Look, as for Nikon vs. Canon and MP, I personally feel like even 10MP is too much for an APS-C camera if you're primarily a high ISO shooter, which is why I elected to not go for the D300 when I made the switch - 12MP is definitely too much for what I do. But look at what Nikon did when they went full-frame - they didn't try to one-up Canon with a 15MP camera, they chose to simply match the 5D in MP and make the high ISO performance as good as possible, this was a quality-first decision and bucks the MP war trend.
But this is all irrelevant. The proof is in the pudding - the 50D's extra resolution adds essentially nothing to the camera due to issues with noise. They rushed it out the door 6 months earlier than they have for the past 5 years or so. Anyone who thinks this is anything other than a desperate attempt to maintain their market share is fooling themselves.
Mike Farren wrote:
Why else would Canon make a camera?!?
Well, no shit sherlock - the point is that they're compromising quality to simply appease the numbers-driven portion of the market. If you're okay with that, more power to you. But I'm more interested in supporting a manufacturer that's putting quality first.
Tentacle wrote:
Then why do both AMD and Intel give their processors rating numbers now?? I recently got myself an Intel Core Quad Q9550. Or I could have gotten an AMD Phenom X4 9950, had I gone with the other platform... They don't advertise in clockspeed anymore.
Both companies stopped optimising for speed and opted for optimising for instructions per clockcycle. Of course they still marginally improve on clock speed too, but the biggest gain is in better IPC.
Maybe you have hit on the key here....AMD and Intel are no longer marketing their products based on speed even though speed continues to be an issue. Maybe cameras needs to be marketed using some metric other than Mp.
Who knows why they choose that? marketing? The old AMD Athlon proved that higher clock speed doesn't mean better performance so I supposed it's logical for them to move away from advertising clock speed as the mark of performance.
What do they really improve in their newer chips apart from increasing clock speed, FSB and cache anyway?
At the time, higher-up management at Intel forced the decision that the new generation of processor (which became the ill-fated Pentium 4) would have sacrifice some IPC for higher clock speeds. Do less each cycle, do more cycles. That was a pure managerial marketing driven decision, one that was opposed by their technical engineers, and it was purely based on the public perception that higher clockspeed = better.
That was also the moment that AMD switched to rating numbers. Eventually, Intel had to abandon the really high clock speeds, and move to rating numbers too. Their new generation of processors was faster, but had a lower clock speed.
As far as what manufacturers do to improve a CPU? Processor architecture is complicated at best. The station of simply directly handling x86 instructions is long gone. They get decoded, translated into sub-instructions, those get re-arranged, fed into one of several pipelines and then results get re-ordered into the proper sequence and retired. There are instructions to deal with IF/ELSE instructions, there is stuff to predict memory access and cache data in advance, and a whole slew of clever tricks. The number of pipelines and the number of stages in a pipeline can greatly affect the work a CPU can do in a single cycle. There is much more to it than just FSB and cache.
How many people on this thread have actually been using a 21 megapixel camera for the last year?
The G10 technology would give us full frame images heading towards 300 megapixels. The limit here is processing power and storage, both inside the camera, and on the desktop. Computers have hit a brick wall and are holding imaging back, at the current rate of development. Lenses also impose a limit. There is an economic and physical limit to this, but 21 megapixels is a great facility, both for large prints, and for cropping. Don't forget, that cropping allows you to take fewer lenses, as long as your lenses are really good.
Then there is the issue of print size. You can see the full benefit of 21 megapixels on Epson A3. A 44 inch wide Epson banner printer now costs about $5000, well within the gear budget of most working photographers. The limit here is that customers seldom want 40x70 inch prints, but a 21 megapixel file looks great at that size, and a 30 megapixel image taken with an L prime at f8 will look even better.
The sRaw1 and sRaw2 formats mean that the high megapixel spec of a sensor is a facility that is useful to the professional in certain circumstances, not a mandatory file size.
People, even revered reviewers like DPR and LL, are making the mistake of comparing 100% crops between 12 and 21 megapixel files. That always favours the lower megapixel file because it is far less magnified. As a result, the benefit of 21 megapixels over 12 is being underestimated.
