Deoldify Vapoursynth filter

[Selur]

Okay, will try to look at it later this evening or tomorrow.

Cu Selur

[Selur]

Had a quick look and a question about:

I updated the filter introducing a "gradient mask" merge.

looking at the code it looks like this is applied as long as dd_method != 0 and thus a dd_color option, is that as intended?

Cu Selur

[Dan64]

Just to provide you an example.

The movie DV+DD_LumaMerge.mp4 was obtained with the following settings

clip = ddeoldify(clip=clip, model=0, render_factor=24, sat=[0.90,0.80], hue=[15.00,15.0], dd_model=1, dd_render_factor=24, dd_tweak_luma_bind=[False, 0.5, 0.5], dd_bright=0.05, dd_cont=1.0, dd_gamma=2.0, dd_method=5, ddcolor_weight = 0.6, luma_mask limit=[0.4, 0.4, 0.3], dd_method_params=[0.6, 1.0, 0.15, 0.15], chroma_resize = True)

The movie DV+DD_LumaMerge+DarkDarkness.mp4  was obtained with the following settings

 

clip = ddeoldify(clip=clip, sat=[0.90,0.80], hue=[15.00,15.0], dd_bright=0.05, dd_cont=1.0, dd_gamma=2.0, dd_method=5, ddcolor_weight = 0.6,  dd_method_params=[0.6, 2.0, 0.15, 0.2, False], luma_mask=[0.4, 0.4, 0.8], dark_darkness=[True, 0.3, 0.3, 0.2, -0.05])

The post-process " dark_darkness " helped to stabilize the colors on dark scenes

Dan

[Selur]

Updated the dev version (same link).
Will look at the examples. Okay, in really dark scenes it helps, now it needs a general stabilization.

Cu Selur

[Dan64]

I really like the new GUI, very nice! 

Thanks,
Dan

[Dan64]

In the last version I shortened the names of dd_tweak_bright, dd_tweak_cont, dd_tweak_gamma in dd_bright, dd_cont, dd_gamma.
I decided to revert the changes, so I attached a new updated with the names equals to the ones implemented in Hybrid.

I checked all the parameters in the GUI. It is only missing a new boolean parameter "dd_method_params[4]" called "invert_clips" if true is inverted the clip order in all the Merge methods (i.e. clipa-> clipb and clipb->clipa). 

Dan

P.S.
here you can find a comparison of LumaMerge using a gradient mask (0.4,0.7)
https://imgsli.com/MjQ3NjIy

[Selur]

I'll look at it.

[Selur]

Updated download (same link).

Cu Selur

[Dan64]

In the last version, the script is not including any more the string: "dd_method_params: list = [0.6, 2.0, 0.15, 0.2, False]" independently from the fact that "invert merger order" is checked or not.

Dan

[Selur]

updated the download