DPChallenge: A Digital Photography Contest You are not logged in. (log in or register
 

DPChallenge Forums >> Hardware and Software >> Canon develops 120 megapixel sensor
Pages:  
Showing posts 1 - 23 of 23, (reverse)
AuthorThread
08/24/2010 10:09:44 AM · #1
New sensor form Canon

No mention of when it will show up in a production camera.
08/24/2010 10:18:50 AM · #2
Oh my! More MP than anyone would know what to do with!
08/24/2010 10:31:49 AM · #3
Originally posted by moriadelacroix:

Oh my! More MP than anyone would know what to do with!


Bah, I know just what I'd do with it.. (aside from filling multiple hard-drives with data.)
08/24/2010 10:33:35 AM · #4
You do have to wonder though, if my comparatively low-resolution 15 megapixel sensor can show the limitations of all but the very best glass, what kind of lens will you need to actually take advantage of 120 megapixels of resolution? That's going to need some very excellent glass, and incredibly precise AF sensors..

Message edited by author 2010-08-24 10:33:54.
08/24/2010 10:47:23 AM · #5
I don't know if these ever plan to get into production cameras. Canon produced a 50 MP sensor in 2007 and we haven't seen anything from that yet.
08/24/2010 10:59:17 AM · #6
Originally posted by coryboehne:

You do have to wonder though, if my comparatively low-resolution 15 megapixel sensor can show the limitations of all but the very best glass, what kind of lens will you need to actually take advantage of 120 megapixels of resolution? That's going to need some very excellent glass, and incredibly precise AF sensors..


True, dat.
Keep in mind, though, that it *only* represents a 2.3x increase in linear resolution compared to a 15Mpx APS-C sensor. So it is a huge increase in the amount of information recorded, but a suprisingly modest step in linear resolution. Still, the difference in detail in the recorded image would be stunning.
Now, with respect to lenses, yes, there definitely will be high demands on glass. Many of the best current lenses actually will be up to the task, at least over most of the frame and when stopped down 1 or two stops from wide open. But it's not entirely a bad thing if the sensor outperforms the lens by a narrow margin. This is called "oversampling", and will result in images with smooth edges when viewed at 100%, rather than the jagged (pixelated) edges we are so used to due to undersampling (sensor has less resolving power than lens). Oversampling also completely eliminates moire, and reduces demands on demosaicing algorithms. Even with a very unsophisticated algorithm, the results will be almost indistinguishable from a very sophisticated (read "processor intensive" algorithm. This is because the color does not vary significantly across a two-pixel distance.
08/24/2010 11:01:52 AM · #7
Originally posted by kirbic:

Originally posted by coryboehne:

You do have to wonder though, if my comparatively low-resolution 15 megapixel sensor can show the limitations of all but the very best glass, what kind of lens will you need to actually take advantage of 120 megapixels of resolution? That's going to need some very excellent glass, and incredibly precise AF sensors..


Keep in mind, though, that it *only* represents a 2.3x increase in linear resolution compared to a 15Mpx APS-C sensor.


Fritz, I appreciated your thoughtful analysis of the benefits of oversampling. But is there a short explanation for how the math works with this statement?
08/24/2010 11:29:39 AM · #8
Originally posted by mycelium:

Originally posted by kirbic:

Originally posted by coryboehne:

You do have to wonder though, if my comparatively low-resolution 15 megapixel sensor can show the limitations of all but the very best glass, what kind of lens will you need to actually take advantage of 120 megapixels of resolution? That's going to need some very excellent glass, and incredibly precise AF sensors..


Keep in mind, though, that it *only* represents a 2.3x increase in linear resolution compared to a 15Mpx APS-C sensor.


Fritz, I appreciated your thoughtful analysis of the benefits of oversampling. But is there a short explanation for how the math works with this statement?


Simple, it's the same idea as aperture calculations..

Effectively linear means the measure along one side (at least in this usage), so, we will assume a square sensor (which it is not, but the results don't differ by much, and it's the theory you want...)

So, if we assume a square sensor, we take the square root of 120 and compare to the square root of 15, giving us 10.95 vs. 3.87, which means a linear increase of 2.83 or so... Clearly that doesn't exactly match 21.gif kirbic's number, but the idea should be obvious...

Make sense?

Basically in this case, he's telling us that there are about 2.3 times as many pixels along each side of the image as when compared to a 15MP sensor..

Message edited by author 2010-08-24 11:30:41.
08/24/2010 12:57:56 PM · #9
Originally posted by scarbrd:

I don't know if these ever plan to get into production cameras. Canon produced a 50 MP sensor in 2007 and we haven't seen anything from that yet.


Yeah, it's just a proof of concept thing. Showing they can do it. Gives people to look forward to.

In the mean time they'll sell us a 5D mk3, mk4, mk5, mk6....
08/24/2010 01:38:15 PM · #10
zoom in on aliens with it...

Message edited by author 2010-08-24 13:38:23.
08/24/2010 01:51:49 PM · #11
Originally posted by hojop25:

zoom in on aliens with it...

Or take macros of bacteria ...
08/24/2010 01:55:04 PM · #12
Originally posted by GeneralE:

Originally posted by hojop25:

zoom in on aliens with it...

Or take macros of bacteria ...


ooooooh
08/24/2010 02:37:11 PM · #13
we are going to need bigger memory cards.
08/24/2010 03:16:30 PM · #14
Originally posted by mike_311:

we are going to need bigger memory cards.

