Visible noise pattern in PJ footage

[Stokestack]

David Mullen remarked on something I noticed but didn't hang out long enough to ask Jim or Graeme about:

"a faint patterning from a sensor that needs re-aligning, that only became visible when an underexposed shot was brightened in post (you saw something similar in some shots in "Deja Vu" shot on the Genesis). Graeme Nattress says that it will be fixed soon."

This was visible in a shot under the plane's dashboard, where the pilot has wedged the teddy bear against the priming handle. It looked just like the pattern you see when bringing up the "exposure" on a Canon raw file, sort of reddish grid lines. Does it arise from some characteristic of CMOS sensors?

I wonder what Red's approach will be on this. I think we all expect increased noise when boosting gain, and diffused and "random" noise is tolerable; a visible pattern obviously isn't going to fly.

[Zk2007]

It's been talked about many times by now. RED is aware and addressing the problem. It seems they even pointed it out to some people at NAB. They know it and we know they know it.
Also remember those were prototype cameras.

[Yuval Shrem]

Also, don't forget that these shots were bluntly improperly exposed (under-exposed about 4 or 5 stops) as a result of a human error, and not because of the lack of light or sensitivity of the sensor. The fact that it was even possible to save these shots with such a mild side-effect is quite impressive by itself!

[Damien Molineaux]

Also, don't forget that these shots were bluntly improperly exposed (under-exposed about 4 or 5 stops) as a result of a human error, and not because of the lack of light or sensitivity of the sensor. The fact that it was even possible to save these shots with such a mild side-effect is quite impressive by itself!

I can't forget, since I didn't know. Interesting indeed, where did you get this info from (and why do you think it's common knowledge ?) ?

Cheers,
Damien

[martinnoweck]

Hi Damien,

Jim mentioned it in another thread.

Martin

[David Mullen ASC]

It was a topic of discussion at the booth. Graeme Nattress told me that it had something to do with how the vertical sensor site rows were selected for being discharged or something (now I can't remember) -- any variations caused each row to be slightly different in exposure, causing a fine line pattern. But he said that it could be fixed by realigning the sensor, which they were doing.

I'm sure you could do some research on how images are dumped from a CMOS sensor over the internet and learn more.

[Graeme Nattress]

Different sensors have different characteristic noise patterns. CMOS can make these vertical lines when not calibrated out. We were in the process of calibrating this out just before NAB (just after the PJ shoot), so it may even be fixed by now, but it's certainly in the process of being fixed.

I'd not worry about it, and indeed, that's why Jim pointed it out. We know about it and it's being fixed.

If you're interested in sensor technology in general, Dalsa have some good white papers on CCD and CMOS. Obviously Mysterium isn't a Dalsa sensor, but if you're after basic principles and understanding the papers are excellent and well balanced between the CCD and CMOS technologies.

Great to have you on our forums David, and superb to get to meet you on wednesday, so thanks again.

Graeme