Just received the ColorHug 2! Feedback

[François Simond]

Hi everyone!

After ordering it as early as possible I was happy to receive a ColorHug 2 about an hour ago.

This is still pre-production hardware, everything is pre-release as well, it was marked clearly when I ordered it.

Device serial: 002734, Batch 16, Calibrated on 215-04-18 by Richard.

So out of the box, the good:

• The colorimeter powers on and announces itself on USB as: USB HID v1.11 Device [Hughski Ltd. ColorHug2]

Issues encountered so far:

• The USB flash drive provided doesn't boot on any of my computers
• lsusb output is empty: Bus 001 Device 010: ID 273f:1004 <nothing>
• Windows client: firmware updater and CCMX loader, ColorHug Display Analysis all crash quickly after launching and identifying the sensor (tested in a Vbox Windows 7 VM)
• In the software provided on the USB flash drive partition COLORHUG, the Linux directory is empty

Next steps I'll try is to compile http://people.freedesktop.org/~hughsient/releases/colorhug-client-0.2.7.tar.xz

I failed for now due to gusb >= 0.2.2 missing on my Ubuntu 14.04 distro.

I'll continue report on my findings here, this is just a start and I can't wait to start measuring things.

[Giacomo Catenazzi]

Hello,

for lsusb, please submit the data to http://www.linux-usb.org/usb-ids.html

ciao

    cate

--
You received this message because you are subscribed to the Google Groups "colorhug-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to colorhug-user...@googlegroups.com.
Visit this group at http://groups.google.com/group/colorhug-users.
For more options, visit https://groups.google.com/d/optout.

[Richard Hughes]

On 23 April 2015 at 14:25, François Simond <simond....@gmail.com> wrote:
> The USB flash drive provided doesn't boot on any of my computers

Uncool; what happened when you booted? Did you manage to get to the
Fedora logo? Are you using a UEFI secure boot by any chance?

> lsusb output is empty: Bus 001 Device 010: ID 273f:1004 <nothing>

I think that's from usb.ids; you might need to update the package that
provides that. The perils of being an early adopter :)

> Windows client: firmware updater and CCMX loader, ColorHug Display Analysis
> all crash quickly after launching and identifying the sensor (tested in a
> Vbox Windows 7 VM)

I test in a kvm instance with Windows XP and Windows 7 with USB
passthrough. Can you be a bit more specific than crash?

> In the software provided on the USB flash drive partition COLORHUG, the
> Linux directory is empty

I'll check this now, thanks.

> Next steps I'll try is to compile
> http://people.freedesktop.org/~hughsient/releases/colorhug-client-0.2.7.tar.xz
> I failed for now due to gusb >= 0.2.2 missing on my Ubuntu 14.04 distro.

Yes, and you might need a new colord too depending on what they ship.
Would you be able to try Fedora 22 beta LiveCD? That's got most of the
stuff required already installed, and you can easily install
colorhug-client using dnf on the command line.

> I'll continue report on my findings here, this is just a start and I can't
> wait to start measuring things.

Cool, me too. Thanks for the feedback so far.

Richard

[Richard Hughes]

On 23 April 2015 at 14:35, Giacomo Catenazzi <cate...@gmail.com> wrote:
> for lsusb, please submit the data to http://www.linux-usb.org/usb-ids.html

We (Pascal IIRC?) tried doing that several months ago, I assumed I was
stuck in some kind of moderation queue or something.

Richard

[François Simond]

On Thursday, April 23, 2015 at 3:35:37 PM UTC+2, Richard Hughes wrote:

On 23 April 2015 at 14:25, François Simond <simond....@gmail.com> wrote:
> The USB flash drive provided doesn't boot on any of my computers

Uncool; what happened when you booted? Did you manage to get to the
Fedora logo? Are you using a UEFI secure boot by any chance?

On both my laptop (Clevo W170HR) and desktop (Motherboard: Gigabyte H55M-UD2H with regular old BIOS) all it does after selecting the USB flash drive is a carriage return.

Dumping the content to files gives: "x86 boot sector" for the whole block device and "x86 boot sector" as well for the first partition.

Partitions seem alright, sfdisk -d output is:

# partition table of usbkey-1.img

unit: sectors

usbkey-1.img1 : start=     2048, size=  7636992, Id=83, bootable

usbkey-1.img2 : start=  7639040, size=   192512, Id= b

usbkey-1.img3 : start=        0, size=        0, Id= 0                                                                                                  

usbkey-1.img4 : start=        0, size=        0, Id= 0 

 

> lsusb output is empty: Bus 001 Device 010: ID 273f:1004 <nothing>

I think that's from usb.ids; you might need to update the package that
provides that. The perils of being an early adopter :)

Yes and both the kernel and Virtualbox identify it nicely anyway.

