Deoldify Vapoursynth filter

[Selur]

I keep the settings and just change the 'ColorModel' value.

[Dan64]

As explained in the user guide, now the DeOldify version of "Artistic" and "Stable" are merged at 50% with model "Video" to improve the stability of these models.

So now are at least 50% more similar to "Video" model.
The blending using another model will smooth further the differences.

If you want to see the difference is better to selected the single color model.

See attached images.

Dan

[Selur]

Hmm,.. there is a difference between deoldify vs ddcolor, but none inside these categories
deoldify:
https://imgsli.com/NDQ0OTY5
https://imgsli.com/NDQ0OTcx
dd:
https://imgsli.com/NDQ0OTcw
bug if you say this is to be expected that is fine.

Cu Selur

[Dan64]

I can see that there are very little differences, but I was expecting more visible differences, your results are unexpected.

Please try the scripts included in the attached zip archive and let me know if you get the same results of images included.

Dan

[Selur]

Those outputs look different as I would have expected.
tile.TR_ddcolor_comp:

tile.TR_deoldify_comp:


Cu Selur

[Selur]

tile.TR_deoldify_comp, changing:

# clip_col = havc.HAVC_main(clip=clip, Preset="slower", ColorModel=models[i], ColorTemp="none", ColorFix="none", FrameInterp=0, ColorMap="none", ColorTune="none", BlackWhiteTune="none", EnableDeepEx=False)
        clip_col = havc.HAVC_main(clip=clip, Preset="veryslow", ColorModel=models[i], CombMethod="Luma-Masked", VideoTune="verystable", ColorTemp="very high", ColorFix="none", ColorTune="none", BlackWhiteTune="none")

shows a black picture and outputs:

Error on frame 0 request:

Traceback (most recent call last):
File "src/cython/vapoursynth.pyx", line 3222, in vapoursynth.publicFunction
File "src/cython/vapoursynth.pyx", line 3224, in vapoursynth.publicFunction
File "src/cython/vapoursynth.pyx", line 837, in vapoursynth.FuncData.__call__
File "F:\Hybrid\64bit\Vapoursynth\Lib\site-packages\vsdeoldify\colormnet2\__init__.py", line 191, in colormnet_client_color_merge
colorizer.set_ref_frame(img_ref)
File "F:\Hybrid\64bit\Vapoursynth\Lib\site-packages\vsdeoldify\colormnet2\colormnet2_client.py", line 64, in set_ref_frame
self.server.SetRefImage(frame_bytes, frame_propagate)

File "xmlrpc\client.py", line 1122, in __call__
File "xmlrpc\client.py", line 1461, in __request
File "xmlrpc\client.py", line 1166, in request
File "xmlrpc\client.py", line 1178, in single_request
File "xmlrpc\client.py", line 1284, in send_request
File "http\client.py", line 1175, in putrequest
http.client.CannotSendRequest: Request-sent

adjusting tile.TR_ddcolor_comp the same way, also throws an error:

Error on frame 0 request:

Traceback (most recent call last):
File "src/cython/vapoursynth.pyx", line 3222, in vapoursynth.publicFunction
File "src/cython/vapoursynth.pyx", line 3224, in vapoursynth.publicFunction
File "src/cython/vapoursynth.pyx", line 837, in vapoursynth.FuncData.__call__
File "F:\Hybrid\64bit\Vapoursynth\Lib\site-packages\vsdeoldify\colormnet2\__init__.py", line 191, in colormnet_client_color_merge
colorizer.set_ref_frame(img_ref)
File "F:\Hybrid\64bit\Vapoursynth\Lib\site-packages\vsdeoldify\colormnet2\colormnet2_client.py", line 64, in set_ref_frame
self.server.SetRefImage(frame_bytes, frame_propagate)
File "xmlrpc\client.py", line 1122, in __call__
File "xmlrpc\client.py", line 1461, in __request
File "xmlrpc\client.py", line 1166, in request
File "xmlrpc\client.py", line 1179, in single_request
File "http\client.py", line 1420, in getresponse
http.client.ResponseNotReady: Idle

=> will try whether rebooting the system helps => no same output

Cu Selur

[Selur]

using:

clip_col = havc.HAVC_main(clip=clip, Preset="veryslow", ColorModel=models[i], CombMethod="Luma-Masked", VideoTune="verystable", ColorTemp="none", ColorFix="none", ColorTune="none", BlackWhiteTune="none")

with

ColorTemp="none"

instead of

ColorTemp="very high"

the outputs look more similar.

Cu Selur

[Dan64]

The parameter ColorTemp is used to apply a temporal color stabilization. Using ColorTemp='Very High' it is expected that the differences between the different color models is smoothed.  Even using the more crazier color model, with ColorTemp='Very High', the colors will be tamed (stabilized). Try using  ColorTemp='Low', you should see some more difference.

Dan

P.S.
Updated the HAVC User Guide with ColorTemp explanation.

[Selur]

Will try later and report back.

[Selur]

Yes, as expected using low temp stability (or none) does lead to more differences: https://imgsli.com/NDQ1MTgz

Cu Selur