AVS Forum banner

HCFR - Open source projector and display calibration software

2M views 16K replies 1K participants last post by  nathan_h 
#1 · (Edited)
Hi and welcome to the open source HCFR software thread.
Updated Jan. 27th - 2018

I love working and tinkering on the program but given the amount of time I've invested over the past year on improvements I'm going to start accepting donations. If you feel the program has been of value to you since I started working on it about 3 years ago please consider a contribution via the link below. Thanks!

https://www.paypal.me/zoyd
(or via direct paypal to vze262tr@verizon.net if paypal.me is not supported in your country)

Thanks to the following for recent donations! [LarryInRI, Barsk, 22point, LastButNotLeast, Dominic Chen, Brian M., Wayne L., Andrej S., Arturs J., Douglas P., John G., mo949, Eric V., David S., BlueChris, EXT64, Nihat S., Luis E., J. F., Miiku Joao F., Luis M., Nate W., Don H., Stahinja M., Jeff H., Pedro C., Howell A., Myat M., Carlos F., Webster D., SiegeX, Jörg O., Enrico B., Gert M., Marlon K., Jacob R., John G., Kari T., Ryan T., Franck P. ]

HCFR uses ArgyllCMS for the operation of the probes, so please help ensure that ArgyllCMS remains available in the future, by making a financial contribution to the argyllcms project:

www.argyllcms.com

---

This thread was originally started by JohnAd who had the nice idea of creating an open source fork of the original HCFR Colormeter software to integrate the existing GUI and all of it's calibration features, with the meter handling code of ArgyllCMS. Since ArgyllCMS is continually adding and improving meter support, HCFR benefits by staying current with these developments. The software is frequently referred to as HCFR or colorHCFR where HCFR is short for "Home Cinema-FR", a French forum where some of it's members originally developed the code to support their home-brew colorimeter in 2005. Along with the meter handling improvements introduced by ArgyllCMS, I have been working on improving and adding features to the color engine and internal pattern generator. Please use this thread for bug reports, usage questions and suggestions for improvement. Anyone wishing to help with code development should join the developers mailing list.

The latest version of HCFR (3.5.1.4) can be found here. Changelog

Help Wanted (pm zoyd)
  • Language translation
  • Update French and German with recent changes
  • Full Spanish, Italian, others? needed
  • Help file updates
  • PDF Quick Guide

User wl1 has provided some useful links for beginners to display calibration as well as links to helpful posts within the thread. Please take a look before posting to see if your question has already been answered. There is also a FAQ section below that may be helpful.

Useful Links
Display Calibration - Part I
Curtpalme - Calibration Guide for Dummies
Basic Guide to Color Calibration using a CMS (updated and enhanced)
ArgyllCMS Probe installation notes
Spears and Munsil
Choosing a Color Space | Spears & Munsil
The Secrets Blu-ray Player HDMI Benchmark Part 1 and Part 2
Original version 2 avsforum discussion thread
Another version 2 discussion thread
How-to: Use HCFR to profile your colorimeter

Links to Tools & Utilities
AVSHD709 Test Disk
GCD Disk
Mascior's Calbration Disc
dispcalGUI (front-end to ArgyllCMS)
Ted's LightSpace CMS Calibration disk (now with pointers for use with HCFR) [Set-up Instructions]

Thread Summary (from Post 792, New Scientist)
Post No. Comment
#151 to #155 - installing i1Display Pro

#159 - use of i1Display 3 3/1/2012 (march 1st)

#161 - removal of non-free code (for accessing meter)

#162 - x-rite driver (.dll) use

#165 - spotread

#167 - spotread

#174 - problems with i3D3

#180 - observations re use of i1D3

#183 - reply to #180

#189,#196 - x-rite drivers

#198 - x-rite drivers

#202 - installation i3D3

#226 - a first attempt at calibration

#227,#236 - i1D3 not requiring calibration

#231 - ccss

#248,#253 - i1D3 not requring drivers from x-rite to work with HCFR

#257,#258 - i3D3 with HCFR

#263 to #270 - analysis of #258

#275 - spreadsheet for setting gamma

#277 - spreadsheet uses

#325 - HCFR v3.0.1.0

#397 - using i1D3

#416 - graphs from testing i1D3

#425 - EDR and i1D3ccss too
l
#426 - v3.0.2.0 issues re i1D3

#428 - generate .ccss files from i3D3 disk

#442 - v3.0.3.0

#487 - v3.0.4.0

#507 - what is .ccss

#509 - refresh and i1D3ccss.exe

#514 - what is .ccss and .EDR

#519 - i1D3ccss usage

#528 - where should the .ccss files go?
#534 - ARGYLL_COLMTER_CAL_SPEC_SET environment variable
#540 - some addtional ccss file info (plasma displays)

