Deoldify Vapoursynth filter

[Dan64]

[code]    :param dd_render_factor:  ddcolor input size equivalent to render_factor, if = 0 will be auto selected
                                (default = 24) [range: 0, 10-64]

"0, 10-64" <- this is ugly

I don't understand why is ugly, it is a standard notation to say that the available values are: 0,10,...,64
see for example the image below


 
Dan

[Selur]

No, not the notation, having a non continues range of values is the ugly part.
(first ugly thing is not starting at 0 for 10-64 and then instead of using 65, which is out of the range, using 0)

Cu Selur

[Dan64]

Ps.: send you a link to a dev version which is adjusted to vsdeoldify-2.0.0_2024-03-12.
[/quote]

I tested your version.

Model 0: it is Ok

Model 1: The values of Saturation/Hue for Deoldify should be disabled. The Weight should not be visible (like Model:0)


The generated script should be:
ddeoldify(clip=clip, sat=[1.0,1.0], hue=[0.0,0.0], dd_method=1, dd_model=1)

Models: 2,3,4

It is missing “a” in “Chrom Merge”


I will move below the Deoldify panel the DDColor settings, and I will left on the left-side only a new dedicated "Merge Settings" panel (in Model 2, should be added the new parameter “alpha”)


Brighteness should be a float and not an int as now



overall the new Deoldify panel is nice  

Dan

[Selur]

I will move below the Deoldify panel the DDColor settings, and I will left on the left-side only a new dedicated "Merge Settings" panel (in Model 2, should be added the new parameter “alpha”)

I have no clue what you want to say with that sentence. (lost in translation)

(in Model 2, should be added the new parameter “alpha”)

def ddeoldify(
    clip: vs.VideoNode, model: int = 0, render_factor: int = 24, sat: list = [1.0,1.0], hue: list = [0.0,0.0],
    dd_model: int = 1, dd_render_factor: int = 24, dd_tweak_luma_bind: list = [False, 0.0, 0.0], dd_bright: float = 0, dd_cont: float = 1, dd_gamma: float = 1.0,
    dd_method: int = 2, dd_method_params: list = [0.5, 0.6, 0.15, 0.2], chroma_resize: bool = True, device_index: int = 0, n_threads: int = 8, dd_num_streams: int = 1,
    torch_hub_dir: str = model_dir
)

no alpha there,....

Model 1: The values of Saturation/Hue for Deoldify should be disabled. The Weight should not be visible (like Model:0)

will fix.

It is missing “a” in “Chrom Merge”

will fix.

I will move below the Deoldify panel the DDColor settings, and I will left on the left-side only a new dedicated "Merge Settings" panel (in Model 2, should be added the new parameter “alpha”)

Merge settings, are a subset of DDColor settings, since the are only available if DDColor is used.

Brighteness should be a float and not an int as now

will fix.

Cu Selur

[Selur]

After some guessing, I uploaded a new dev version (same link).
Going to bed now.

Cu Selur

[Dan64]

[quote pid="21674" dateline="1710358980"]

(in Model 2, should be added the new parameter “alpha”)

[/quote]
Sorry, I added the "alpha" parameter in the new version (here attached).
There is also a new parameter "luma_temp_stab", but is not working as expected, so please don't add in the GUI.

Thanks,
Dan

[Selur]

Will look at it after work.
Does 'alpha' work as intended?

Cu Selur

[Selur]

Updated the dev build (same link), rearranges the ui elements.

Cu Selur

[Dan64]

I like the new GUI.

Thanks,
Dan

[Dan64]

Hello Selur,

   I hope you feel better.
   I updated the ddeoldify filter and added another merge method: Luma Masked Merge
   Probably this is the more effective method to fight against ddcolor psychedelic effect.

   With this method the clips are combined using a luma masked merge, the pixels of ddcolor with luma < luma_mask_limit will be filled with the pixels of deoldify, moreover if the parameter ddcolor_weight > 0, the final clip obtained using the luma mask will be merged again with deoldify. This last merge is necessary to blend the effects of masked merge.

  This is an example of image obtained using Luma Masked Merge,  with luma_mask_limit = 0.6 and ddcolor_weight = 0.4

 

   I made some small change to the parameters in input to ddeoldify.

   Given that "clipb_weight" was used by all the merged methods I removed it from "dd_method_params" list and renamed in "ddcolor_weight".
   I removed "luma_temp_stab" and added "luma_mask_limit" which is used by the new method.  

   Unless are found bugs I'm not planning to add others changes, so I hope that this is the last 2.0 version.

   Thanks again for the nice GUI in Hybrid for this filter.

Dan