Pages: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
I released the new ProPainter version
v.1.2.4
Dan
P.S.
If you look in more detail your file "mask.png" (by opening it with PIL and converting it to np.ndarray) you will discover that mask_np.min()=1. So in your file the black is represented by the number 1 and not 0 as should be (while mask_np.max()=255 is correct). This anomaly prevented the filter from correctly recognizing the mask hedges. The new added option will fix also this issue.
Nice, thanks!
Will bundle a new torch-add-on and a new dev and upload them. => both are up
Cu Selur
Hello Selur,
please find attached the new RC version of ProPainter v1.2.5.
Main changes:
1) all clip formats are supported (*) no need of RGB24 conversion before calling the script.
2) the filter misc. SCDetect () is automatically loaded, no need of loading code inside the script.
Dan
(*) in case of high big depth formats they are converted in YUV444PS before the RGB24 conversion to preserve the high bit depth on luma.
Nice.
SCDetect: But the dependency is still there?
Since it seems like miscfilters are likely to be dropped soon, you might want to try
https://pastebin.com/Bn2qBT8d instead. (also
here)
created a
test dev which does support the new version, did two quick tests which worked fine, will test some more tomorrow.
Cu Selur
Hello Selur,
I attached the new RC2, now SCDetect() is implemented using PlaneStats, should be still fast enough.
The misc.SCDetect() plugin is not more necessary.
Dan
Will test after work and report back
Cu Selur
Did some testing and everything seems to work fine.
Cu Selur
Good, released the new version
v1.2.5
Dan
I'll package and upload a new torch add-on with it and a dev version which is adjusted to it.
Cu Selur
Using:
Code:
from vspropainter import propainter as ProPainter
clip = ProPainter(clip, img_mask_path="C:/Users/Selur/Desktop/mask.png", length=500, mask_dilation=8, mask_enhance=False, mask_region=(640,256,0,200))
i get:
Code:
clip = ProPainter(clip, img_mask_path="C:/Users/Selur/Desktop/mask.png", length=500, mask_dilation=8, mask_enhance=False, mask_region=(640,256,0,200))
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "f:\hybrid\64bit\vapoursynth\Lib\site-packages\vspropainter\__init__.py", line 92, in propainter
return propainter_inpaint(clip, length, clip_mask, img_mask_path, mask_dilation, mask_enhance, neighbor_length,
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "f:\hybrid\64bit\vapoursynth\Lib\site-packages\vspropainter\__init__.py", line 150, in propainter_inpaint
clip, orig_fmt = convert_format_RGB24(clip)
^^^^^^^^^^^^^^^^^^^^^^^^^^
File "f:\hybrid\64bit\vapoursynth\Lib\site-packages\vspropainter\propainter_utils.py", line 261, in convert_format_RGB24
clip = vs.core.resize.Bicubic(clip, format=clip.format.replace(bits_per_sample=8))
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "vapoursynth.pyx", line 1142, in vapoursynth.VideoFormat.replace
File "vapoursynth.pyx", line 2855, in vapoursynth.Core.query_video_format
vapoursynth.Error: Invalid format specified
I first used a 1bit mask and then the 8bit mask I attached,...
Any idea what the issue is?
Cu Selur
Pages: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18