mmColorTarget v3.0
This location is for Registered Users Only.
Perhaps you need to login or register.
13.1, 12.2, 12.1, 12.0, 11.3, 11.2, 11.1, 11.0, 10 or later
Linux, Mac, Windows


Useful Nukepedia Tool #8. Explore this video:
Usage:
- Make sure you have Numpy installed, compatible and accessible with your Nuke version. (see Blogpost with additional Information )
- Plug two images showing a MacBeth Colorchecker into the source and target inputs.
- Corner Pin the source and target sampler to its corresponding ColorChecker and adjust the sampler Size
- (Switch inputs using the 'currentView' dropdown)
- Hit the 'Calculate Matrix' Button to get the output ColorMatrix.
Change Log:
v3.0
- BREAKING: Using colorspace transforms to retrieve the reference values instead of the hardcoded patch values. This allows easier addition of future colorcharts by only adding their sRGB reference values. This change will deprecate the use of the mmColorTarget_colorspaces.json file and will start using mmColorTarget_colorspace_transforms.json instead.
- Support for colorcharts:
- ColorChecker24 - Before November 2014 (ColorChecker2005)
- ColorChecker24 - After November 2014
- SpyderCHECKR 24
- ColorChecker Passport Video
- Python 3 support
- Added normalization option
v2.0:
- More intutive UI with callbacks
- Fixed a problem introduced in Nuke 9.0v8 that broke the entire functionality
- The reference target now comes with a dynamic colorspace selection instead of fixed sRGB Values. (generously provided by Thomas Mansencal at colour-science.org)
- Additional sampling method 'no clip' (EXPERIMENTAL!)
- More stability due to no more expression-based-sampling in the background
Tested successfully on Windows 7/10 (Nuke 7-9) and CentOS 6/7/8 (Nuke 7-13)
Mac Users: The Gizmo should also work on Mac, but you need to find/compile a compatible version of numpy yourself, since i don't have a Mac. Check the UNT Video for an example installation.
Comments
Thanks
Daniel
the other way would be to add the following to your init.py file:
import sys
sys.path.append("/Library/Pytho n/2.7/site-pack ages")
It's still possible that you'll get an error if the compiler version doesn't match, which is the point where i can't help any further, since I didn't get enough feedback from Mac users yet. But it should at least find the module now and tell you what's going on.
I hope it helps, and please share your information if you suceeed.
Marco
I've found the NumPy installer (1.6.1) for Python 2.7 for Mac. I'm running a mac osx 10.6.3 with NukeX 8.0.3.
http://stackoverflow.com/questions/7338051/install-numpy-on-mac-os-x-lion-10-7
It works fine with osx 10.6
I've checked in the OSx command line if the package got installed and yes, it worked and found all the files
I found the folders inside "/Library/Pytho n/2.7/site-pack ages" and copied to: .nuke/Framework s/Python.framew ork/Versions/2, 7/lib/python2.7 /site-packages" I'v copied there the numpy folder.
When I load the gizmo, it works well but when I try to calculate the matrix there it appears an error: " .../site-packag es/numpy/core/m ultiarray.so: no matching architecture in universal wrapper "
I've tried importing with the init.py command but it does not work.
import sys
sys.path.append("/Library/Pytho n/2.7/site-pack ages")
Is there any way I can fix it??or I'm doing something wrong..?
I would appreciate any help.
Thanks Marco and amazing gizmo!
- Guillem R.
You might wanna check any of those solutions:
http://stackoverflow.com/questions/4527338/how-to-install-numpy-and-scipy-on-os-x
Have you already tried any of the official installers for osx10.6? For example:
http://sourceforge.net/projects/numpy/files/NumPy/1.7.2/numpy-1.7.2-py2.7-python.org-macosx10.6.dmg/download
Providing google help is really the best i can do right now :(
Very thanks Marco! I've been searching for this for a long time!
I'm the first Mac user to get this working??
I've just added the new site-packages created on the /Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages
Never underestimate google help
Thanks again!!
