Tentacle wrote:
Isn't this much more of a parallel problem where glass quality plays its role alongside things such as resolving power of the sensor, AA filter, noise reduction, anti-reflection coatings and all the rest?
So you're saying that to get more resolving power out of a sensor you have to increase the resolving power of the glass a corresponding amount? Or, to get noise reduction algorithms that perform better you have to stick a better lens on the camera? Consider that for a moment...
There's no correlation between sensor and glass performance, but there is an interaction between them. Meaning if you increase sensor performance, but glass performance doesn't increase, overall IQ increases slightly. However, if you increase the performance of both glass and sensor, and perhaps other components, overall performance increases exponentially.
Simply stated, if you stick an 85 1.2L II on a Digital Rebel 300D and shoot a photo, it will be worse than the same photo shot through the same lens using a 1Ds2. Further, given the ability of a camera like an SD700 to produce an image approching the quality of 30D with a smaller sensor and more compromised glass, it' a safe assumption that the L glass on a DSLR is not the bottleneck to overall IQ performance.
(in other words, we are in aggressive agreement on the subject)
But erhh, the thing is, I was approaching this from the digicompact side, since this sprouted from the G7 and mentioned S70 and SD700IS. Looking at it from that angle, glass would not not be a problem untill all others were solved, it would become one before that point already. A properly engineerd package will have all components "good enough" to have the desired performance. You don't design an Über-lens to go with a poor performing sensor, that'd be a waste of resources.
So, there is a (design/engineering) correlation between the two if you're talking about a digicompact. You were thinking dSLR and L-glass. OOPS, that's a different ballgame. From that angle, you're right too.
No, I understand your point completely, I'm saying that even given the compromised design of a compact point and shoot, the lens is still far better than the other systems in the camera.
Let me explain it this way: I'd be willing to bet Canon could hold the lens design constant in any given Powershot model and improve the other systems 1 or 2 generations and still not outperform the lens design. Another good example of this is comparing a Nikon D80 to a Sony Alpha. Both cameras use the same sensor, but because of the more aggressive noise reduction algorithms in the Nikon, it has less noise but appears softer out of the camera than the Alpha.
I think the same thing happens with current P&S cameras. Their softness in the image is not so much a function of the glass as it is a function of the noise reduction required for their small sensors.
My point with DSLR's is that the same assumption holds. The current crop of DSLRs is nowhere near capable of outperforming current L glass.
3.9 mpixel Powershot G3: f/2-3 with 35-140 mm eq.
5 mpixel Powershot G5: f/2-3 with 35-140 mm eq.
7.1 mpixel Powershot G6: f/2-3 with 35-140 mm eq.
All on a 1/1.8" CCD. (And then the G7: f2.8/4.8 with 35-210 mm eq...)
Here's the "but": There's no telling if Canon saw fit to re-specify tolerances, quality checks, glass quality, coatings etc. Even if the actual design of lens elements and groups stays the same. Part of it remains unknown.
Another thing to ponder: Some L-primes stay tack sharp when you add a 2x teleconverter. That means that the resolving power of the lens is at least twice as that of the (current) sensors. Other L-primes don't fare so well and lose some sharpness which hints that some lenses are only (oh, the pain!! ) at less than twice the resolving power of the current best sensors.
On a different note: I hope this isn't too off-topic for a Photokina speculation topic
For those who are claiming that glass is nowhere being a limiting factor, could I respectfully suggest that you take a look at a few image corners across Canon's lens range. L glass is already limiting the sensor - not the other way around.
I predict a sharp increase in medium format digital over the next two years as the traditional benefits of MF are re-discovered (and at an ever-more affordable price).
There are several reasons the small aka "digicams" produce good results even with their built in lenses.
1) The small sensor and corresponding short focal length lenses have huge DOF hence small errors in focus dont affect image quality at all.
2) The lack of a mirror inherent in SLR's combined with the short focal lengths used minimize the effect of user induced vibration hence, good sharpness even with less than perfect technique.
Sneakyracer wrote:
There are several reasons the small aka "digicams" produce good results even with their built in lenses.
