I have to chuckle that a few of the same people who are saying "wow! this camera is a whole new benchmark in dslrs" are also saying "how can Canon be so stupid!".
I think both are somewhat untrue! Apart from the huge increase in processing power, which as Ziggy has explained makes the 1DX more reliable and more responsive, this camera doesn't do any more than what the 5DII and 7D do individually.
But it is clearly a different camera to a simple hybrid of those two cameras, even if such a thing were possible. In Canon's launch media they emphatically draw attention to the 1DX's versatility, and this is something that attracts a number of responders to this thread. But Canon even more emphatically draw attention to the 1DX's advantages for the sports photographer - reliability, responsiveness, speed, high ISO and video. Reach is critical for the sports photographer, but not as critical as for the wildlifers, for reasons such as the relative closeness of the subject, and the relative predictability of the subject's position within a known limited space. Therefore much cropping is also less likely to be needed. Sporters' shots are mainly action, they need fast shutter speeds-high ISO and fps which the 1DX has, while wildlifers' shots have a lower proportion of action. Wildlifers typically don't use video, whereas for sporters it's a high value addition to their reporting tools. In addition, and yet another difference, is that sporters typically shoot in a much greater range of light levels than wildlifers, from daylight to twilight to artificial, therefore bigger pixels and high ISO are very desirable.
Why are the wildlifers bellyaching that this camera is not for them?! It is clearly NOT MEANT for them! Does that make Canon mad? Has Canon abandoned wildlifers? I don't see the logic of such ideas. Nor of a belief that a camera which will take the 1DIV another notch is no longer a possibility because of the 1DX. Seems like hysterical "sibling rivalry" to me!lust
Sign me up. I've been waiting for the an update to the 1D series and this is it - It'll make a great wedding anniversary present to myself, oops... I mean for my wife. I still love my 5D mII but now can use that for video and this for photos. Thank you Canon.
OK, there is a lot of speculation in what I'm about to say, but I think the reason for the resolution being 18MP may be video related.
One of the things Canon has touted about the new sensor is reduced moire, this is one of the main problems with DSLR video at the moment. It is caused by several factors, but one of them is that the pixels on the sensor do not translate directly to the 1080P frame.
If the sensor was designed with this in mind, for example 3x the horizontal and vertical resolution of the 1080p frame, with exactly 9 pixels to each video pixel, this might result in a sharper image with less moire.
Depending on various factors, this would result in a still 3x2 frame size of approx. 18MP.
I get 15.7mp by my math, did I do something wrong?
1080 x 3 = 3240 vertical pixels
3240 x 3 / 2 = 4860 horizontal pixels
3240 x 4860 = 15.7mp
1DX is 3456 x 5184, so I don't see how that fits your theory. (interestingly that is the same pixel dimensions as the 7D)
-Jack
An "accurate" reproduction of a scene and a good photograph are often two different things.
No you have to divide twice because the crop factor applies to both vertical and horizontal dimensions. That is, a FF sensor is 1.3x wider and 1.3x taller than an APS-H sensor.
So, 18mp/1.3/1.3 = 10.65mp. That's just a smidge more than the 1DIII.
This is also why the 5DII has the same pixel size as the 20D. 21/1.6/1.6 = 8.2mp.
-Jack
An "accurate" reproduction of a scene and a good photograph are often two different things.
This has been an interesting discussion on pixels and crops; sure has given me some insight. What I am really wondering about is the relationship between pixel size (i.e. 5uM, 7uM, etc.) and IQ. I can appreciate that fewer pixels means less resolution, all other things being equal. And that further, a single pixel just registers a triad of numbers (RGB) that in itself has nothing to do with resolution. However, the accuracy of those numbers does affect IQ. For example, under low light situations, noise is reduced by various algorithms that use adjacent pixels to help determine what numbers are correct and what numbers are spurious due to noise. So in this situation, I can imagine that a FF sensor with good noise characteristics might produce better images that a sensor with a higher pixel density but worse noise problems. (I have written my own denoising software, so I know this to be the case.) And certainly, comparing my ISO=3200 images from my FF 5D2 to my 1.3 CF 1D3, the images are much better from the former even after a 55% crop.
It also seems to me, subjectively at least, that this is true at ISO=100 as well, but it is not clear to me why this should be true. I have done similar comparisons between my 5D2 and my 1D4 as well, with less obvious differences, though I still think the 5D2 gives me better IQ. Because of this, I put up with a lot of the 5D2's limitations as a sports camera and use it all the time to shoot pro sports events, along with my 1D4.
So in thinking about the 1Dx (especially since I am at the top of the waiting list at Calumet) and what it has to offer, I am not sure that there is much advantage in the ISO range I work at (up to 6400 in some stadiums, which is OK on the 5D2 and 1D4). My lenses give me enough reach for almost all shots, save perhaps those in the far outfield, and the bump in FPS I'm sure could be useful but not often for the way I shoot.