Well you're not necessarily the first Mac user ever to get it working, but you sure are the first to give me some positive feedback.
So thanks a lot for sharing!
Could you give us a tip for work with spydercheckr 24 without hide 2 rows ( it are inverted in this color card)
Thank you!!
1. It does not manage levels well - I would look at adding an exposure, levels, or some other way of balancing brightness before applying the chart. I fixed that by manually adjusting levels first.
2. It crashes in Nuke 9.0.3. Numpy works in 9 OK - I was able to test it - so something else isn't working.
Thanks again!
I was running into the same need - I created a new version of the tool that allows chrominance calibration without affecting luminance if you're curious to try it out:
https://gist.github.com/jedypod/798b365ea64e8121999e7036ae7e0217
Quote:
I found this morning the new video. Thanks for making it. I followed along on OS X 10.11.5 with Nuke 10.0v1 Non-Commercial.
I still can't make it work.
The gizmo is coming with a .json file, is that also belonging in the gizmos folder?
When I hit "calculate" I get the following error:
'NoneType' object has no attribute 'knob'
Did I install it okay? Is there a way to check inside of Nuke?
And what does this error message means?
Thanks in advance,
Best regards
Daniel (toodee)
you are right, in the normal licensed version it works fine.
Thanks.
It would be nice that the Foundry would actually give the user some hints, that a restriction causing an error in the NON-COM. version.
Current problem is:
'NoneType' object has no attribute 'knob'.
It is a Multivariate Linear Regression: https://en.wikipedia.org/wiki/General_linear_model and uses numpy.linalg.ls tsq: https://docs.scipy.org/doc/numpy/reference/generated/numpy.linalg.lstsq.html
One thing that seems to have always some discrepancy when comparing different documentations are the RGB values on the Macbeth chart. For example, here on Nukepdia the values used on mmColorTarget tool are different to the ones of V_ColorRenditio nChart... Also different documents seems to have different information on the normalized values of RGB for the same illuminant:
http://www.babelcolor.com/index_htm_files/RGB%20Coordinates%20of%20the%20Macbeth%20ColorChecker.pdf
https://www.cs.utah.edu/~aek/classes/cs7650/part03.html
http://xritephoto.com/ph_product_overview.aspx?ID=824&Action=Support&SupportID=5159
Can anyone expand on this differences and how / where these values of this tool were obtained?
Absolutely :) As explained in the description they were generated by myself, the file that yielded the value is there: https://github.com/colour-science/colour-nuke/blob/master/colour_nuke/notebooks/mmColorTarget_dependencies.ipynb
They are based on XRite ColorChecker 2005 values as described here: https://github.com/colour-science/colour/blob/cdbffd063b0c44bb32d752b01647137871434851/colour/characterisation/dataset/colour_checkers/chromaticity_coordinates.py#L114-L114
Now colour rendition charts are never produced equal, so it is normal to find discrepancies between them, not only that but there are also differences in the various RGB colourspaces / whitepoints values depending where you source your data.
This reminds me that I should regenerate the Notebook/JSON file with the latest values content from Colour.
Have it working in Nuke 10.5 in Windows 7 at work, I got the numpy build from the link in Marco's blog above
It drops in just fine and functions well. When I save and open the script again i get a slew of "no such knob" errors and the property window looks like an empty node. Any ideas?
Thanks,
Michael
Please do you know what could be the problem?
thank you for this very useful tool.
Is there any way to keep the lumincance untouched? Often I would like to just get it neutrlized without the patches hitting absoloute value (like .18 for middle grey).
Usualy I am going with the "no clip" choice instead of direct, because the direct mode tend to do funky stuff to overbrights. Is this recommended?
Thanks,
Tobi
thank you for this very useful tool.
Is there any way to keep the lumincance untouched? Often I would like to just get it neutrlized without the patches hitting absoloute value (like .18 for middle grey).
Usualy I am going with the "no clip" choice instead of direct, because the direct mode tend to do funky stuff to overbrights. Is this recommended?
Thanks,
Tobi
I've been struggle trying to make it work on windows.
Theres is a step by step installation for this OS?