I think RAW is 48-bit color, so for a 120MP sensor you should end up with 120 (MP) x 6 (bytes/pixel) = 720MB nominal uncompressed size/image (360MB if you are shooting 24-bit JPEG or TIFF).

Note: 48-bit color is also known as 16-bit -- that is 16 bits (or 2 bytes) per RGB channel.

Likewise 24-bit color is the same as 8-bit (1 byte)/channel color -- what on monitor settings is known as "millions of colors."

24-bit color yields about 16.7 million colors -- not all of them distinguishable by the human eye!
08/25/2010 03:58:30 PM · #15
I wonder how this would affect noise. And wouldnt it be better to concentrate efforts on a new less noisy sensor? How about a sensor that gives the same performance we get at iso 50? :D

And how about a sensor similar to the fuji s5 ccd that improves dynamic range?
08/25/2010 04:51:16 PM · #16
Originally posted by dmadden:

I wonder how this would affect noise. And wouldnt it be better to concentrate efforts on a new less noisy sensor? How about a sensor that gives the same performance we get at iso 50? :D

And how about a sensor similar to the fuji s5 ccd that improves dynamic range?


While I agree with you, from a selling to the public standpoint, it's all about megapixels.
08/25/2010 06:32:37 PM · #17
Originally posted by dmadden:

I wonder how this would affect noise. And wouldnt it be better to concentrate efforts on a new less noisy sensor? How about a sensor that gives the same performance we get at iso 50? :D

And how about a sensor similar to the fuji s5 ccd that improves dynamic range?


without a doubt we need progress on noise, especially fixed-pattern noise and noise related to read-out (normally appears as banding). We also need additional progress on DR, although we are in fairly good shape there. We certainly don't want back-sliding, and I don't think we'll get that, but with the focus on more pixels, I fear we won't see much progress.
DR is all about what's referred to as full-well capacity, which is simply the maximum number of electrons that can be stored at a pixel location. The physical size of the pixel is important, but so is the "charge density" of the storage. If pixels get smaller and we want greater storage, we need either greater depth (more volume vertically) or greater density, or a combination of both. It will be interesting to see what the next three to five years bring.
08/26/2010 11:24:11 AM · #18
How about that Fuji sensor on the S3/S5, I though Nikon would have capitalized on it. Was that CCD technology the way forward, in DR improvement?
08/26/2010 11:48:53 AM · #19
Originally posted by dmadden:

How about that Fuji sensor on the S3/S5, I though Nikon would have capitalized on it. Was that CCD technology the way forward, in DR improvement?


No. It was a good interim solution. The real path forward is to increase the full-well capacity of the pixels themselves.
08/26/2010 12:01:23 PM · #20
Just for perspective, navigate to this page //www.dpchallenge.com/camera.php?CAMERA_ID=1668 to see a number of DPC challenge entries that scored higher than 7 (one) or higher than six (nine) using a camera with a 2MP sensor.

Of course, astrophotography and other specialized applications might actually be able to use huge numbers of pixels.
08/26/2010 02:43:12 PM · #21
Originally posted by bob350:

Just for perspective, navigate to this page //www.dpchallenge.com/camera.php?CAMERA_ID=1668 to see a number of DPC challenge entries that scored higher than 7 (one) or higher than six (nine) using a camera with a 2MP sensor.

Of course, astrophotography and other specialized applications might actually be able to use huge numbers of pixels.


Not to cavel, but try making a large print out of any of those... :)
08/26/2010 05:06:39 PM · #22
Originally posted by coryboehne:

... try making a large print out of any of those... :)


Actually, you make exactly the point that I was trying to indirectly suggest: the need for a particular quantity of pixels depends on the intended use.

In the mid 1980's, my brother-in-law was amazed that I had just acquired a computer with a gigantic 10 megabyte hard drive. His comment was: "who would ever need more than a couple of megabytes?" Of course, now that drive wouldn't hold even one of the common programs we use every day (or even one high resolution multi-layered image for that matter).

There will be good uses for very high megabyte sensors, and electronic equipment will evolve to hold and process the resulting images efficiently. Fun to think about while waiting to see what actually enters into production.

08/26/2010 05:10:31 PM · #23
Originally posted by bob350:

Originally posted by coryboehne:

... try making a large print out of any of those... :)


...

In the mid 1980's, my brother-in-law was amazed that I had just acquired a computer with a gigantic 10 megabyte hard drive. His comment was: "who would ever need more than a couple of megabytes?" Of course, now that drive wouldn't hold even one of the common programs we use every day (or even one high resolution multi-layered image for that matter).

...


LOL, or for that matter, even 1 shot from my camera in RAW. As a matter of a fact many of my photos exceed 20Mb quite easily.. :) Those 720Mb 120MPix images would just murder card and drive space..

Then again, that would make us careful about what we did during composition of the shot and would force more pre-thought in our photography :)
Pages:  
Current Server Time: 05/22/2019 07:45:37 AM

Please log in or register to post to the forums.


Home - Challenges - Community - League - Photos - Cameras - Lenses - Learn - Prints! - Help - Terms of Use - Privacy - Top ^
DPChallenge, and website content and design, Copyright © 2001-2019 Challenging Technologies, LLC.
All digital photo copyrights belong to the photographers and may not be used without permission.
Proudly hosted by Sargasso Networks. Current Server Time: 05/22/2019 07:45:37 AM EDT.