#552 - ccmx file

#600 - i1D3 not showing in Sensor list

#610 - "Use measured colors", in Preferences -> General, option

#678 - i1D3 and i1 Display Pro differences

#682 - comment re issue with v3.0.4.0 and i1D3

#688 - which driver to use

#717 - i1D3ccss tool

#732 - do I need drivers or not?

#786 - spectral sample files

#824 – Steps to load Driver

#841 – i1D3 reds on Plasma - matrix

#876 – ColorMunki Display guide

#898 - 904 – Pana Window sizes

#1008 – Comparison of HCFR 2.1 and 3.0.4.0

#1083 – HTPC Cal. Suggestion

#1105 – Updating and removing Argyll Drivers

#1114 – Plasma Essential Reading

#1122 –i1D3Pro - initial accuracy study/background to meter matrices

#1146 – Matric Correction file info

#1150 – Import Matrix file into HCFR

#1205 – 2 point V 10 point Calibration

#1215 – Warming up meter

#1241 – profiling Meters

#1250 – Matrix file for Display LT2

#1258 – errors in HCFR ? Red push?

#1260 – GT30/VT30 Experiences

#1270 – WARNING – HCFR Matrix different than CP or CM

#1280 – Zoyd Steps in - Installation package and baseline Integration value

#1285 and 1299 – The fixes explained – instability @30%IRE

#1303 – Argyll Driver suggestions

#1336 –build from Zoyd – USB fixes

#1337 – Plasma/LCD – definitive Refresh settings and Calibration of meter (>50%)

#1344 – Lots of changes to HCFR

#1352 – List of more changes – BT1886 gamma target

#1354 – More info link on BT1886

#1357 & 1360 & 1370 – Answers about changes – and gamma recommendations

#1380, 1390 – another build

#1398 – Fork of a Fork

#1411 – Luminance and Delta E – Yes or No? Colour or Grey?

#1420 & 1426 – 3.0.4.1 Changes

#1424 – Use of GCD Disk Colour Checker Pattern

#1512 & 1515 & 1558– Version 3.0.4.2 Changes

#1521 – Run HCFR with Admin Privileges

#1523 – Projector questions and suggestions
#1537 – Explanation of Change to HCFR calculation of gamma
#1647 – LCD and LD ON/OFF? Plasma and ABL – guidleines
#1648 – Preferences -> General -> gray levels rounding assumption to unchecked? AVCHD,DVD=checked, GCD,internal patterns=unchecked
#1650 – Can I use matrix from D3 for LT2? No – see 1651

#1663 – BT.1886 or not? Depends on if you have 10 point control

#1702 – BT1888 is not meant to be flat

#1722 – Samsung D & E -75%A and 75%S recommended

#1774 – Explanation of affect of Patterns to gamma and light output results


FAQ:
Do I need to install a driver for the i1Display Pro (i1D3)?
No. This meter as well as the Huey, ColorMunki Display, and ColorHug use the system HID driver.

Which probes are supported?
Any probe that ArgyllCMS supports.

How do I install the right driver for my probe?
Assuming your device is not one of those above, the driver .inf file is located in the drivers sub-directory of the main HCFR folder (by default this is c:\Program Files (x86)\HCFR Calibration\Drivers). You do not need to install the full ArgyllCMS program but useful driver installation hints can be found here. After plugging in your meter go to the system devices in your control panel, find the meter and right click to update the driver. Point the update dialog to the HCFR drivers folder.

Which deltaE formula is the most accurate?
All the deltaE formulas represent a measure of the distance between your measured color and the reference color. So they are just rulers with different scales, use the one recommended if you aren't sure. For gray scale measurements the recommended formula provides more sensitivity to chromaticity (x,y) errors so it's easier to use for "tuning" your controls. The dE94 or dE2000 formulas are generally the most acceptable for "proofing" as they are the most uniform in perceptual space and a unit of error in one part of the gamut has the same visual impact as the rest of the gamut.