 

> Windows client: firmware updater and CCMX loader, ColorHug Display Analysis
> all crash quickly after launching and identifying the sensor (tested in a
> Vbox Windows 7 VM)

I test in a kvm instance with Windows XP and Windows 7 with USB
passthrough. Can you be a bit more specific than crash?

Here's what's given as Problem details, I'm not sure it will help much.
I don't have access to a real windows machine for now, but I could later.

Problem signature:

  Problem Event Name: APPCRASH

  Application Name: colorhug-flash.exe

  Application Version: 1.0.0.0

  Application Timestamp: ff444642

  Fault Module Name: StackHash_0a9e

  Fault Module Version: 0.0.0.0

  Fault Module Timestamp: 00000000

  Exception Code: c0000005

  Exception Offset: 044ad2e9

  OS Version: 6.1.7601.2.1.0.256.1

  Locale ID: 1033

  Additional Information 1: 0a9e

  Additional Information 2: 0a9e372d3b4ad19135b953a78882e789

  Additional Information 3: 0a9e

  Additional Information 4: 0a9e372d3b4ad19135b953a78882e789

Read our privacy statement online:

  http://go.microsoft.com/fwlink/?linkid=104288&clcid=0x0409

If the online privacy statement is not available, please read our privacy statement offline:

  C:\Windows\system32\en-US\erofflps.txt

 

> In the software provided on the USB flash drive partition COLORHUG, the
> Linux directory is empty

I'll check this now, thanks.

> Next steps I'll try is to compile
> http://people.freedesktop.org/~hughsient/releases/colorhug-client-0.2.7.tar.xz
> I failed for now due to gusb >= 0.2.2 missing on my Ubuntu 14.04 distro.

Yes, and you might need a new colord too depending on what they ship.
Would you be able to try Fedora 22 beta LiveCD? That's got most of the
stuff required already installed, and you can easily install
colorhug-client using dnf on the command line.

Fedora 22 beta, alright! I'll install it in a VM for convenience.

 

> I'll continue report on my findings here, this is just a start and I can't
> wait to start measuring things.

Cool, me too. Thanks for the feedback so far.

As soon as I get the client working I'll add support for it in my measurement software using colorhug-cmd and will share a comparison with other sensors :)

 

Richard

[François Simond]

Yes, and you might need a new colord too depending on what they ship.
Would you be able to try Fedora 22 beta LiveCD? That's got most of the
stuff required already installed, and you can easily install
colorhug-client using dnf on the command line.

Fedora 22 beta 3, in a VirtuaBox VM works perfectly for my initial usage with 

colorhug-cmd get-readings-xyz 0

I am surprised to observe that the factory calibration matrix changes the values a lot, I thought that as the new sensor is a XYZ sensor the matrix would almost be an identity matrix.

I wonder why but maybe the answer is in the sensor datasheet already.

May I ask Richard what you are using to determine the factory calibration matrix?

colorhug-cmd get-calibration 0
index: 0
supports LCD: 1
supports LED: 1
supports CRT: 1
supports projector: 1
description: Factory Calibration
( 0.57  -0.07   -0.00   )
( 0.05  0.52    -0.03   )
( -0.10 0.03    0.78    )

So it's a good start, I will now switch to the next step to integrate support in my measurement software.

colorhug-cmd design makes that very easy!

[Richard Hughes]

On 23 April 2015 at 16:43, François Simond <simond....@gmail.com> wrote:
> Fedora 22 beta 3, in a VirtuaBox VM works perfectly for my initial usage
> with
> colorhug-cmd get-readings-xyz 0

Cool.

> I am surprised to observe that the factory calibration matrix changes the
> values a lot, I thought that as the new sensor is a XYZ sensor the matrix
> would almost be an identity matrix.

Yes, I thought that was a little odd too. The matrix is very close to
unity, but not quite.

> May I ask Richard what you are using to determine the factory calibration
> matrix?

I'm using a NEC screen that has primaries superbly close to sRGB and
then using the Four-Color Matrix Method to generate the device matrix
in slot 0.

> colorhug-cmd design makes that very easy!

You can also use C, Java, Python asynchronously using libcolorhug if
that's any use.

