Sony CMOS -- SSE too?!?

[Steve Mullen]

When I first saw the posted image of a lighting strike captured using a Sony CMOS HDV camcorder I notice what looked very much like "horizontal SSE." The bottom 1/3 looked darker than the top 2/3 of the image.

Since the Sony CMOS chip has 3 read-out ports -- the 1/3 division made sense -- just as the twin read-out ports on the HD100 CCDs give rise to 1/2 -- 1/2 split-screen.

This got me to thinking about Image Sensors and "their artifacts." For example, CCDs smear but CMOS does not.

So I've started "The Sensor Sessions" at: http://www.gyhduser.com/

The first two sessions cover CMOS. Then we'll look at Interlace Scanning for CCD and CMOS chips.

Then I'll cover Progressive scanning and lastly I'll look inside the HD100 CCD/DSP system to provide an undstanding of SSE and how to control it.


Obviously, I can't post a copy of each Session here -- so I'll post this one link for those who are interested in such topics.

[Serge Victorovich]

Originally Posted by Steve Mullen

The bottom 1/3 looked darker than the top 2/3 of the image.

The bottom 1/3 is a Sony Trinitron feature:)
Steve,video captured from component has same defect too?
May be problem not a cmos imager but a dsp?

[Steve Mullen]

Originally Posted by Serge Victorovich

The bottom 1/3 is a Sony Trinitron feature:)
Steve,video captured from component has same defect too?
May be problem not a cmos imager but a dsp?

I expect it was 1394.

The JVC moves RGB via two paths and can get SSE. Each read-out can get slightly different data.

As I said in my story, I was speculating based up the 1/3 split of the displayed image and the fact Green is moved from the CMOS chip via 3 paths to the DSP.

With a Bayer RGB filter there is 2X more green pixels to read out. With 1440 (and some claim 1920) pixels per line that pushes the Green clock rate VERY high. I'm assuming Sony reads G out via 3 paths for the same reason JVC reads out of 2 paths -- to allow a reduced clock rate.

But, that seems to open up the door to differences between the paths and to SSE.

[Thomas Richter]

Please correct me if I'm on the wrong track, but could the darker lower third in the lightning image filmed with the HC1 could just be a rolling shutter artifact?
As I understand it the CMOS of the HC1 is read out line by line, not as a whole. The rolling shutter technique increases the time light can fall on the sensor but goes with nasty artifacts.

In my understanding the illumination caused by the lightning varied over the time of the frame or field (haven't seen the footage). As the lines in the lower third of the image were read out later it is possible that they caught less light during exposure.
One simple way to test this hypothesis is to check how the HC1 records pictures of a strobe light.
By the way, I would prefer the risk of SSE over the rolling shutter issue - but hey, you get what you pay for

[Steve Mullen]

Originally Posted by Thomas Richter

Please correct me if I'm on the wrong track, but could the darker lower third in the lightning image filmed with the HC1 could just be a rolling shutter artifact?
As I understand it the CMOS of the HC1 is read out line by line, not as a whole. The rolling shutter technique increases the time light can fall on the sensor but goes with nasty artifacts.

I now think you are exactly correct and my idea was wrong.

I have extensively revised CMOS Session #2.