How can I import Chromapure probe profile data in HCFR?
Calculate the matrix correction using the spreadsheet from this post where z=1-x-y and then enter it in the sensor matrix tab.


How do I use the Casino Royale images to check my calibration?
There is a nice guide here on what to look for at those particular timecodes.


What do the different parameters for the BT.1886 gamma formula do?
The recommended value (default) for the exponent in the formula is 2.4. This value will be used when you enter 0 for absolute in the BT.1886 box. If you want to tailor the function to make the picture have a little more or less contrast then enter a different number here. If you enter 2.4 for example the actual exponent used will be adjusted so that the target gamma has a value of 2.4 at 50% stimulus. You might use that for a dark environment and 2.2 for a brighter environment.

The other parameter (% Input Offset) will change the shape of the curve near black up through about 20% stimulus. The default value will target a gamma that should provide plenty of detail in the shadows. If you find the shadow detail is being lifted too much when you calibrate to this curve, adjust this value lower (try 50%) and this will raise the target gamma near black and darken the shadows. Note that in most cases you must have multipoint gamma adjustment ability to get a good match to a BT.1886 curve. Also note that if you have a zero black level (e.g. OLED) the target point gamma curve will be flat.


What should I use to generate test patterns?
Most people start out using DVD/Blu-ray test disks to generate their test patterns and run HCFR in manual mode. The assumption being that the BD player is somehow altering the test pattern and not following industry standards for the rendering and transmission of video and that these errors must be "calibrated out". However, this assumption is usually overblown and most often you'll find that while there may be some single bit rounding errors being generated they are not large enough to have any perceptual effect on the signal. I recommend that once you get a good feel for the calibration process using a DVD/BD that you move on to automatic pattern generation (see next question) and then cross-check against your BD player results for sanity.


How do I set-up automatic test patterns?
HCFR has three options for automatic test pattern generation, it's own generator, madTPG, and the Chromecast device. The first two require some set-up of the video card to ensure accuracy while the third requires no set-up. All three options are very accurate test pattern generators with the following characteristics:

HCFR generates all of it's triplet targets in 8-bit video levels (16-235) so when used in a video level path both the internal generator and madTPG will provide technically reference triplets. When used in a PC level (0-255) path the internal generator will expand the video levels to PC level and this will produce rounding errors. While small it is recommended that you use madTPG if you want to calibrate in PC levels because it will dither the output levels and reduce or eliminate any rounding errors. A good summary of the various levels combinations can be found here(h/t @SiegeX).

The Chromecast device produces Y'C'C' output and while not technically a reference device because the levels it produces have rounding errors, it is also dithered to reduce these errors below perceptual limits. The advantage to this device is cost ($30) and ease of set-up (it really is "plug-and-play")

To set-up your laptop/PC for pattern generation (either internal or madTPG) you will need to ensure your video card has linear gain tables loaded (i.e. no profile installed). madTPG does this for you automatically, otherwise you'll need to research your card settings to determine how to do this. It is also easiest to get the correct levels set-up if you force the card to output it's normal 0-255 range when connected to your display over HDMI as some cards will default to 16-235 levels, which in some cases is undesirable. The madVR package has a utility for doing this (madLevelsTweaker) or your card may have this option in it's control panel.

Once you have the card set-up select 16-235 in the HCFR view images settings (or 16-235 in the madVR settings) for a display set to video levels. If the display is set to expect PC levels choose 0-255 in HCFR or the same in madVR. There are no level settings required when using the Chromecast as your pattern generator.

All three pattern generators have settings for window size as percent of screen area and background level as percent average picture level. Which settings you use depends on your particular display so it's best to consult an owners thread for specific recommendations.
 
See less See more
#8,363 ·
Forgive me I'm a noob. So which one of the patterns in the pictures I posted should I use? Thanks
I don't think it matters what is around the patterns as long as the pattern in centers scales from 0% - 100%.
That is the area you are measuring.
 
#8,365 ·
Ok guys 3 more questions and then I'm going to start calibrating:

1. For meter settings I've selected:
Display type plasma
Reading type display
Observer type Default
I need to hit calibrate right? What should be on the screen when I hit calibrate?

2. For setting primary and secondary colors I want to use the 100% color pattern in "ColorHCFR Windows"?

3. With all the debate about 2.2 gamma vs 2.4 gamma I am going to target my gamma line at 2.3 all the way across. Is this a good strategy?
 