I'm sure there are pixel peepers out there who have done all these studies. I have seen a few, but not sure how well they translate into overall IQ. Maybe I should just rent the 1Dx and take it for a test spin before plunking down the $$$...
This has been an interesting discussion on pixels and crops; sure has given me some insight. What I am really wondering about is the relationship between pixel size (i.e. 5uM, 7uM, etc.) and IQ. I can appreciate that fewer pixels means less resolution, all other things being equal. And that further, a single pixel just registers a triad of numbers (RGB) that in itself has nothing to do with resolution. However, the accuracy of those numbers does affect IQ. For example, under low light situations, noise is reduced by various algorithms that use adjacent pixels to help determine what numbers are correct and what numbers are spurious due to noise. So in this situation, I can imagine that a FF sensor with good noise characteristics might produce better images that a sensor with a higher pixel density but worse noise problems. (I have written my own denoising software, so I know this to be the case.) And certainly, comparing my ISO=3200 images from my FF 5D2 to my 1.3 CF 1D3, the images are much better from the former even after a 55% crop.
It also seems to me, subjectively at least, that this is true at ISO=100 as well, but it is not clear to me why this should be true. I have done similar comparisons between my 5D2 and my 1D4 as well, with less obvious differences, though I still think the 5D2 gives me better IQ. Because of this, I put up with a lot of the 5D2's limitations as a sports camera and use it all the time to shoot pro sports events, along with my 1D4.
So in thinking about the 1Dx (especially since I am at the top of the waiting list at Calumet) and what it has to offer, I am not sure that there is much advantage in the ISO range I work at (up to 6400 in some stadiums, which is OK on the 5D2 and 1D4). My lenses give me enough reach for almost all shots, save perhaps those in the far outfield, and the bump in FPS I'm sure could be useful but not often for the way I shoot.
I'm sure there are pixel peepers out there who have done all these studies. I have seen a few, but not sure how well they translate into overall IQ. Maybe I should just rent the 1Dx and take it for a test spin before plunking down the $$$...
The differences between the Canon 1D MKIII and the Canon 5D MKII, with regard to random/high-ISO sensor noise are more than just FF versus crop, although that certainly plays a part.
The total concept of "random sensor noise", and the noise level that we see at the in-camera image level (RAW, JPG, TIFF mostly) has to do with the sensor itself, the sensor design (CMOS, CCD, LiveMOS, etc.), distance from sensor to LNA (low noise amplifier), internal shielding, LNA performance, signal bias voltage, analog to digital converter (ADC) accuracy and efficiency, image bit depth, image processor, demosaicing algorithm (type and accuracy), image file storage type, etc.
Trying to separate any of these underlying issues from the total system in a camera is beyond the abilities of most camera users, so it's best to assume that we should limit ourselves to the "output" efficiencies of each camera and the accumulative effect of the system in total.
Assuming that all other parts of the system in a camera are equal, and looking at just the pixel density as the only variable, it's easy to understand the reason why more pixel density results in more random sensor noise (and conversely lower pixel density results in less random sensor noise).
Lower pixel density tends to equalize the random nature of sensor noise because you are taking a single sample of a larger site. That single data access point is accumulating data from a larger sample and the larger sample is using a natural type of normalization (averaging) of the randomness of the media. The same thing occurs with longer time duration sampling (longer time exposures) and/or more samples normalized together (multiple short exposures combined into a single image). All of these mechanisms and techniques will normalize/average the input data resulting in less randomness of the data, thus less random sensor noise.
When you sample a smaller pixel site there is less opportunity for averaging the data, and a greater likelihood that random noise values will have a greater influence on the measured values at that photosite. You can compensate with either longer exposures and/or multiple exposures combined (when appropriate).
Software noise reduction, including the camera's internal firmware noise reduction, uses various different algorithms to try to achieve similar results. (Basically the best software tries to separate random "and" predictable noise from image detail, and then the software uses normalization against the noise component.)
More modern cameras even seem to use a type of image masking, biased towards shadow tones, to help reduce the noise that's more visible in the shadows (both chrominance and luminance noise). This level of processing occurs before the RAW file is generated and the user has no control over the masking or degree of NR, making the resulting RAWs much better with regard to both visible and measurable noise and operationally transparent to the user.
Assuming that all other parts of the system in a camera are equal, and looking at just the pixel density as the only variable, it's easy to understand the reason why more pixel density results in more random sensor noise (and conversely lower pixel density results in less random sensor noise).
