This forum uses cookies
This forum makes use of cookies to store your login information if you are registered, and your last visit if you are not. Cookies are small text documents stored on your computer; the cookies set by this forum can only be used on this website and pose no security risk. Cookies on this forum also track the specific topics you have read and when you last read them. Please confirm whether you accept or reject these cookies being set.

A cookie will be stored in your browser regardless of choice to prevent you being asked this question again. You will be able to change your cookie settings at any time using the link in the footer.

[HELP] Removing chroma rainbowing
#1
I've tried most of the rainbow and chroma noise reduction filters but none of them seem to remove the rainbowing on a very old U-matic TV master. Any suggestions?


Attached Files Thumbnail(s)
   
Reply
#2
Would need a small video sample to suggest anything, since those artifacts are usually changing over time.
----
Dev versions are in the 'experimental'-folder of my GoogleDrive, which is linked on the download page.
Reply
#3
A few suggestions to start with:
'Temporal Degrain' + 'CAS'
[Image: Basic-VSR-CAS.png]
'MLDegrain' + 'CAS'
[Image: MLDegrain-CAS.png]

Basically any of the DeGrain filters seems to work fine and remove most of the chroma artifacts on default setting, increasing filteirng for chroma will probably do more, or on could try some ml based filters.
BasicVSR++(required torchAddon)
[Image: Basic-VSR.png]
BasicVSR++(required torchAddon) + GrainFactory3
[Image: Basic-VSR-Grainfacotry3.png]

If it's mainly about the chroma noise, try for example HQDN3D and increase the spatial and temporal chroma strength.
Hqdn3d(clip=clip, lum_spac=0.00, chrom_spac=7.00, chrom_tmp=10.50):
[Image: Hqdn3d-clip-clip-lum-spac-0-00-chrom-spa...-10-50.png]
Cu Selur
----
Dev versions are in the 'experimental'-folder of my GoogleDrive, which is linked on the download page.
Reply
#4
Thank you, I will try Smile
Reply
#5
Thanks, HQDN3 works very well. I also need to upscale to 1080p.

What would you suggest?

What I have done so far:

Step 2. De-interlace SD file using QTGMC and convert to square pixels (SD PAL)
Step 2. Remove chroma noise (HQDN3)

Step 3. Upscale to 1440 x 1080p.

I need the very best quality for a possible Blu-ray release. I have tried several upscale options, should this be done at the same time as the chroma de-noise, or as a separate step?
Reply
#6
Quote: I have tried several upscale options, should this be done at the same time as the chroma de-noise, or as a separate step?
I would do all the steps you mentioned (including the deinterlacing) in one script and not create an intermediate file, especially when not using a lossless format to same the intermediate steps in.
But depending on your system and general workflow doing each filtering separately with intermediate files is also okay, as long as lossless formats are used.

About the resizing:
a. before thinking about upscaling I would look into cleaning the clip up a bit by adding MCDegrainSharp, TemporalDegrain or SMDegrain after HQDering and some deblocking (Delock_QED) before it and may be even add some denoising after the degraun.
b. since the clip hasn't much details to begin with using an ml based resizer like RealESRGAN, SRMD, RealSR and similar will probably lead to more trouble than it is worth (assuming you are not doing per Scene resizing; <- not possible in Hybrid). So for resizing I would probably go for NNEDI3 as resizer and try whether it helps to so a stepped resizing.

Cu Selur
----
Dev versions are in the 'experimental'-folder of my GoogleDrive, which is linked on the download page.
Reply
#7
Another question. The resolution of the video is 720 x 576 PAL, or 768 x 576 when converted to square pixels. Media info reports the following:

[Image: Screenshot-2022-03-18-at-16-13-26.png]

What should the correct aspect ratio be when upscaled to 1080p? It's a UK TV programme shot on video, so I would guess it should be 1.33:1, but when converted to square pixels it works out wider than 1.33:1.
Reply
#8
Tons of possibilities influenced by what happened to the video, whether it was cropped or not, what PAR type was used when or converted,....
Can't answer that, but may the the links in the https://forum.selur.net/thread-597.html might help.

Unless the input is wrongly flagged, Hybrid should to it's job when told to convert to suqare pixel.

Cu Selur
----
Dev versions are in the 'experimental'-folder of my GoogleDrive, which is linked on the download page.
Reply
#9
I've just been reading through this thread.


Quote:In an ITU rec.601 PAL video capture (virtually all PAL capture devices follow this standard) the 4:3 image is in a ~702x576 portion of the 720x576 frame. So if you want very accurate upscaling you should crop to 702x576 then upscale to 1440x1080. As was pointed out, if you upscale the full 720x576 frame to 1440x1080 the aspect ratio will be about 2 percent off.

So If I do the crop to 702 pixels, and converto to square pixels, that should be correct?

I just found this. Wow.
Reply
#10
Quote:I just found this. Wow.
Yes, thats one of the links you should read if you want to try to keep it accurate.

It all depends on what really the source is and what happend with the video,... (also not all equipment always produced the aspect ratio as it was specified Wink)
Personally I only deal with newer content where all 720 pixels are active pixels and cropping 18 pixels (-> 18/720 = 0,025 ~2%) doesn seem like a good way to go.
Also not that if your input PAR is ITU, make sure to let Hybrid know it,.... Wink
-> In my book as long as I can't spot a distortion, aspect ratio is fine.

Cu Selur
----
Dev versions are in the 'experimental'-folder of my GoogleDrive, which is linked on the download page.
Reply


Forum Jump:


Users browsing this thread: 2 Guest(s)