Tim Speciale wrote:
You guys have no idea what cameras are capable of and where we'll be in a few years.
Perhaps not, but from reading all the posts here and other forums, I sort of get the general idea where people want those cameras to be in a few years. If Canon wanna keep their sales up, they better have similar idea.
CUclimber wrote:
Just wait until you hear RED's product info on November 13th.
Unless it's about RED making a DSLR, it's probably irrelevant. If they are, it'll be interesting to see what they'll do to penetrate this oligopoly market.
10MP-12MP is all I need - could always use more DR though.
For my printing needs, I have not felt the need to complain yet .... and I have made 24"x36" prints with the old faithful 11MP 1Ds that I am very happy with. Even the older Rebel XTi has been quite capable of producing 24"x36" prints.
What was Canon thinking when they decided to put a 14-15MP sensor on not just the 50D, but also the G10 and the model 990-something P&S.
chez wrote:
Come on Tim, don't leave us hanging...enlighten us with your wisdom.
I remember when a 750mhz computer was unreal. It was the same feeling i have towards a 15mp camera. The digital camera is not even a decade old and you people expect them to stay at the current sensor size.
Tentacle wrote:
If the 50D has better shadow noise performance, then the DR gets extended only at the dark side, not at the highlight side. But still it could have a better overall dynamic range.
The point is the same. The per-pixel dynamic range is the sensor's dynamic range. You can't get extra dynamic range by combining pixels. This is true of shadows as much as highlights-- in a resulting image either the shadows will be blocked up or random noise when the pixel's DR has been exceeded, no?
n0b0 wrote:
Perhaps not, but from reading all the posts here and other forums, I sort of get the general idea where people want those cameras to be in a few years. If Canon wanna keep their sales up, they better have similar idea.
Not likely. Most consumers are just drones. Ask anyone in marketing...
Tim Speciale wrote:
I remember when a 750mhz computer was unreal. It was the same feeling i have towards a 15mp camera. The digital camera is not even a decade old and you people expect them to stay at the current sensor size.
Car's been around for more than a hundred years and only recently did manufacturers really start researching into alternative fuel source. My point is, sensor size will stay the same unless it's in the manufacturers' best interest to enlarge them.
Just as I cannot say that sensor size will stay the same with any degree of certainty, you also can't say that they won't stay the same, at least in the foreseeable future.
The point is the same. The per-pixel dynamic range is the sensor's dynamic range. You can't get extra dynamic range by combining pixels. This is true of shadows as much as highlights-- in a resulting image either the shadows will be blocked up or random noise when the pixel's DR has been exceeded, no?
The bright end of the DR is determined by the saturation of photo sites. The dark end of the DR isn't determined by "no charge" but by the darkest recognisable signal above background noise. If you manage to lower background noise you will detect smaller signals above the background, hence extending DR.
It's also the reason why DR is (or ought to be) linkt to quality. DR at high quality (a higher S/N ratio for the lowest signal) is smaller than DR at low quality, where the the smallest non-dark signal rises less above the background noise.
Some files are a bit underexposed. But there is a ISO 800 image of some bottles with red caps which shows visible banding in its shadows (upper right corner). There are also ISO 1600 shots of swans against a dark background which show a very limited DR IMO. Heavy noise/blotches/banding in the shadow areas and the whites are (almost) clipped.
BTW I converted the files with ACR 4.6 - don't know if DPP does a better job.
Tentacle wrote:
The bright end of the DR is determined by the saturation of photo sites. The dark end of the DR isn't determined by "no charge" but by the darkest recognisable signal above background noise. If you manage to lower background noise you will detect smaller signals above the background, hence extending DR.
Yep. But the point is STILL the same. If a pixel's DR ends at a certain point, combining the pixel with others won't raise it somehow. If a pixel's signal is indistinguishable from background noise, the same will be true of any four such pixels taken together, and so on.
Tentacle wrote:
But where did pixel binning get into the mix?? The 50D at and above ISO1600 performs better than the 40D without pixel binning.
I honestly don't know how, if at all, pixel binning influences S/N ratio.
It got into the mix with brainiac's comment about measuring dynamic range per image, instead of per pixel. I never said the 50D had worse DR than the 40D or anything like that. My point was just that the dynamic range of the individual pixels and the whole sensor is the same. (Which means that while a camera may exceed the DR of a prior camera with older technology, in general denser-packed pixels lead inescapably to a loss of DR.)
jvarszegi wrote:
It got into the mix with brainiac's comment about measuring dynamic range per image, instead of per pixel. I never said the 50D had worse DR than the 40D or anything like that. My point was just that the dynamic range of the individual pixels and the whole sensor is the same. (Which means that while a camera may exceed the DR of a prior camera with older technology, in general denser-packed pixels lead inescapably to a loss of DR.)
I am not sure that I believe this. Here are two versions of a file. One was cast to bitmap, and then unbitmapped and interpolated. The other is straight. I see increased noise, and much worse tone, but no less overall DR. I don't know if this is relevant. http://cyberphotographer.com/1ds3/vamp.jpg http://cyberphotographer.com/1ds3/vampbmp.jpg
It seems to me that there is a very direct relationship between resolution and tone. In the case of film, for instance, a tonal image is built up of very very small binary bits. Obviously, diffusion dither and interpolation are involved in the case above, but what is a Bayer mosaic but an interpolation? It's not entirely clear to me that pixel DR and image DR are exactly the same thing when factors like noise, Bayer interpolation, colour filters, and so on, are involved.
parents give me when they hear my "professional" camera with that big white lens is ONLY 4 MP (1D Classic)... 
If a pixel's DR ends at a certain point, combining the pixel with others won't raise it somehow. If a pixel's signal is indistinguishable from background noise, the same will be true of any four such pixels taken together, and so on.