! The only oddity was a 1DE difference in the blue primary (the rest was spot on after the corrections).
You didn't give any details on how the CS2000 measurements were obtained. I hope they are from your display using the same pattern generator, otherwise they are of no use for profiling.
Synnove's Avatar Synnove 03:49 AM 06-13-2014
Quote:
Originally Posted by zoyd View Post
You didn't give any details on how the CS2000 measurements were obtained. I hope they are from your display using the same pattern generator, otherwise they are of no use for profiling.
They were played via Davinci Resolve/Decklink using the same slides/display settings.
First ... 123  124  125 126  127  ... Last

Up
Mobile  Desktop