Artifacts on Blu-Ray that resemble low bitrate blocks and or banding?

[killerteengohan]

I have a Blu-Ray I am trying to fix up but it has got to be the worst looking Blu-Ray I have worked with yet.

Not only is it very soft/blurry looking but it seems to have alot of artifacts that resemeble low bitrates and or banding on DVD's.

Is this something I can clean up without adding grain to cover it up? Whats up with this? I didnt think a Blu-Ray would allow its video to look this poor unless it was an old upscale but this is a newer series, not an upscaled older one.


Here is a cropped screenshot of the untouched source. You cant miss it if you look right in the guys hair in the middle.

https://forum.videohelp.com/images/imgfiles/UsGWWVS.png

[killerteengohan]

Here are some full size untouched shots that may also help.


https://forum.videohelp.com/images/imgfiles/IbxjYS0.png

https://forum.videohelp.com/images/imgfiles/yzvGLFn.png

https://forum.videohelp.com/images/imgfiles/tTpX1Sc.png

https://forum.videohelp.com/images/imgfiles/0C271sm.png

[jagabo]

Since it appears to be worst in the dark parts of the picture try something like:

src=last
ColorYUV(gain_y=-128, cont_u=-128, cont_v=-128)
GradFun2dbMod(str=0, strc=0)
ColorYUV(gain_y=256, cont_u=256, cont_v=256)
GradFun2dbMod(str=0, strc=0)
Overlay(last, src, mask=src.ColorYUV(cont_y=200))

The net result is some grain added though. Play around with the mask to tune it to your liking. You can also follow up with noise reduction but some of the banding will come back.

[killerteengohan]

Okay well gradfun2dbmod is a dithering/debanding filter if I'm not mistaken.

Whats the ColorYUV thats being used twice for? is that the mask your referring to? Is this adjusting colors?

[jagabo]

Yes, gradfun2dbmod is a debanding/dithering filter. It doesn't work well when the steps between bands are too large*. So I used ColorYUV to decrease the brightness by a factor of two, deband, then double the brightness, and deband again. That results in a video with good debanding and some dithering. Finally, I built a mask based on brightness and use that to overlay the debanded image with the original image. Where the frame was dark the debanded image is displayed, where the image is bright the original image is retained, in between is a blend of the two. When I say you should play with the mask I mean you should select for yourself at what intensities and how steep the boundary between debanded and original image is. Also try leaving out the U and V arguments in the ColorYUV calls. They cause a slight shift in some of the colors, especially those with low saturation. See which you like better.

I only tried the sequence on the first of your images. I don't know if it works well with the rest.


* Try replacing the sequence:

ColorYUV(gain_y=-128, cont_u=-128, cont_v=-128)
GradFun2dbMod(str=0, strc=0)
ColorYUV(gain_y=256, cont_u=256, cont_v=256)
GradFun2dbMod(str=0, strc=0)

with a single call to GradFun2dbMod(). You'll see it doesn't work nearly as well.

[killerteengohan]

Thanks Jagabo!! I originally asked for and hoped for a way without adding grain, but this seems to look acceptable.

I appreciate it!