| Author | Thread |
|
|
08/12/2005 06:11:01 PM · #1 |
I have a simple question:
Does the quality of a digital camera effect the overall 'grainyness' of a specific ISO? In other words, Is my Canon A75 going to have the same amount of grain at ISO 400 when compared to a high end camera?
Hope that makes sense. |
|
|
|
08/12/2005 06:12:35 PM · #2 |
No. Camera brands always try to reduce the amount of grain when they're making a new camera.
|
|
|
|
08/12/2005 06:14:53 PM · #3 |
Point and Shoots will have a lot of noise at high ISO's when compared to DSLR's
For example my 1D at ISO 3200 might be equivalent in terms of noise to a Canon G2 at ISO 200 |
|
|
|
08/12/2005 06:16:44 PM · #4 |
| Hmm..so therefore, ISO is relative to the camera you are currently using? There is no standard for what a certain ISO should be? |
|
|
|
08/12/2005 06:17:10 PM · #5 |
Originally posted by doctornick:
Point and Shoots will have a lot of noise at high ISO's when compared to DSLR's
For example my 1D at ISO 3200 might be equivalent in terms of noise to a Canon G2 at ISO 200 |
I don't know that the difference is so extreme! (perhaps more like 800 to 100). |
|
|
|
08/12/2005 06:18:01 PM · #6 |
Originally posted by Gil P:
Originally posted by doctornick:
Point and Shoots will have a lot of noise at high ISO's when compared to DSLR's
For example my 1D at ISO 3200 might be equivalent in terms of noise to a Canon G2 at ISO 200 |
I don't know that the difference is so extreme! (perhaps more like 800 to 100). |
It actually is, I'll shoot with both and post it here. |
|
|
|
08/12/2005 06:18:11 PM · #7 |
Originally posted by mattmunger:
I have a simple question:
Does the quality of a digital camera effect the overall 'grainyness' of a specific ISO? In other words, Is my Canon A75 going to have the same amount of grain at ISO 400 when compared to a high end camera?
Hope that makes sense. |
A dslr at the same ISO value would have considerable less noise. I've seen postings here in the forums, for example, of the rebel XT getting excellent results at ISO 1600. I've found that with the A75, I can't go over iso 100 and still get a usable photo, unless it's going to be resized quite small for the web.
|
|
|
|
08/12/2005 06:20:22 PM · #8 |
Originally posted by mattmunger:
Hmm..so therefore, ISO is relative to the camera you are currently using? There is no standard for what a certain ISO should be? |
OF course there is! Iso speed in Digital is a play on words, the real term is GAIN but in order to make comparables with tradition film, camera makers have preset "gains" that more or less resemble the "speed" of film.
so, iso100 is the same for everyone... what changes is how your camera treats it...and this is where it gets all fuzzy.... In film, a Kodak 200asa film will react the same in any camera....
So, Iso in digital is more or less a "concept" and not really an actual true mesure. |
|
|
|
08/12/2005 10:45:31 PM · #9 |
Yeah I really try and keep the ISO as low as I can with my A75. Although NeatImage and the like can really help with those photos that you have to shoot higher.
|
|
|
|
08/12/2005 10:50:40 PM · #10 |
| I couldn't quite believe my eyes when updated from Powershot S1 IS to my new Nikon D50. S1 IS was way too noise already at ISO 200, D50 gives fine pictures even with ISO 1600. And that allows you to use much shorter exposure times.. which is good. I think it's been one of the greatest improvements from P&S to DSLR. |
|
|
|
08/12/2005 11:48:17 PM · #11 |
Very short answer:
Yes there's a standard, but it's about light, not noise. P&S cameras will be much noiser than good DSLRs... as in three stops noisier or more.
Longer answer:
ISO level is a measurement of sensitivity to light, and affects the metering on a camera. It *is* a standard... though some cameras actually have broken metering (i.e. there's one Nikon that meters at approximately ISO64 when it claims to be at ISO100, for instance, but I don't remember which model).
For most settings on most cameras, higher ISO levels than the base one are achieved by boosting analog gain prior to digitization. This causes some noise to be generated, and the smaller the pixel pitch you have (e.g. the larger the number of total pixels for a given size sensor, or the smaller the sensor for a given number of pixels), the larger that noise is likely to be in absolute terms, though better cameras have better ways of mitigating this without losing detail. The highest ISO levels (e.g. ISO3200, "H" on Canon 20Ds) are often artificial, which is to say that they're a mathematical doubling of the post-digitization number only, and are only useful if you're shooting straight to jpg, not using RAW.
