Fluorescent banding in RED footage. . .

[Stephen Pruitt]

Hey guys. . .

. . . in looking over some footage of a location we cannot revisit, we see some banding from old fluorescent ballasts (we used our new Kino bulbs) in the footage. Is there any post process that will take this problem out of our precious footie?

We just didn't see it on the monitor while we were shooting. I don't know why, but we just didn't see it. It's definitely there. . . faint flow bands rising up in the shot.

I can't believe it.

:-(

Thanks.

Stephen

 

[ericyoung]

Hate to say it, but it's going to be a monster job to even try to tackle it without a reshoot. The trouble is that the lighting will affect different elements in the picture depending on where they were in relation to the light. So no overall frame correction will completely solve the banding, which gets us into selective area correction and travelling masks for any moving elements! The amount of time and effort for even a probably partial solution would be better spent begging, borrowing or even paying for a reshoot at the location, or at another alternative location IMHO.

 

[Stephen Strangways]

Yeah, unless some clever company like The Foundry comes up with some brilliant plugin to remove it automatically, you're looking at a very time consuming selective colour correction perfectly sized and blended to match the banding, and animated at the proper speed. It's going to be a very manual job. It should be rather consistent if you are always shooting under the same ballast, but if you move throughout the shot and are under different banks of fluos, the banding could change slightly.

 

[Stephen Pruitt]

Wow. . . I'm surprised that no one makes a fluoro fix plug-in. You could vary the intensity and size and speed of the band and there you are.

I'm afraid that a reshoot may not be possible at this location, which is really unfortunate, because we have the same location at night. It is very slight banding, so maybe the moving mask might work in this case.

Alas. . . never had this happen before. We just didn't see it on the monitor.

Thanks.

Stephen

 

[Nick Gardner]

you can usually check for this by framing up a shot with out talent and watch t5he histogram or better yet the wavefrom monitor. If you have discharge lighting issues it will pulse.

Nick

 

[Sergei Franklin]

Shoot with a 144 degree shutter angle in 60Hz countries. (at 23.98 fps)

 

[Stephen Strangways]

This can also happen with dimmed LEDs that use cheap dimmers.

 

[Dan Hudgins]

Temporal noise reduction

Temporal noise reduction

I have had some success with reducing "moving pattern noise" by using Temporal noise reduction, so if your "flicker" bands are moving that means that you can average them out by using Temporal Noise Reduction on parts of the image that are not moving.

NeatVideo (tm) may help as it seems like it can reduce frame to frame noise.

I've noticed that you don't see it on subjects that are moving as much, mostly on backgrounds.

My Temporal code can average a two second window, its part of the new de-Bayer program I am finishing up now getting the docs made before its release...

Depending on the 50/60Hz "roll time" you may be able to optimize the moving average window to "fit" the movement of the flicker since your flicker has some coherence, whereas the ones I have been dealing with (from another camera) relate to low Hz digital noise and move around "randomly"...

You might make DPX frame set and run it through ARRI Relitivity to see if that helps as well.

If the bands stand still, then you might make a "mask" and subtract that from the images.

If the bands are uniform and move slowly, then you might make a animated "mask" but that would take some skill.

Here is the link for NeatVideo (tm),

http://www.neatvideo.com/

I think their code is 8bpc, but you can ask them, I have only run 24bpp BMP through their test demo, I got some motion artifacts and such so I went on to write my own code for static camera shots that does 48bpp for DPX and TIF processing, and that will be a freeish download that works with "any" size frames (within the 2GB that Windows (tm) lets the program use). Temporal noise reduction for moving camera shots works better at 60p or faster since the motion blur at 24p from fast camera and subject motion makes matching between frames hard so the motion estimation becomes an issue. Temporal works better with 60p video since you can stop the shutter down to 5 degrees at 60p, whereas with a rolling shutter and 24p you don't want the shutter that small with rapid motion of the camera or subject. Also it may be better to use Temporal on shots before editing trim to length since some run-up and run-out frames help avoid a pickup of noise at the head and tail of the shot since you trim those parts off if you process the whole take, then do the edit trim.