View Full Version : Chromashift Question


echooff
14th February 2005, 19:23
First a little detail info:
I have a vhs tape of different games of my sons’ football season. The person who made it copied a little of each game from different vhs tape onto this tape. The red chroma (which is the most noticeable) from the leg sox bleeds to the left and right heavily. The leg shape is clear but the fuzzy red extends for some distance.

My question:
It is my understanding chroma filters shift the chroma one way or the other. Since mine is bleeding everywhere is this worth trying to correct? Is it even possible?

Wilbert
14th February 2005, 22:21
It is my understanding chroma filters shift the chroma one way or the other.
Yup.

Since mine is bleeding everywhere is this worth trying to correct? Is it even possible?
No, I don't think it is possible to correct it a little bit.

Si
14th February 2005, 23:25
As Wilbert says, it will just move it around.

From personal experience, I find it better to just de-saturate Red.

The colours end up a bit non-natural but IMO the overall result looks better.

regards

Simon

Didée
15th February 2005, 09:55
Often one can achieve good results by applying aWarpSharp with big radius & strength, together with the cm=1 option, and then use just the chroma of that.

Sniplet of an older script:

o=last
warp1=o.mergechroma(o .blur(1.0).unsharpmask(255,1,1).aWarpSharp(depth=20.0, thresh=0.75, blurlevel=3, cm=1))
warp2=o.mergechroma(warp1.blur(1.0).unsharpmask(255,1,1).aWarpSharp(depth=12.0, thresh=0.75, blurlevel=2, cm=1))
# interleave(o,warp1,warp2) # visualization
warp2

This did a pretty good job on the bleeding colors of a problematic VHS capture (however not mine).
There was additional code to fight color artefacts that aWarpSharp was producing on the outmost frame borders - but the above should suffice to give a fair starting point.

echooff
15th February 2005, 18:22
Thank you Didée. I'll give that a try.

scharfis_brain
15th February 2005, 19:03
@Didee: after some more tests I made with that CTI-replacement, I found it being quite godd one large colored objects.

But small objects will completely loose their chroma.
It has gotten killed by this function.

Soulhunter
15th February 2005, 20:24
Originally posted by scharfis_brain
@Didee: after some more tests I made with that CTI-replacement, I found it being quite godd one large colored objects.

But small objects will completely loose their chroma.
It has gotten killed by this function.

So, a script that warps large objects strong but small ones less !?!

I already smell BlindChromaWarp... :D


Bye

Didée
16th February 2005, 09:07
scharfi: the question is, in case of strong color bleeding, do small objects have any "own" chroma at all?

Perhaps one could fiddle something to find edges that are present in a small neighborhood only. But I've no time to look into that currently ... plus I lack appropriate test sources.

scharfis_brain
16th February 2005, 17:19
I need to search for this (or another) video clip which showed the chroma killing when I tried to CTI it.