Richard

[Neil McGovern]

On Thu, Apr 23, 2015 at 02:35:36PM +0100, Richard Hughes wrote:
> On 23 April 2015 at 14:25, François Simond <simond....@gmail.com> wrote:
> > The USB flash drive provided doesn't boot on any of my computers
>
> Uncool; what happened when you booted? Did you manage to get to the
> Fedora logo? Are you using a UEFI secure boot by any chance?
>

Same here on a X220 with bios booting - all I get is a underscore.

Is there a download image so I can try flashing the USB stick again by
any chance?

Neil

[Richard Hughes]

On 24 April 2015 at 15:34, Neil McGovern <ne...@halon.org.uk> wrote:
> Is there a download image so I can try flashing the USB stick again by
> any chance?

I've uploaded the one with the partitions here:
https://rhughes.fedorapeople.org/temp/ already -- I'll also upload the
LiveCD one too now, so give it a few mins to upload if you want that.

Richard

[michael...@gmail.com]

On Friday, April 24, 2015 at 4:43:34 PM UTC+2, Neil McGovern wrote:

Same here on a X220 with bios booting - all I get is a underscore.

All I get is a blinking underscore, too. I tried various USB settings but no luck so far.

Cheers,
Michael

[michael...@gmail.com]

On Thursday, April 23, 2015 at 3:26:53 PM UTC+2, François Simond wrote:

Hi everyone!

After ordering it as early as possible I was happy to receive a ColorHug 2 about an hour ago [...]

 
I want to share my initial findings as well. For me, going through the steps at http://hughski.com/owner.html worked fine until I finally ran into a problem in the last step, the actual calibration. Using "high quality" mode I got the following message:

Calibration failed! The profile could not be generated. You can remove the calibration device.

I later tried again using the "medium quality" mode, which worked but produced a very red profile.

All testing was done on Fedora 21 so far, as I still cannot get the USB stick to work. I will try Fedora 22 Beta next and see how it works there.

Cheers,
Michael

[michael...@gmail.com]

On Saturday, April 25, 2015 at 4:19:17 PM UTC+2, michael...@gmail.com wrote:

I will try Fedora 22 Beta next and see how it works there.

I had better luck using the Fedora 22 Beta. While gnome-control-center would crash often, I was able to produce profiles in those cases when it did not crash (about 1 in 3 tries). After applying the profile on my Fedora 21 installation I noticed the following, though:

---
The profile has the following problems:
• One or more of the primaries are invalid
---

Any idea what could be wrong?

Cheers,
Michael

[Graeme Gill]

michael...@gmail.com wrote:

> • One or more of the primaries are invalid

> Any idea what could be wrong?

You didn't link to, or attach the profile, so it's hard for anyone to help you.

Graeme Gill.

[michael...@gmail.com]

Right, my mistake. I attached the file now.

Michael

[Graeme Gill]

michael...@gmail.com wrote:

> I attached the file now.

Without trying to analyse it in more detail,
it is corrupt or malformed. iccdump says:
" tag 14 is out of range of the nominated file size".

Tag 14 sig 'DevD' claims to be 34760 bytes in size, while the file
itself is only 3012 bytes long. So first guess is a software bug
in whatever created the profile.

Graeme Gill.

[Richard Hughes]

On 28 April 2015 at 01:03, Graeme Gill <gra...@argyllcms.com> wrote:
> Without trying to analyse it in more detail,
> it is corrupt or malformed. iccdump says:
> " tag 14 is out of range of the nominated file size".

Using the colord tools:

tag 14:
sig 'meta' [0x6d657461]
size 508
type 'dict' [0x64696374]
Dictionary:
License -> CC0
MEASUREMENT_device -> colorhug2
Quality -> high
MAPPING_device_id -> xrandr-Eizo Nanao Corporation-S1910-38547115
DATA_source -> calib
CMF_product -> colord
CMF_binary -> colord-session
CMF_version -> 1.2.9
tag 15:
sig 'dscm' [0x6473636d]
size 40
type 'mluc' [0x6d6c7563]
Text:
en_US: S1910 [28 bytes]

Perhaps iccdump is barfing on the optional dscm tag?

Richard.

[Richard Hughes]

On 26 April 2015 at 11:54, <michael...@gmail.com> wrote:
> The profile has the following problems:
> • One or more of the primaries are invalid

Although the primaries are indeed out of range, it's only by a tiny
amount. I've seen a lot worse for gamut mismatch. The profile looks
mostly sane to me IMHO.

