Rolling shutter example

[Greg Laves]

Originally Posted by Tom Hardwick

Not sure if you read my post through Greg. In less than 4 minutes of video there were 64 flashes. Most DSLRs put out two flashes for each exposure, and some fire long bursts to get the exposure correct. Correcting that little lot is not what I'm paid for.

Yup, I read it. On a lot of EX-1 wedding footage I have seen, not every flash produces the banding effect. I would think that you wouldn't have to fix all 64 flashes. As for the multiple flashes from the DSLR's, that is the camera's attempt at reducing red eye and typically, the pre-flashes are lower intensity. I am surprised that the pre-flashes caused any problem for the EX-1.

[Tom Hardwick]

Phil - the alternatives you mention (JVC, Panasonic & Canon) all use CCDs in the Z5/7 price bracket, so don't suffer the CMOS failings. But to get this in perspective, it's really only electronic flash (and police car / ambulance lights etc) that cause this nasty partial frame over-exposure.

So you've got to ask yourself this - how much of my footage is flash frame filled? I suspect for most of us the answer's 'not much', in which case buying into the early years of CMOS isn't too difficult a decision.

tom.

[Gary Nattrass]

Has no one produced a plug in to cure this in post prod?

[Tom Hardwick]

It would be very difficult to do Gary. I took my last couple out into the night with the stills tog. I used 1/3rd sec shutter speed on the Z1 as he fired away at them with flash. I then added motion blur on the timeline and this (a surprise to me) produced a beautiful 'sine wave' of light - gently ramping up, staying bright on the couple and then slowly fading away.

Would've been unusable if only a third of the frame had been exposed, but then maybe CMOS does indeed light the whole frame at slow shutter speeds. Anyone?

It really looks as if this is an effect I've spent an age on in post production but it was literally one click of the Canopus filter combined with the very slow shutter speed (and rock-solid camera of course).

I'm just mighty glad I wasn't using CMOS chips at this time as their better low-light capability might have induced me to use a shorter shutter speed, and not created this effect. And not once did I get CCD smear from the outside lights, so CCDs still suit me and the type of work I do just fine. CMOS has been perfect for DSLRs because (up til now) they havent shot movies.

tom.

[Dave Blackhurst]

Originally Posted by Gary Nattrass

Has no one produced a plug in to cure this in post prod?

I played with the flash remover plugin for Vegas from BlueFX - it was "somewhat" sucessful at removing RS flash artifacts... introduced some odd glitches from what I saw in parts of the frame with fast motion, but it's worth a look.

I'd think that a plugin is possible - just has to be able to correct brightness/contrast over PARTIAL frames. The challenge is that the RS causes different portions of the frame to show the effect at different times depending on the phase relationship of the flash to the CMOS "cycle". You get everything from a partial frame to a full frame (on rare occaision), so you have all these "partial" overexposed frames to deal with.

[Tom Hardwick]

Originally Posted by Dave Blackhurst

you have all these "partial" overexposed frames to deal with.

Beautifully put Dave. And when the film rolls I see these frames in the same way as I can see a 'left-in-frame' from sloppy editing, or see a single frame of subliminal advertising.

I'm sure as CMOS matures the effect will be minimised, but right now it's akin to the vertical CCD smear so common on cameras such as the Sony PDX10. The XHA1 and Z1 have all but eliminated it.

tom.