VHS PAL Capture - Very minor pink noise appearing on sports capture

[Robertp]

Hi

I am wondering if anyone on here could help me in regards to some chrome noise.
I have been capturing old VHS tapes from the 1980's using a VideoMate DVB-T300 analogue capture card.
I am capturing via Virtualdub and using Huffyuv as the compression format.

Now my issue is very minor but because it is sport, it is much more noticeable when playing a clip opposed to still frame of the issue.


[Attachment 47111 - Click to enlarge]

As you can see, there is pink noise appearing in small sections of the captured video.
I am trying to use avisynth to correct this but nothing so far has worked.

Can anyone help me?

[jagabo]

Pink noise? You mean near the edges of the frame? On/around the players?

[Robertp]

Pink noise? You mean near the edges of the frame? On/around the players?

On/around the players.
On the shoulders you can see some pink. It's almost unnoticeable but it's definitely there.

[Robertp]

the player in the background also has a tiny patch on his leg

[jagabo]

I recommend you post a huffyuv video sample. But basically, VHS has very low color resolution -- about 40 lines across the entire screen. So colors get smeared quite a lot. It's possible to sharpen and shift them for better alignment. I usually separate fields, downscale horizontally, sharpen horizontally, upscale back to full width, and weave. Merge that chroma with the original luma.

AviSource("filename.avi")
luma = last

SeparateFields()
BilinearResize(288,height)
aWarpSharp(depth=10)
Sharpen(0.5, 0.0)
nnedi3_rpow2(2, cshift="Spline36Resize", fwidth=720, fheight=height) # or needi3_rpow2()
ChromaShift(c=-2) # whatever works
Weave()
MergeChroma(luma, last)

[Robertp]

thanks so much for helping me out jagabo. Much appreciated.

Here's a small video sample (51MB) https://drive.google.com/open?id=1jBuK0sSePVqpgxI6FluPind5iMn3EpA3

[jagabo]

The chroma in your sample is fairly sharp. It may still benefit from sharpening but there's a bigger problem that needs to be addressed first: The chroma of the two fields of each has been largely blended together. It was also captured as RGB instead of YUV. You need to figure out where the chroma blending is occurring and fix it. It could be a strong temporal denoiser, a bad YV12 to RGB conversion, or something else. Try setting your capture device to capture YUY2 and see if that helps.

[Robertp]

ok, I have a feeling you may be right about the capture settings.
I'll change the settings to YUY2 capture and let you know if it makes a difference.

[Robertp]

After trying different settings, it appears that by switching to YUY2 and then changing the hue setting in virtualdub capture, there is barely any pink appearing on the videos. It could possibly be the VHS tapes as I have noticed this issue on videos I have received from other people as well.

[jagabo]

VHS has very low chroma resolution so colors are always smeared horizontally. You had an additional problem where the color of the two fields was blended together:


[Attachment 47133 - Click to enlarge]

On the left is a crop of two consecutive fields from the original video. To the right of that are the two corresponding chroma channels. You can see the motion of the ball in the left image. But there there is very little motion in the chroma channels because they have been blended together. You can see evidence of this in the left image -- the brown color of the ball appears in the grass, ahead of the ball in one frame, behind the ball in the other. Ie, the chroma isn't moving with the luma.

There are also horizontal stripes in the chroma. I believe these are Hanovor bars -- after incorrect conversion to RGB.

https://en.wikipedia.org/wiki/Hanover_bars

[Robertp]

VHS has very low chroma resolution so colors are always smeared horizontally. You had an additional problem where the color of the two fields was blended together:


[Attachment 47133 - Click to enlarge]

On the left is a crop of two consecutive fields from the original video. To the right of that are the two corresponding chroma channels. You can see the motion of the ball in the left image. But there there is very little motion in the chroma channels because they have been blended together. You can see evidence of this in the left image -- the brown color of the ball appears in the grass, ahead of the ball in one frame, behind the ball in the other. Ie, the chroma isn't moving with the luma.

There are also horizontal stripes in the chroma. I believe these are Hanovor bars -- after incorrect conversion to RGB.

https://en.wikipedia.org/wiki/Hanover_bars

Would the only way to fix this issue be by buying a certain model VHS player?
Or is this how the original source was recorded on the VHS tape and therefore, nothing can be done?

[jagabo]

The problem is in the capture device/driver/software, not the VHS deck or tape.

[Robertp]

Ok.
Well thank you for helping out jagabo.
My computer is need of an upgrade as it's going on 15 years old now.