rolling shutter

[Alister Chapman]

It's called a rolling shutter because in effect the imager scan continuously rolls to the top of the sensor at the end of each scan and then rolls down the sensor to the bottom before repeating the process again. The sensor is constantly being scanned. It has nothing to do with progressive or interlace.

[Dean Sensui]

In a wedding video that's thoughtfully shot, with good audio and great moments from that most important day, I wonder how many bridal couples will even notice the partial-frame flashes?

[Marty Welk]

Originally Posted by Alister Chapman

It's called a rolling shutter because in effect the imager scan continuously rolls to the top of the sensor at the end of each scan and then rolls down the sensor to the bottom before repeating the process again. The sensor is constantly being scanned. It has nothing to do with progressive or interlace.

I realise that it has nothing to do with a "progressive" picture itself. but look a that word itself as it is used, it does little to describe the differences of video progressive and interlace itself.

progressive
1. Moving forward; advancing.
2. Proceeding in steps; continuing steadily by increments

That doesnt speak english of a "Full Frame of video captured in one pass" at all.
But it has caused about 1 Million questions as to what the differences are between progressive and interlace. Many people still not understanding the differences, or the advantages interlace once had.
Interlace or Interleaved Feilds, is easily a correct wording for the other,and easily explained.
Do they have a word for that in Japan, that didnt translate?

the chip IS scanned progressivly (its in the book), the progressive picture isnt progressive , it is bad form of Videnglish that only serves to confuse. There is no shutter, and the rolling is a progressive scan by techincal definition.
I am just saying the "techwords" that we have adopted are not helpfull at all, they instead confuse. then reading 30 pages of techjunk we get the real meaning.
Still have to exsplain it to the customer if they have questions or what to know why things look the way they do.

Which brings up one more important thing.
Is an Interlaced full frame from a CMOS captured in 2passes, captured interlaced out of 1/2 the lines (because it can be scanned and reset that way), or interpreted out of something after capture.
WHY?
because interlace on the CMOS HAS 2 seperated feilds that both have different feilds of motion. (real true 60feilds from different times)
BUT
interlace or progressive , the sensitivity of the collection , or low light capability, doesnt seem to change? (diode exposure time)
IF
you shoot in 60i or 30P , it doesnt seem to change the exposure capability of the camera. knowing stuff like that, could be usefull for choosing IorP from a cmos when working in lower light, or when worrying about Skew problems.

yesterday, on the David letterman show, they made some techspeek video geeks look like clowns for a skit, sort of shows that we will be keeping our jobs, no regular human could understand a word/abbrev. they said :-)

[Alister Chapman]

My guess is that the sensitivity doesn't change because each single displayed field line is made up of the sum of two scan lines. The summing of overlapping line pairs is done to prevent line twitter and is normal practice in interlace for both CCD and CMOS cameras.

Field 1:===================Field 2:

Display Line1 = sensor scan line 1+2

========================Display Line 2 = sensor scan line 2+3

Display line3 = sensor scan line 3+4

========================Display Line 4 = sensor scan line 4+5

Display line 5 = sensor scan line 5+6

and so on.....

The sensor must be scanned twice (as you would expect) for each interlace frame as flash artifacts normally only appear in one field and not the other.

[Richard Hunter]

Originally Posted by Marty Welk

I realise that it has nothing to do with a "progressive" picture itself. but look a that word itself as it is used, it does little to describe the differences of video progressive and interlace itself.

progressive
1. Moving forward; advancing.
2. Proceeding in steps; continuing steadily by increments

That doesnt speak english of a "Full Frame of video captured in one pass" at all.
But it has caused about 1 Million questions as to what the differences are between progressive and interlace. Many people still not understanding the differences, or the advantages interlace once had.

I agree that the use of jargon gets crazy sometimes, however I do see some logic and meaning in the use of the word progressive to differentiate the whole frame scan from the 2 x interlaced field scan. As in your quotes, progressive scan moves steadily in the same direction until the entire frame is scanned, whereas the interlaced scan has to jump back almost to the start when it gets halfway through the frame.

Richard

[Marty Welk]

Originally Posted by Alister Chapman

The sensor must be scanned twice (as you would expect) for each interlace frame as flash artifacts normally only appear in one field and not the other.

but the confusion comes in when , if it was being scanned 2 times for 2 different Interlace Feilds in 60i, then the rate/speed of the whole chip being scanned is minimum ~60fps.
then combined to make the single 30fps interlace FRAME.

