Very technical question on Canon SD image sensor

[147798]

Does anyone know the answer to this technical question:

In cameras that have movie mode (like the Canon SD series), the movies are only 640X480. Does that mean they are only using 640 pixels by 480 pixels from the CCD to capture the movie image, or do they use the whole CCD chip and downscale it in the camera to 640X480.

Also, on a related topic, given that a movie might be 30 frames per second, is the shutter speed for each frame that slow (1/30), or is the shutter speed possibly higher (say 1/60 or even 1/250), but its only take 30 each second.

I now these are not run of the mill questions, but I am not sure where to go with these. I can't quite seem to find an answer on various forums and website. Thanks in advance if anyone knows the answers. Also, if anyone knows of a good Q&A site that might be able to answer such a question, that'd be really helpful.

 

[taylorwilsdon]

As far as movies go, I'm pretty sure that its not a repeated opening and closing, like it was in olden days, but rather some sort of constantly open recording.

Or I could be completely wrong.

 

[TuffLuffJimmy]

There is no longer a shutter in cameras except in DSLRs. So there is no shutter speed, just frame rate. and I'd expect that it uses the entire sensor and just scales the image.

 

[MacNoobie]

Well it depends on how the designers and engineers made the camera and to be honest I'd expect it to be easier and more battery efficient to read out just 640x480 resolution off the sensor. Do that at what ever rate the user specifies (aka 30fps or 15fps) and maybe what ever resolution the user specifies (I've noticed cameras do 1280x720 video and 800x600 and so forth). I'd be a lot easier to do that (from a technical standpoint) then take the full resolution of the sensor and scale it down it'd just be way too much computational power for the camera (which adds cost) and it'd burn batteries trying to record (which for a camera battery life is a must), scale down, compress, and ultimately write the file to the memory card.. just easier to record, compress and write the file to the card.

I'd be surprised if Canon (or anyone else) choose to use some sort of scaling down from full resolution to movie.

 

[147798]

Thanks for the replies. They all echo my thinking, but at this point it's just that -- thinking. It is all dependent on how the engineers build it.

You can have a chip that scales the image before writing. They don't have to write, scale and re-write. So, scaling the image is possible.

My concern is with these new "12MP" sensors, if they are only taking a literal 640X480 pixels, then the physical pixels are smaller, and could be more noisy.

My guess also is this is why an SD camcorder takes good quality movies but poor pictures -- because the entire 1/3" CCD is filled w/pixels for a 640X480 movie -- but a pocket camera with a high MP count takes poor SD movies -- because only a portion of the 3000X4000 pixel count chip is used. Although, this is just an educated guess. I don't really know.

On "shutter speed" though there are no shutters, there is still the concept of length of exposure before writing the data to memory. That's why you can have burry or clean pictures. Even with my SD800, that sometimes a single frame in a movie can seem blurrier than in other movies. But, I'm not sure if it's my perception, or if the "exposure time" for each frame is variable, or if it's fixed.

Really do appreciated the thoughts and input. I am blindly fishing for anyone who might know these answers, or if anyone knows any good technical websites that might be good sources for this kind of information.

 

[cube]

No, I would imagine that the digicams use pixel binning when shooting movies for increased sensitivity.

Don't forget that there are SD camcorders that also take pictures of over 5 MP.

 

[147798]

No, I would imagine that the digicams use pixel binning when shooting movies for increased sensitivity.

Don't forget that there are SD camcorders that also take pictures of over 5 MP.

In theory. But, I've found at least one that, it turns out, was actually interpolated to a higher MP, though the CCD itself was still more than .3MP (640X480).

Thanks for the bit on pixel binning. That makes a lot of sense, and points me to a new term! I just read up some more at http://www.ccd.com/ccd103.html. This makes a lot of sense. It also leads to another question -- if you down scale on a 12MP camera to take a resolution 1/2 that, do they use pixel binning for that and, if so, does it reduce noise in the output (roughly equating higher MP with more noise in lower light situations, so if you pixel bin, do you end up with a better signal/noise ratio with a 6MP shot than a 12MP shot on the same camera)

 

[ChrisA]

Seeing as the field of view is the same in movie or still mode for a given zoom setting the effective sensor size must be the same in both movie and still modes.

My bet is that the image is not scaled down to 640X480. Likely the scaling is built the algorithm used to interpolate the raw pixels into an image. In other words the image is created at 640x480 by combining raw pixels. This should alow for a high effective ISO and low noise at the same time.