#8,366 · (Edited)
Thanks guys. I will use the bottom one (ColorHCFR Windows 10% Greyscale). The reason for my question is because each pattern gives different measurements leading to different settings.
No, you should not use the window patterns for TVs (such as plasmas) which automatically adjust the brightness based on the displayed content. Full screen patterns would be even worse. That's the purpose of the APL patterns - to keep the average brightness the same when the pattern itself varies from black to white. Same with the colours.
 
#8,367 ·
No, you should not use the window patterns for TVs (such as plasmas) which automatically adjust the brightness based on the displayed content. That's the purpose of the APL patterns - to keep the average brightness the same when the pattern itself varies from black to white. Same with the colours.
That's what I thought. Can you tell me if I should use the "small" or the "large" in the pics I posted a little earlier. Thanks.
 
#8,368 ·
That's what I thought. Can you tell me if I should use the "small" or the "large" in the pics I posted a little earlier. Thanks.
I would use the small patterns, although I don't have a plasma TV to test that. You should display a white pattern and take continuous readings, to see which pattern gives you less fluctuations in the luminance reading.
 
#8,369 ·
No, you should not use the window patterns for TVs (such as plasmas) which automatically adjust the brightness based on the displayed content. Full screen patterns would be even worse. That's the purpose of the APL patterns - to keep the average brightness the same when the pattern itself varies from black to white. Same with the colours.
Ah. ok. So that is the purpose of the surrounding "scenery".
 
#8,370 ·
Does "Change White" under References work? I'm trying a Judd Voss white point ( x .307, y .318) as a test, but selecting Custom white does not enable the White text boxes.
It works, but you have to click on the Apply button to enable to text boxes. Or you can click OK to close the dialog box, and then re-open it.
 
#8,371 · (Edited)
How can one tell if they are using their meter correctly or if the meter is off? I'm using HCFR 3.4.2 with a i1Display Pro rev B (brand new). Using AVSHD709 I could get a rock solid greyscale on a Sony XBR43X800D (quasi-10bit 4K TN panel), but the resulting picture clearly had errors throughout with red and green banding in the greyscale.

Using DisplayCAL on Eizo monitors results in a 6500K sRGB 2.2 calibration that looks awfully blue to the eye.

Using HCFR on a Panasonic Plasma ST50 results in readings that just don't seem correct...

Is there any method to determine if a meter is off without the use of another meter? For instance, when calibrating the ST50 I've tried both "plasma" and "refresh". Both don't seem to read accurately. By this I mean if I adjust Red to be 21% of 100% white, it is quite visibly wrong on the flashing color bars. Same with cyan. If I go by the meter, it is clearly wrong "by eye*". This leads me to believe either

a) I'm setting HCFR up incorrectly
b) my meter is "off"

I've seen reference to @zoyd correction matrixes, but it seems like those have been rolled into HCFR at this point under "plasma", no?

Any help? I'd like to know if my meter is off while I am within the period of return.

*using the blue filter from Spears & Munsil v2 Blu-ray

(additionally on the plasma, the smoothest greyscale delta I can achieve (average sub 2.0) results in a picture that is crushing blacks and clipping white! Is is possible my ST50 simply can't achieve 120cd/m2 with a good picture? If I ignore the meter and set Contrast and Brightness where they should be by clipping patterns then my 10% APL 100% white reading is near 80cd/m2.)

This is a difficult learning experience where I am now second-guessing my equipment! Please--someone set me straight! :eek:
 
#8,372 · (Edited)
How can one tell if they are using their meter correctly or if the meter is off? I'm using HCFR 3.4.2 with a i1Display Pro rev B (brand new). Using AVSHD709 I could get a rock solid greyscale on a Sony XBR43X800D (quasi-10bit 4K TN panel), but the resulting picture clearly had errors throughout with red and green banding in the greyscale.
In most cases, even if a meter does not read a display type correctly, it should still read consistently throughout the range. If you see colour banding in the greyscale, you should set HCFR to measure at least 20 steps to make sure there are no discontinuities between the 10 points.

For instance, when calibrating the ST50 I've tried both "plasma" and "refresh". Both don't seem to read accurately. By this I mean if I adjust Red to be 21% of 100% white, it is quite visibly wrong on the flashing color bars. Same with cyan. If I go by the meter, it is clearly wrong "by eye*".
When you calibrate the primary colours, it's not sufficient to just set the level. You need to make sure the x, y, Y values all match the target. You cannot rely on the colour filters.