Richard

[Rafael Brune]

I received my ColorHug2 yesterday.

The usb-stick also only shows a blinking cursor - haven't tried flashing a new image yet.

But I did give it a try on Windows 7 with dispcalgui (make sure to have zero install the newest version of it and Argyl). It works like a charm and I could successfully calibrate a very old Dell 2405fpw - no apparent red shift.

Next is getting all tools up and running on my Ubuntu install.

[Florian Höch]

Am 29.04.2015 um 20:47 schrieb Richard Hughes:
> Perhaps iccdump is barfing on the optional dscm tag?

I can verify Graeme's findings with my tools as well. Tag 14 is (should
be) 'DevD'. The tag table seems to be (partly) wrong.

Tag data for tag DevD should begin at offet 2412 with size 34760 (beyond
the file size). If I examine the data beginning at the given offset, it
spans over the contents of 'gXYZ' (the second half of it after the 'XYZ
' signature), bXYZ, rTRC/gTRC/bTRC, 'meta' and 'dscm' tags (but those
tags itself are fine).

Florian.

[Florian Höch]

As a follow-up, after removing 'DevD' and 'CIED' entries the profile
parses fine. Red and blue primaries are slightly outside the CIE 1931
'horseshoe' of real colors though.

Red XYZ 53.3734 28.0885 -1.5076 xy 0.6675 0.3513
Green XYZ 28.1423 62.9384 11.2324 xy 0.2751 0.6152
Blue XYZ 13.6198 8.9731 98.2083 xy 0.1127 0.0743

Florian.

[Graeme Gill]

Florian Höch wrote:

> Red XYZ 53.3734 28.0885 -1.5076 xy 0.6675 0.3513
> Green XYZ 28.1423 62.9384 11.2324 xy 0.2751 0.6152
> Blue XYZ 13.6198 8.9731 98.2083 xy 0.1127 0.0743

Although the ICC spec says that the primary values should not be negative, It's
perfectly possible for a best fit matrix to have primaries with such values, due
to measurement variation or non-additivity in the display behaviour. In itself,
it doesn't lead to anything bad. I don't see a reason to impose such a restriction
myself.

If the system the profiles are being used with is being particularly picky in this
respect and the profile was being created using ArgyllCMS colprof, then this problem
can be avoided using the -R flag.

It is highly undesirable to refuse to accept -ve primary values for input profiles
though, as the best fit primaries for these devices are often imaginary.

Graeme Gill.

[Neil McGovern]

I've tried colorhug-live-20150420.img.gz with gunzip then:
dd if=colorhug-live-20150420.img of=/dev/sdb bs=1M,
and I've got the same blinking cursor it seems.

Neil

[Graeme Gill]

I just tried booting the latest Live CD image 20150420 on
my usual test machine, and get:

Warning Could Not Boot
/dev/disk/by-label/colorhug-i686-201504200851 does not exist
/dev/mapper/live-cd does not exist

So still unable to test the latest ColorHug2 firmware :-(

Graeme Gill.

[Florian Höch]

Am 30.04.2015 um 15:16 schrieb Graeme Gill:
> Warning Could Not Boot
> /dev/disk/by-label/colorhug-i686-201504200851 does not exist
> /dev/mapper/live-cd does not exist
>
> So still unable to test the latest ColorHug2 firmware :-(

Have you tried the Windows tools that Richard linked a while ago?
I flashed to the newest firmware with them. Took two tries because the
first attempt failed and put the ColorHug2 into bootloader mode, second
attempt worked fine.

Florian.

[Graeme Gill]

Florian Höch wrote:

> Have you tried the Windows tools that Richard linked a while ago?

Yes, it failed to work for me, no matter how many times I tried it:

bin/colorhug-cmd.exe flash-firmware colorhug2-2.0.3.bin
usb[04:02]: USB error on device 273f:1004 : Input/output error [-1]

and the ColorHug2 was left flashing red and green.

Graeme.

[Rafael Brune]

Using fedora rawhide I can't get the gnome-color-manager to work.