This is complicated by the fact that digitization is nonlinear, and that visible noise is really a matter of signal to noise ratio, not just a factor of added noise at the gain stage, but I suspect this is more than you wanted to know already. |
|
|
|
08/13/2005 10:35:43 AM · #12 |
| OK, another question. Does the physical size of the sensor have anything to do with the amount of noise? For instance, My 6mp D70 vs a 6mp P&S. Are the sensors the same physical dimensions and does this contribute to the noise factor? |
|
|
|
08/13/2005 10:54:05 AM · #13 |
Originally posted by DannyM:
OK, another question. Does the physical size of the sensor have anything to do with the amount of noise? For instance, My 6mp D70 vs a 6mp P&S. Are the sensors the same physical dimensions and does this contribute to the noise factor? |
Here is a good read concerning your question. |
|
|
|
08/13/2005 12:03:37 PM · #14 |
Nobody seems to have mentioned yet CMOS vs CCD :)
|
|
|
|
08/13/2005 12:06:50 PM · #15 |
Originally posted by riot:
Nobody seems to have mentioned yet CMOS vs CCD :) |
or JFET
|
|
|
|
08/13/2005 02:57:03 PM · #16 |
| Wow. Far more complicated than I ever imagined. But I guess since everything is digitized in a digital camera, everything must be simulated. Therefore, can't camera makers eliminate this simulation and give us crystal clear pictures at any ISO? |
|
|
|
08/13/2005 03:10:27 PM · #17 |
Originally posted by mattmunger:
Wow. Far more complicated than I ever imagined. But I guess since everything is digitized in a digital camera, everything must be simulated. Therefore, can't camera makers eliminate this simulation and give us crystal clear pictures at any ISO? |
no, however they're getting better at it!
|
|
|
|
08/13/2005 03:15:34 PM · #18 |
Originally posted by mattmunger:
Therefore, can't camera makers eliminate this simulation and give us crystal clear pictures at any ISO? |
No, the sensor is a physical device with electgricity flowing through it. Heat and random data are an inevitable consequence. The larger the physical area devoted to each pixel and the more efficient the sensor is at dissapating heat the less noise it will produce. That's why larger/more expensive sensors have less noise.
For the ISO rating to have any meaning, the manufacturer must calibrate the sensor so that under a specified set of conditions, the sensor will yield an image similar to that obtained with exposure to (say) ISO 100 film. That exposure setting will be called ISO 100 for that sensor. Then, if you want a shot at ISO 200, the signal from the sensor will be amplified (doubled) before being recorded. This process, similar to cranking up the volume control on your stereo, will make the "noise" more obvious.
The quality of the sensor ought to be described like audio recording equipment; the signal-to-noise ratio is a useful comparison. |
|
|
|
08/14/2005 05:30:39 PM · #19 |
This discussion is making me wonder why we don't see even experimental actively cooled sensor cameras. It wouldn't be especially complicated or difficult to attach a heatsink to the back of the sensor with a heatpipe going to a small refrigeration unit, maybe even something so simple as a small heatsink and fan?
I know the idea of having a fan whirring in your camera might be a bit unsettling but there's more than one way to skin a cat in this situation. Even sticking a little peltier to the back of the sensor might have dramatic effects if done right... anyone explain why this has never been tried, or if it has, show me?
|
|
|
|
08/14/2005 05:38:41 PM · #20 |
Originally posted by riot:
anyone explain why this has never been tried, or if it has, show me? |
well, this is a shot in the dark, since I don't really know anything about it, but I'd guess that sensor dust would become unmanageable if there were a constant flow of air in its vicinity. Then again, like I say, I don't know what I'm talking about. |
|
|
|
08/14/2005 06:07:14 PM · #21 |
I'm talking behind the sensor if anything, there'd be no air flowing over the sensor (indeed, that would be quite a dust issue). I'm thinking more of actively cooling the sensor from behind, one way or another.
|
|
|
|
08/14/2005 06:13:04 PM · #22 |
| Astronomical cameras are actively cooled (like an auto engine) to allow extremely long exposures. |
|
|
|
08/14/2005 06:15:55 PM · #23 |
Originally posted by riot:
I'm talking behind the sensor if anything, there'd be no air flowing over the sensor (indeed, that would be quite a dust issue). I'm thinking more of actively cooling the sensor from behind, one way or another. |
Active cooling of the sensor is often done in astrophotography; it's been on the market for years. There are potential problems associated with it though. If you cooled the sensor enough to make a really obserable difference in noise, you woudl cool it below the dew point of the ambient air. That would cause condensation on the sensor, which of course would not be good for image quality ;-)
The best alternative is to provide a direct path for heat flow from the sensor to a heatsink. The camera manufacturers definitely design with that in mind today.
|
|
|
|
08/14/2005 06:38:00 PM · #24 |
Originally posted by kirbic:
Originally posted by riot:
I'm talking behind the sensor if anything, there'd be no air flowing over the sensor (indeed, that would be quite a dust issue). I'm thinking more of actively cooling the sensor from behind, one way or another. |
Active cooling of the sensor is often done in astrophotography; it's been on the market for years. There are potential problems associated with it though. If you cooled the sensor enough to make a really obserable difference in noise, you woudl cool it below the dew point of the ambient air. That would cause condensation on the sensor, which of course would not be good for image quality ;-)
The best alternative is to provide a direct path for heat flow from the sensor to a heatsink. The camera manufacturers definitely design with that in mind today. |
Aha, so it is done :D I'll investigate further! I was just thinking that for modern dslrs, they can get to be fairly hefty things... a small active cooling system would not take much more room or mass, and would certainly have a lot of room to play around in the area between a sensor's normal temperature (with passive cooling) and dewpoint.
|
|
|
|
08/14/2005 06:42:38 PM · #25 |
The "play" between ambient temp and dewpoint is probably not enough to make a significant difference in image quality, especially when you factor in the additional noise and power drain likely to be involved.
You can try sticking you camera in a plastic bag (maybe with some dessicant packets) and keeping it in an ice chest as you drive around -- why not make some tests?
Message edited by author 2005-08-14 18:43:07. |
|
