I was shooting with a RED MX recently on a location that featured a lot of venetian blinds.
And I started to notice I was getting a strange horizontal flare. Well I thought it was a flare, but it seems like something more is going on.
It’s a kind of fault that can be reproduced. All you need is a venetian blind, a background behind the blind that is 3 or more stops over your base exposure and something dark(ish) in frame.
At two stops over you can *just* start to see this fault coming in, but it’s barely noticeable. By +3 you’ll definitely see it.
I even realised it doesn’t even have to be as thin as a venetian blind.
I’ve also been able to replicate this fault on an Alexa, but the clip below is from a RED.
Speaking with RED I got a vague explanation that it would happen anytime a hot source like this was near the edges of the frame and not to shoot venetians with hot backgrounds.
I’d still love to know the technicalities of what’s going on here if anyone knows ?
johnbrawley 
Hi John, it sounds like they were suggesting it’s edge diffraction but it isn’t that. To me it looks like light bouncing backwards off the inside of a lens element or filter which you would see in areas of such different contrast. It looks so precisely mirrored that it’s probably a filter (if you were using one).
<a href=http://imageshack.us/photo/my-images/269/2011060622h16m22s187.png
<a href=http://imageshack.us/photo/my-images/269/2011060622h16m22s187.png
Hi. It’s not a filter problem.
There wasn’t a filter used in this test.
hello there. mailing from Poland.
in my opinion it’s a reflection of a whole picture FROM SENSOR itself, which is bounced from sensor’s surface to back element of the used lens and bouncing this time from back element to the sensor, and so on. it’s “endless” process.
Try to COVER part of the shot, so just i.e. middle part will be bright like 3+ over. and then you will see how it looks. if you can measure the difference between full black and lighter stripes on black somehow, they should be in range of 4% difference regarding black level – if i remember this correctly.
all the best,
filip
This could be complete FUD; a result of years of reading useless tech-briefs, and allowing them to simmer in my mind into a spicy gumbo of flavorful tech-bits that taste nothing any of the original ingredients… but, I recall some technical notes on CMOS sensors, and how when overexposed in high-contrast situations, particularly on the edge of frame, electrons will actually seep horizontally within the photosites of the raster line. CCDs are known to do a similar thing, but vertically, and much more pronounced. Unfortunately, I can’t remember where where I might have found any of that stuff to actually lend credibility to my statement.