1) The small sensor and corresponding short focal length lenses have huge DOF hence small errors in focus dont affect image quality at all.
2) The lack of a mirror inherent in SLR's combined with the short focal lengths used minimize the effect of user induced vibration hence, good sharpness even with less than perfect technique.
Precisely why I just bought my son and daughter in law a new Canon Powershot A640 to take pictures of my new grandson.
i will give you the corners issue.. that is a 35mm design issue.. But lik e a tennis racquet, the sweet spot of the lens is nowhere near being strained.
someone should test these lens to full resolution... obviously this would not involve a camera but would involve some other glass....
KIDERAL wrote:
i will give you the corners issue.. that is a 35mm design issue.. But lik e a tennis racquet, the sweet spot of the lens is nowhere near being strained.
someone should test these lens to full resolution... obviously this would not involve a camera but would involve some other glass....
You hit it on the head. The corners issue was around with film too, this isn't new news.
foto-z wrote:
For those who are claiming that glass is nowhere being a limiting factor, could I respectfully suggest that you take a look at a few image corners across Canon's lens range. L glass is already limiting the sensor - not the other way around.
I predict a sharp increase in medium format digital over the next two years as the traditional benefits of MF are re-discovered (and at an ever-more affordable price).
DSLRs are the new medium format.
Medium format is the new large format.
dazedgonebye wrote:
DSLRs are the new medium format.
Medium format is the new large format.
Must respectfully disagree. The advantages of LF were/are far more than film size. The ability to swing/tilt/rise and fall on both the front and rear standards were often far more critical to creating a beautiful image than was the increase in film size. Until we get truly combine digital with the movements available on a view camera (at an affordable price) we will be giving up a great deal in terms of our ultimate creative opportunities.
That said, there are lots of situations where a view camera is just useless.
jwil wrote:
You hit it on the head. The corners issue was around with film too, this isn't new news.
The vignetting issue was, and is, the same for 35 mm film and digital. The edge sharpness issue is worse on FF sensors than on film because film is more forgiving to light striking it at an angle.
Leica tackled this problem with the new M8 rangefinder digital camera: instead of placing all microlenses directly over the photodiode they are gradually offset as you get closer to the edge of the frame. See http://www.dpreview.com/articles/leicam8/
So all the problems inherent to full frame performance on 35 mm digital, can be compensated in some cases. I don't see the microlens offset trick be as effective in dSLRs though since the position of the rear lens element isn't fixed, whereas on a rangefinder camera it's very much up close.
Dream of a rumour, courtesy of Aussie Geof:
'For what it's worth I heard 40D:
* Still 5 FPS - plenty anyway
* Digic III - for lower noise and higher ISO 3200+ as well as better dynamic range (16 bits likely)
* Better AF - more points and faster (courtesy Digic III)
* Anti-Dust,
* CMOS Sensor etc. Interestingly this was either an 'improved' version of the 10.1mp in the 400D or a 12mp one they were testing - but probably reserved for the 'next generation' (50D I guess)
* Still 1.6 Crop
I also heard that Canon are going to split the 5D into two models to test our the demand for prosumer full frame without compromising demand for the 1.6 crop. There was going to be a smaller (closer to 30D size) and cheaper version of the 5D - same sensor, Digic III, wider AF points (given full frame) and probably Anti-Dust (some have suggested this as a 7D). And a upgraded 5D wiith a similar size to the current 5D but 16.2mp (or so) sensor ('very similar' to the 1Ds Mark II) and the samle Digic III, wider AF points, 5 FPS etc.
My source - who visited Canon Japan a couple of months ago (and gave me a very accurate preview of the 400D) said they basically had replacements for their full EOS range in the works and that 'everything would be replaced by February / March 2007'. He said Canon were very excited about the combinaiton of new sensors and Digic III giving better dynamic range and 'much better' high ISO performance.' http://forums.dpreview.com/forums/read.asp?forum=1019&message=20067035
Dunno what they mean by 'better DR' - if they are just talking about better tonal gradations by using more bits to represent the range, or if they have managed to fundamentally improve the sensor to give more DR like the Fuji S3.