In dmesg I get segfaults:
[ 1416.601910] gnome-control-c[2988]: segfault at 0 ip 00007ff84f7cb3e0 sp 00007ffe89e0cdc8 error 4 in libgtk-3.so.0.1700.1[7ff84f6b6000+72d000]
[ 1680.297927] gnome-control-c[3194]: segfault at 8 ip 00005618c141c8f7 sp 00007ffdb1fe9ef0 error 4 in gnome-control-center[5618c13b4000+387000]
[ 1729.579255] gnome-control-c[3298]: segfault at 8 ip 0000562d5acc98f7 sp 00007ffe40968760 error 4 in gnome-control-center[562d5ac61000+387000]
[ 1783.599760] gnome-control-c[3365]: segfault at 8 ip 000056451fa9e8f7 sp 00007ffcaba911b0 error 4 in gnome-control-center[56451fa36000+387000]

And in the console when running gnome-control-center and try start a calibration run:
(gnome-control-center:3365): color-cc-panel-WARNING **: failed to start calibrate: failed to get output
(gnome-control-center:3365): color-cc-panel-CRITICAL **: cc_color_calibrate_setup: assertion 'calibrate->priv->device_kind != CD_SENSOR_CAP_UNKNOWN' failed


Trying to run  /usr/libexec/colord-session -v just gives me:
00:06:33    Verbose debugging enabled (on console 1)
00:06:33    CdMain: acquired name: org.freedesktop.ColorHelper

And it will stay there indefinitely.

Regards,
 Rafael



Trying to run

[Thomas Daede]

After an ordeal with the post office, I managed to receive my ColorHug2 as well. However, on Fedora 22, the colorhug-refresh utility is the only program that works, and produces bogus (though consistent) results. This seems to be the bug I'm hitting: https://bugzilla.redhat.com/show_bug.cgi?id=1204282

colorhug-refresh output:
http://a.pomf.se/hawnar.png

(with a spectrophotometer, brightness is 350 nits and color temperature is about 65k)

The provided flash drive does not boot on my machine. Also, the LiveCD mirror is down, so I can't try a newer image.

[john...@gmail.com]

Got my ColorHug2 and finally am able to play with it a bit now that I am on vacation. I had briefly tried to get all the required packages on my native Mint 17 install, but couldn't get system to recognize the CH2 when I plugged it in (did see a red/green flash on plug-in though). Moved onto the live usb, and I have the same comments as everyone else: empty linux dir, when trying to boot flashing underscore, etc. But I flashed it with livecd-colorhug-20150420.iso and was able to boot into the live os no problem. 

CH2 was recognized, but oddly fedora 22 didn't pick up my wireless adapter (lenovo t430s) so at the time didn't check for firmwware updates. I was able to run dispcalGUI though all the way through but my generated profile has a similar "One or more of the primaries are invalid" as Michael. I've attached it for inspection.

Will play more with everything sporadically over the next week and will keep my eye on this thread. Let me know if I can be of any assistance. Thank you Richard and everybody for the product and community!

Cheers,

[thomas.s...@gmail.com]

Hi there,

got my ColorHug 2 today as welland immediately tried it with the Windows tools - which work perfectly on my install of Windows 10 (10074). The live USB didn't boot on any of my computers, it was stuck on a blinking cursor with BIOS emulation or straight up skipped to my regular OS in UEFI mode. I get the exact same behavior if i try to boot other live sticks unless i've written them with Ultra ISO.
I'm downloading the 20150420 image right now, is that the most recent one?

Cheers,
-Tom

(if it matters, the PCs i get the blinking cursor on are fairly similar systems, one has an Asus P8Z77-V LX w/ i5-3570K, the other uses an ASUS P8H77-M w/ i3-2120)

Am Donnerstag, 23. April 2015 15:26:53 UTC+2 schrieb François Simond:

Hi everyone!

After ordering it as early as possible I was happy to receive a ColorHug 2 about an hour ago.

This is still pre-production hardware, everything is pre-release as well, it was marked clearly when I ordered it.

Device serial: 002734, Batch 16, Calibrated on 215-04-18 by Richard.

So out of the box, the good:

• The colorimeter powers on and announces itself on USB as: USB HID v1.11 Device [Hughski Ltd. ColorHug2]

Issues encountered so far:

• The USB flash drive provided doesn't boot on any of my computers

• lsusb output is empty: Bus 001 Device 010: ID 273f:1004 <nothing>

• Windows client: firmware updater and CCMX loader, ColorHug Display Analysis all crash quickly after launching and identifying the sensor (tested in a Vbox Windows 7 VM)

• In the software provided on the USB flash drive partition COLORHUG, the Linux directory is empty

Next steps I'll try is to compile http://people.freedesktop.org/~hughsient/releases/colorhug-client-0.2.7.tar.xz

I failed for now due to gusb >= 0.2.2 missing on my Ubuntu 14.04 distro.