When it is in progressive the chip only has-to be scanned at ~30Fps , to make the single 30fps progressive frame.

That would provide for 2times the photo diode exposure time, aka 2 times the low-light capability. Scanning at 60 vrses scanning at 30, reguardless of the final frame assembly.

we can sort of SEE that there is something different IN the scan rates , they cant both be ~60fps because we can alter the "shutter" in 30 , which sets the cmos exposure time right? We can set all kinds of "shutter" speeds in the camera for 30P 30 40 50 60.
plus we can observe that it does have 2 feilds in the interlace at 60i.

So why IF the chip is being scanned at ~60 (for interlace) Vrses ~30 is there no chip sensitivity differences?

This comes from many thing i have read, not fitting together in total "logic" when you assemble them, then test the camera itself.
Plus my CCD cameras did increase huge in light sensitivity when the 60i signal was shuttered to a (still interlace output) 30fps , for those it may have been frame doubling post processing not chip exposure time. We cant have frame doubling here being a factor, if we have motion between the 2 feilds still, and the Sqew ammount changes with the scan rate, and the chip is scanned "rolling".

Side notes: A analisis of these cameras that also did not make sence, indicated that at 24fps , the "sqew" ammount reflected a scan rate of the chip from top to bottom at 60fps.
How can the camera have 24fps motion blur, 24fps frame judder, ability to set a shutter speed below 60fps that changes the picture, AND still have a squew that reflects only 1/60th of a second.
it is a virtual impossibility, in my mind, but i have an idea how they may have assumed that, based on output they view, or else the device is working different than they are saying.

* all of this discludes the idea that my brain doesnt function normally, :-) or there is a missing piece of information.

[Alister Chapman]

Did you not read my previous post?

Because in interlace for each single line displayed, 2 lines are summed, ie added together, thus the signal from one pixel is added to the signal from the pixel below which gives TWICE the signal, thus twice as much sensitivity. So if you have twice as much sensitivity but only half the exposure you get the same sensitivity as an exposure with half the sensitivity but twice the exposure.

The scan speed is simply the rate at which the readout cycles from top to bottom of the sensor. You slow the frame rate by adding a delay or pause between the start of the each scan sequence. So if the scan rate is 1/60th and you shoot at 24P (1/24th) the delay period is 1/40th of a second. So that means you have a period of 1/40th of a second where the sensor is not being scanned but is gathering light and this is why you get less flash band issues at slower shutter speeds and frame rates.

Fast shutter speeds are achieved by reseting the sensor rows to zero in time with the sensor scan such that the time (integration period) between each row being reset and then scanned is equal to the shutter speed. Hence the shutter roll as you get a narrow band of active pixels rolling down the sensor when using a fast shutter. This is why flash banding is worse at higher shutter speeds.

[Marty Welk]

line summing also doesnt make sence to me.

they claim the resolution drop from 30-60 is from 900-700 , and i have interlace full frame pictures here that dont seem to be slapping together lines, at least not as badly as thier picture showed, with all the "twitering" and jaggies and such.

http://home.comcast.net/~TVV0/MiniStage.jpg <--- that is interlace, it has not been de-interlaced.
zoom up to 200% and it doesnt look neer as bad as the pictures they were showing of the EX1, where they showed obvious issues with the interlace lines, thier picture however it was done, did show that some nasty stuff was going on in the interlace , this picture is Post compression besides.
On the Blue toys RED nose, i can see a completly wacked out line there.

i should probably get a solid picture of Both 60i and 30p of the exact same thing, and see what i am missing.

[Marty Welk]

now i see
Scene30Pvs60ion toplayer.psd
(8megs Psd layered compatability file)
this is an adobe photoshop with the Top layer being 60 interlace (not de-interlaced)
it is closly matched to the other picture, being 30p below it.
In "layers" window in photoshop, a person can unView the top layer and see the differences.
Switching the layer to diference will also show differences, although it is not aligned that well, and there is many compression artifacts.

What i noticed so far. in interlace
Red seems to have the biggest issues
is all a bit muddy
the kool aid thing basically goes to heck all over it
loose lines of color hoping off the color
resolution on the furry stuff goes muddy
the color box edges hold mostly with the red going wackey
the artifact noise is different, slightly less
full black to white edges look similar