I am also a bit wary of talk of improved noise via processing - could mean a hideous mush due to compulsory NR.
The extra bits and improved AF do not surprise me - that after all is the point of having more processing power!
Regards,
DaveMart
Now that Tentacle took up Leica M8 here: They say that they chose 1.33x crop sensor for 3 reasons, one of which was the prohibitive cost of a FF sensor (for a >$4500 camera!). At least here the info of the cost benefit of APS-H format in Canon's white paper seems to apply.
It will be very interesting to see which of Canon's contradictory messages about Canon's next $4000-$5000 camera will prove correct: Is it the sensor cost/Nikon competition based APS-H message or is it the future vision/marketing intention based FF message?
Dunno what they mean by 'better DR' - if they are just talking about better tonal gradations by using more bits to represent the range, or if they have managed to fundamentally improve the sensor to give more DR like the Fuji S3.
I am also a bit wary of talk of improved noise via processing - could mean a hideous mush due to compulsory NR. The extra bits and improved AF do not surprise me - that after all is the point of having more processing power!
There's NR and there's NR. Read the Canon white paper on FF for details. Canon starts to tackle the problem on-chip: CDS (Correlated Double Sampling) reduces fixed-pattern noise and Complete Charge Transfer Technology reduces the random noise. It's a matter of measuring residual charge and using that value to compensate the output value. Both mechanisms do so without averaging/smearing adjecent pixels. Improving these will not get more mush.
Then there's noise prevention by reducing semiconductor crystal defects. Hmmm, wait, Canon is making its own scanners and steppers, right? So they are in a position to improve chip quality here too. I don't expect Canon to make it's own wafers, so they do rely on third parties, but here too there are continuous improvements in silicon purity.
As for DR? Again, the white paper suggests there is room for real DR improvement on-chip, not just finer stepping by better A/D converters. This is without simply going to bigger pixels, which would mean more photodiode area, and thus more capacity to store charge. Going bigger is the easiest way to increase real DR, but that's not really an option in the drive for more mpixel, if you want to stay within the existing sensor format.
(And that's why I think/expect APS-H not to disappear too soon.)
I am not at all surprised about the 40D info. The AF upgrade should not come as a surprise to all: http://www.dpreview.com/reviews/canoneos30d/
We see the D30/D60 had 3 pt AF, 10D had 7 pt AF, 20D/30D had 9 pt. So, Canon is constantly upgrading their AF unit.
It remains to be seen whether Canon will implement an auto ISO mode and weather sealing (like the D200 and K10D).
thw2 wrote:
I am not at all surprised about the 40D info. The AF upgrade should not come as a surprise to all: http://www.dpreview.com/reviews/canoneos30d/
We see the D30/D60 had 3 pt AF, 10D had 7 pt AF, 20D/30D had 9 pt. So, Canon is constantly upgrading their AF unit.
It remains to be seen whether Canon will implement an auto ISO mode and weather sealing (like the D200 and K10D).
The AF upgrade is more or less of a given, I would have thought, since the 400D uses that from the 30D.At least to the level of the 5D, but likely IMO to a different system as hinted in the link I gave to take advantage of the power of DIGIC 3
How I would like a better position of AF points!
Check out the arrangement on the Pentax K10D, and the number of cross-sensors.
Judging by the noises form Chuck on the article about Canon futures, where he discussed a move to ISO 6400 and future photograhy without flash but just going to higher ISO, I think that auto-ISO will indeed be a feature of the next generation of Canons.
Let's hope they don't bury it in the menus.
I'll have to re-read that article on Nu-core, when I am feeling more intelligent! ;-)
I think Aussie Geof's suggestions sound very realistic and might be one of the most probable rumour I've read in the past few days... (or even weeks... )
The idea of splitting the 5D into two different models sounds also rational for the reasons mentioned and it would also close the gap between the current 5D and 30D.
) at less than twice the resolving power of the current best sensors.
)