Good Frame/Bad Frame adaptation (Chroma faults)

[JohnMyler]

Hello, I have a question regarding computer analysis and distribution between frames. I am curious as to whether any software packages (or filters) can do the following:

Say there was a one minute movie sequence. Of 60 seconds, imagine over half have damaged chroma (for instance patterning or banding). If the breakdown of this sequence was good frames-bad frames-good frames etc, could any software analyse the pixels and fix the chroma fault? For instance if frame *A* (0:00:00:01) was good and frame *B* (0:00:00:02) was bad, obviously over such a short distance the pixels would be near identical in both, could any software therefore recognise which pixel hue and saturation had changed, and change it back?

I hope this makes sense, just very curious - it almost feels too difficult to do but like it logically could be. Any guidence appreciated.

Many thanks
John

[redwudz]

I may be possible with one of the many VirtualDub filters. I've seen a few that may be able to do some of that. Take a look at this site: http://neuron2.net/

For a easier to seach listing, go to 'Other' then click on 'Dmitry Popov's Filter Database'.

Another one to look at there may be 'flaXen VHS Filter 1.0 '.

[JohnMyler]

Hi, thanks for the reply. Could you recommend a filter, I have downloaded some, but none have the desired effect. They seem more geared to mass color changes rather than individual pixels. I may have been a little unclear in my post so I have mocked-up this quick example.
It's only a hyperthetical question, so this is just a photoshop fake.

As you can see I want to know if the software can recognise the same pixels in both and re-correct the hue/sat values of the pixels.

Thank you
John

[JohnMyler]

If anyone has any ideas or even untested theories, would be great to hear them?