I say, bring on 30 megapixel. There's no evidence that Canon has been compromising on dynamic range to achieve high pixel counts, so why not? It's hard to tell a 1D3 and 1Ds3 apart, in terms of dynamic range and noise. And a D700 seems to have less dynamic range than a 1Ds3, not more.
Nobody has to buy or use 21 or 30 megapixels, so what's the problem? Stick with your current camera, or buy an old one. Obsolete cameras have never been so good and so cheap.
brainiac wrote:
So where does that leave the megapixel war, and this tenacious fallacy that more megapixels involves some kind of pact with the devil?
I think the tenacious fallacy (what does that mean anyway?) is geared towards recent developments in the world of 1.6x crop cams (think 50D) rather than 21MP FF...
Daan B wrote:
I think the tenacious fallacy (what does that mean anyway?) is geared towards recent developments in the world of 1.6x crop cams (think 50D) rather than 21MP FF...
From wikipedia:
"A fallacy is a component of an argument which, being demonstrably flawed in its logic or form, renders the argument invalid in whole."
From m-w.com:
Tenacious: tending to adhere or cling especially to another substance.
It's a myth that just won't go away no matter how many times it turns out to be unfounded.
I've seen no evidence that the 50D has less dynamic range or worse noise PER IMAGE than a 40D. Every time someone cites 50D inferiority it turns out to be a link to a comparison of 100% crops, which is METHODOLOGICALLY WRONG.
I'm a sports shooter and have always been happy with my 1DMkII's but I borrowed a 1DsMkII for an event last month and with a 400/2.8 IS the IQ was amazing. The brightness of the viewfinder also blew me away - it was as if someone had switched the light on. The only problem was the buffer & fps. I know I missed some great shots whist waiting for the buffer to clear.
I'm now longing for a 20+ megapixel camera with 10 fps and a pair of Digic IV's. I guess one day it's coming but who knows whether a pact with the devil will be required or not.
brainiac wrote:
The limit here is processing power and storage, both inside the camera, and on the desktop. Computers have hit a brick wall and are holding imaging back, at the current rate of development.
I don't think so my friend. I know for a fact that there are digital painters who uses 9000 x 9000 canvas size for their matte painting. That's 81MP being manipulated in real time. If you have the money, there are workstations that can even handle playing 4k resolution videos. Desktop computers have certainly not hit any wall and holding imaging back.
If there's any limiting factor, I'd say it's in the software. Many of them probably still aren't optimised for multicore/multi processor and don't take advantage of 64bit hardware which limit RAM usage.
It always amazes how many people on Fred Miranda queue up to criticize "pixel peepers" and yet so many of the same people go on to demand ever higher mega pixel counts!
I think they have a long, long way to go on lenses before they can up the pixel density. 4.5 MP/cm2 is already way more than enough.
I can see canon upping their APS-H and FF camera lines to that density... but going past 4.5 is isn't really going to make for better photos. But they do need to keep coming out with new bodies, and they need to have reasons for people to want to buy them...
there is a lot of room to go on high ISO sensitivity, noise reduction, and other camera features.
Although, realistically, you'de be better served buying a used 1D2 in good shape than a 50D...
I have a 1D classic for my main camera, and my back up camera is a 20D. I like the small file size of the 1D, because I can shoot in RAW and one day's worth of sports events doesn't take up my whole hard drive. And I can use the 8fps without guilt of filling up my card instantly
It seems that pushing one envelope will spill over into advancing others. For example: if the 21+MP camera start out-resolving the lenses they'll have to improve upon the resolving power of the lenses. Pushing limits like this leads to discoveries and breakthroughs that help in other areas yet untested. I, for one, hope no manufacturer stops pushing further with each new camera released.
If a pixel is blown, it's blown, no? Combining it with other blown pixels will not create information.
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.
n0b0 wrote:
I'm hoping that once they start to outresolve lenses, Canon would enlarge the sensor size of their crop bodies to around 1.3 to 1.5 of the FF.
Back when the 5D was introduced, Canon already stated its intention to move to full frame for everything except entry-level dSLR.
Given the rumour that Canon has developed lithography equipment that can produce a chip the size of a full frame sensor in a single exposure pass (so without having to align multiple lithographic exposures per layer, the so-called 'field stitching') that shouldn't be too far off. Doing an FF sensor without field stitching dramatically cuts production costs.