I just installed gizmo, phyton, numpy; modify init.py
I can bring the node but can't calculate matrix because it says: This gizmo requires the Python module 'NumPy'.Please install it first and make sure Nuke can import it.
But it is installed.
Any help please?
Thanks before hand.
into your ".nuke" directory?
Thanks sir!!!
Finally I stumbled with a very useful post that helped me to make it work in a easy way
Quoting dapoer vfx:
This is a trick:
1.Install las Houdini Version
2. Go to: C:\Program Files\Side Effects Software\Houdin i 17.5.293\python27\lib\site-packages
3.Copy all content to: C:\Program Files\Nuke10.0v3\pythonextensions\site-packages
4. WhOA LA!!
Thanks to:
https://jurajtomori.wordpress.com/2018/11/28/nuke-tip-installing-numpy-for-nuke-python-on-windows/
Hope it helps!
Same situation here..! Tried on Win 10. I hope someone else here can give solution. Thanks
WHOOO...HOOOOOO ..Thank a tons ..Obrigadooo... Rodrigo !! ...Works like a champ !!
It was long overdue but I have updated the JSON file to use the "ColorChecker24 - After November 2014" formulation: https://github.com/colour-science/colour-nuke/blob/master/colour_nuke/resources/mmColorTarget_colorspaces.json and the notebook to generate those is here: https://github.com/colour-science/colour-nuke/blob/master/colour_nuke/notebooks/mmColorTarget_dependencies.ipynb
Cheers,
Thomas
It's great to see that you are keeping this project alive! It's great, thanks again! :)
Now to a question. We've run into plates shot with the "new" ColorChecker Passport Video edition, that obviously doesn't work with the current version. Any plans on making it compatible with this color checker version as well since it seems to be used more and more?
Cheers,
David
You are the second person this week to ask me about the ColorChecker Passport Video! So I haven't spent too long looking for them, but I could not find the colour specification for the samples.
That being said, one issue I see with that particular model is that it is extremely biased toward grays and skin tones which will be detrimental to all other colours if you were to do a 3x3 matrix-based colour correction.
Technically, when captured under uncontrolled lighting conditions, colour checkers should only be used to perform white balance, the reference colours are assuming a particular illuminant, e.g. D65 and if you were correcting a shot lit with tungsten, you would most likely do worse than good.
Cheers,
Thomas
We have received the full colour specifications for the ColorChecker Passport Video directly from X-Rite. Would you be interested in having a look at them, and see what you could do with them at all? Or is that not interesting to you because of what you stated above?
Cheers, David
I don't have any objections to generate the data however what are the usage conditions for that X-Rite data?
We should probably continue that convo via Email anyway :)
[email protected]
Cheers,
Thomas
Hi, I've done everything to make the gizmo work with nuke 12.0 v4
I have downloaded numpy from this link
https://www.marcomeyer-vfx.de/posts/2014-04-11-mmcolortarget-nuke-gizmo/
and I copied it C: \ Program Files \ Nuke12.0v4 \ pythonextension s \ site-packages
but at the time of calculating matrix
I get a message that says
'NoneType' object has no attribute 'knob'
Hi, I've done everything to make the gizmo work with nuke 12.0 v4
I have downloaded numpy from this link
https://www.marcomeyer-vfx.de/posts/2014-04-11-mmcolortarget-nuke-gizmo/
and I copied it C: \ Program Files \ Nuke12.0v4 \ pythonextension s \ site-packages
but at the time of calculating matrix
I get a message that says
'NoneType' object has no attribute 'knob'
Thanks to some help from @Thomas, I have found that the version of Numpy that ships with Houdini is compatible with Nuke 12.2v4 (macOS). Previously the Apple shipped Numpy worked fine with Nuke 11.
If anyone could just say the complete workflow to install this good script correctly, it would be awesome!
Thanks a lot.
I think you need to have "numpy" library in your .nuke folder.
I can't have correct values in ACES workflow.
It contains some additional features so it requires some more testing, so please bear with me ;)
RSS feed for comments to this post