Is this blue chroma bleeding?

[hermetic_charm]

Every time a strong blue shows up in this video it seems to spill out a bit to the left. Screenshots below plus clips attached of raw capture and post edit.

I am new to this, but I think this is color bleeding? Although I have also seen some posts about edges/ghost/halo can present like this sometimes too.

Is this something I could correct with ChromaShift?

Screenshot from raw capture:
blue_bleed1.png

Screenshot after some CC and filtering and encode to mp4:

You must be logged in to view this content; either login or register for the forum. The attached screen shots, before/after images, photos and graphics are created/posted for the benefit of site members. And you are invited to join our digital media community.

[Aya_Rei]

Looks like chroma bleed to me, try Chromashift in order to shift the U (Blue/Yellow) by 4 pixels to the right or so.

Edit - Quickly did it myself

blue_bleed_correction.png

[lordsmurf]

It's not just blue. You only notice blue most. Look again, look carefully. For example, girl's right arm (left-most in image).

Shift right by 2-4 pixels, maybe up to 2 pixels.

ChromaShift() in Avisynth x86/32-bit
ChromaShiftSP() in 64

http://avisynth.nl/index.php/ChromaShift
http://avisynth.nl/index.php/ChromaShiftSP

[hermetic_charm]

Thank you both! I played with this for a little bit this morning but need to leave the house now.

LS - the girl's right arm, is it the colors from the couch? There is a little red, but the flower pattern brownish sticks out to me in the same place on her arm.

Shift 2 helped but not enough, 4 mostly did the job but the issue was then I started noticing red color bleed in the opposite direction (e.g. bleeding to the right of objects, needs to be shifted left).

I will take a look again when I get back home and try to add some screenshots/clips.

[aramkolt]

Of note, certain TBCs can let you set a YC delay so that this is addressed at the time of capture. It's a little hard to tell sometimes as often it seems that the a actual width of the chroma information is either wider or narrower than the rest of the Luma data, so while you can get it aligned in one area of the image, it may then go out of alignment elsewhere, or when doing that you might find that one edge of the image lacks color because the chroma information was shifted. It's also possible to have chroma data shifted vertically, though TBC adjustments for that are much less common, though not completely unheard of. Chroma shifts seem more common happen on non-first-gen tapes or low production budget commercial tapes that have had dubbing back and forth for editing purposes where no one was really watching the chroma alignment very closely, or their hardware may just not have had adjustments for it even if they noticed it.

Yours are probably first generation home movies though from the looks of it.

You'll know for sure you have a YC delay issue if say the blue part of the shirt was extending into the left, but then the right side of the shirt lacked blue color by the same width assuming the border to their right was either mostly white or black and lacked any blue itself.

[hermetic_charm]

aramkolt, good information to know. I do see a white/colorless line along the right side of the blue shirt, I can't tell if it's a result of YC delay or just a highlight in the image though. While ChromaShift seems to correct the blue bleeding to the left it does not seem to correct the white line on the right. I haven't made any TBC adjustments myself, in this chain there is a line TBC in the DCR-TRV530, that goes to frame TBC tbc-1000.

Returning to this morning and the issue with the red shifting to the right after using ChromaShift, I am attaching some new clips and screenshots to show this. You can see below, the ChromaShift(C=4,L=0) does a good job with the blue on the left side of his shirt, L=2 actually caused more bleeding at the top right side of the shirt, so I decided to not do any shift in the vertical.

At the same time though, ChromaShift brings out some reds on the right side of the red shirt. It does look like there may have been some red bleed to the right even BEFORE the ChromaShift, but the ChromaShift makes it more pronounced. This, and aramkolt's comment above, got me thinking maybe the red is not actually bleed but some other edge artifact? And the ChromaShift is just making it worse because as it shifts everything to the right it actually makes that red artifact bigger. Either that or chroma can bleed in more than one direction?

To try and keep the blue shift but set the red back, I also tried doing something like:

ChromaShift(C=4,L=0)
ConvertToRGB32()
ChromaShift(R=-4)

And a few variations, but this just made the image look ridiculous like a 3D movie or something.

Raw capture:
raw_capture.png

Raw capture plus chromashift:

You must be logged in to view this content; either login or register for the forum. The attached screen shots, before/after images, photos and graphics are created/posted for the benefit of site members. And you are invited to join our digital media community.

CC and filters:
raw_capture_plus_chromashift.png

CC and filters plus chromashift:

You must be logged in to view this content; either login or register for the forum. The attached screen shots, before/after images, photos and graphics are created/posted for the benefit of site members. And you are invited to join our digital media community.

[timtape]

Most of the color in analog consumer video formats had way less horizontal resolution that the B/W picture. So wouldnt we expect the edges of the color information to be blurry and extend beyond the limits of the B/W image on both sides of an object?