I've seen reference to @zoyd correction matrixes, but it seems like those have been rolled into HCFR at this point under "plasma", no?
Ideally you should get the correction matrix specific to your TV (ST50). You may be able to find them posted somewhere.

(additionally on the plasma, the smoothest greyscale delta I can achieve (average sub 2.0) results in a picture that is crushing blacks and clipping white!
A smooth greyscale does not provide any information regarding the black level and white level. You need to look at the gamma curves, and the near black and near white graphs.
 
#8,373 · (Edited)
In most cases, even if a meter does not read a display type correctly, it should still read consistently throughout the range. If you see colour banding in the greyscale, you should set HCFR to measure at least 20 steps to make sure there are no discontinuities between the 10 points.

When you calibrate the primary colours, it's not sufficient to just set the level. You need to make sure the x, y, Y values all match the target. You cannot rely on the colour filters.

Ideally you should get the correction matrix specific to your TV (ST50). You may be able to find them posted somewhere.

A smooth greyscale does not provide any information regarding the black level and white level. You need to look at the gamma curves, and the near black and near white graphs.
Thanks for the input. I guess you're saying you think its unlikely the meter is off?

I've sourced what seems to be a matrix for the ST50 and I'll give it a go.
1.0383 -0.015199 -0.012862
0.030022 0.96707 -4.8218e-003 (or -0.0048218)
-4.2114e-004 (or -0.00042114) 6.6268e-003 (or 0.0066268) 0.95594

Will have to further explore primaries. Was following the old Curte Palme guide.

But what does it mean when the Delta E of the primaries decreases, yet looks increasingly wrong on the flashing color bars test?
 
#8,374 · (Edited)
#8,375 ·
Discovered a few things:

I have a P50UT50--who knew?
Using the correction matrix really helped.
Utilizing the service menu I used a 2pt greyscale adjustment, initially based on numbers found online.
They read very red, and behold I could see the red in a ramp at the area it was reporting.

Managed to get a flat(ish) greyscale with all Delta E under 2. Other people report that the UT50 nearly nails Rec709 in Cinema mode. Mine certainly doesn't read that way...
 
#8,376 ·
For TVs with 10-pt grey scale control, "Absolute Y w/ gamma" is the easiest way - all you need to do is to adjust R/G/B controls at each IRE so that the bars show 100%, and both the grey scale and the gamma will be correct. If the TV only has 10-pt gamma control (such as the Epson projectors), it may be easier to adjust grey scale and gamma in separate steps, in which case "Absolute Y w/o gamma" may be a better choice.


Hi Dominic,


I noticed that this selection box is always grayed out for me with the 'w/o gamma' selection checked. Any idea why that may be the case?
 
#8,377 · (Edited)
I noticed that this selection box is always grayed out for me with the 'w/o gamma' selection checked. Any idea why that may be the case?
The "Greyscale dE handling" is not selectable when the Color Difference Formula is Recommended. You have to first select a different Color Difference Formula, then click on Apply, before you can change the Greyscale dE handling option.
 
#8,379 ·
Is there any method to determine if a meter is off without the use of another meter?
It is possible to sanity check an i1d3, but it will only tell you if something is grossly wrong.

You need an incandescent lamp of the classic variety (i.e. not Quartz Halogen), and measure
it in ambient mode (e.g. by using ArgyllCMS "spotread -a"). The measured color temperature
should be close to 2850K, and close to the black body locus.

If I do this with my i1display Pro for instance, I get 2932K CCT and 0.002 Duv.
(2879 Visual color temperature and 3.5 Delta E2K).

With my ColorMunki Display I get 2834K CCT and 0.000 Duv.
(2820 Visial Color Temperature and 0.9 Delta E 2K)
 
#8,380 ·
No, you should not use the window patterns for TVs (such as plasmas) which automatically adjust the brightness based on the displayed content. Full screen patterns would be even worse. That's the purpose of the APL patterns - to keep the average brightness the same when the pattern itself varies from black to white. Same with the colours.
I've finished my calibration on my VT60. I used the small ABL patterns for grey scale and gamma. I also used the small ABL 100% colors pattern to set primary and secondary colors based on what you said above. "Same with the colours". I used the correct pattern for colors or should I have used 100% color windows?
 
Top