Lower pixel density tends to equalize the random nature of sensor noise because you are taking a single sample of a larger site
It's interesting that the 1D3 has the largest pixel size (7.2uM) but the worst noise of my three bodies. I guess this is where the 'all other things being equal' assumption fails. Random noise (and I assume you are referring to thermal noise, not shot noise, as the latter is not likely to be the case) can be introduced anywhere in the system. true pixel-level thermal noise is relatively easy to clean up as it appears as pixel-level aberrations. Banding and noisy patches are the hardest to fix, as large amounts of the picture are affected. I believe the latter is due to cross-talk in the electronics, or other thermal properties in the route between the pixels and the A-to-D conversion. In comparing the noise on my 5D2 to my 1D3, I find roughly the same amount of pixel noise on the former, but not much noise from the other components. Part of this might be due to the relatively lower cross talk in the FF sensor circuitry because of the larger spacing. In any case, part of what I incorporated into the denoising code I wrote was the ability to apply noise reduction in a weighted fashion to various parts of the sensor, which takes into account the inhomogeneous distribution of noise on most sensors. IOW, more noise reduction is applied to sensor sectors that are inherently more noisy.
Another factor to consider is the effect of noise on AF. We all know that the speed and accuracy of AF is proportional to light level: low light is slower (or nonexistent), bright light is faster. If the 1Dx follows this pattern (and they claim it does) then overall IQ might actually be better in tight crops that wildlife shooters routinely do than in a sensor with more pixels. Obviously the advantage is inversely proportional to light levels; but in my experience, my crops are far more often limited by small AF errors than absolute resolution. YMMV.
If you graph pixel density and noise over time (sensor generations), pixel density goes up and noise goes down, generally. Cameras of later generations have better noise and higher pixel density than earlier generations. The 1DX is especially interesting because it breaks this trend. So we will see a sensor that is of the latest generation but with less pixel density. It will hopefully mean significant improvements in noise and DR. They did this with the G series a couple years ago.
-Jack
An "accurate" reproduction of a scene and a good photograph are often two different things.
If you graph pixel density and noise over time (sensor generations), pixel density goes up and noise goes down, generally. Cameras of later generations have better noise and higher pixel density than earlier generations. The 1DX is especially interesting because it breaks this trend. So we will see a sensor that is of the latest generation but with less pixel density. It will hopefully mean significant improvements in noise and DR. They did this with the G series a couple years ago.
That's what I am hoping... I am glad to see that Canon broke this trend, and am very excited to see what this means ITO IQ and other matters.
That's what I am hoping... I am glad to see that Canon broke this trend, and am very excited to see what this means ITO IQ and other matters.
Me too. A lot of people seem to have not realized or have lost sight of that. (my frustrations are also fueled by the bench racers in the dpreview forums)
Furthermore, if you graph pixel density vs noise for the same sensor generation, I believe generally there is an inverse relationship. i.e., D7000 vs 7D, D700 vs 5DII, etc.
-Jack
An "accurate" reproduction of a scene and a good photograph are often two different things.
Data values are greater per unit area with greater pixel density. Read noise is increased by sum of pixels, binning reduces it but decreases effective resolution. Local contrast, DR and signal-noise ratio is kept relatively low.
Signal to noise ratio is greater per unit area with larger pixel size. Initial resolution is relatively low, but binning NR threshold can be relaxed, maintaining a relatively high resolution, and local contrast and DR remain relatively high.
Lens CoC affects local contrast, too.
Overall, given equal Av & Tv & focus & ISO & lens & stability of gear, for cropping purposes larger pixels give better IQ, providing that in camera algorithms are taking the greatest advantage of signal over noise.
SNR can be a mathematical value, or a subjective value. Seems to me that what you guys are talking about is a subjective value, ie eyeballing. In that case, post processing then becomes a relevant factor as well, eg how the RAW was converted. For instance, Lr's defaults enhance without you ever touching a slider, and what Lr does might not be comparable to what ACR does, etc. On top of that there is the quality of the display you are using to make judgements about IQ. In short, any variation in any post processing and viewing factor will affect subjective IQ assessments.
Viewing prints is a different viewing environment again, and has it's own factors affecting IQ.
A very complex dynamic, and so subjective rating of IQ has a very limited relationship to the underlying tech.
I get 15.7mp by my math, did I do something wrong?
1080 x 3 = 3240 vertical pixels
3240 x 3 / 2 = 4860 horizontal pixels
3240 x 4860 = 15.7mp
1DX is 3456 x 5184, so I don't see how that fits your theory. (interestingly that is the same pixel dimensions as the 7D)
Video is 16x9, not 3x2. Obviously in video mode the rectangle that is recorded is a different shape.
It doesn't seem to be clear exactly which pixels are used for the video image, but it might even be possible that some pixels outside the 3x2 area are utilised. I hadn't realised it was the same as the 7D, but it is interesting that that is a camera designed with video in mind as well. That would support my theory that this resolution is beneficial for video and that's why they chose it.
Video is 16x9, not 3x2. Obviously in video mode the rectangle that is recorded is a different shape.
Of course. I'm just trying to get to a 3:2 shape from 3x extension of HD video.
It doesn't seem to be clear exactly which pixels are used for the video image, but it might even be possible that some pixels outside the 3x2 area are utilised. I hadn't realised it was the same as the 7D, but it is interesting that that is a camera designed with video in mind as well. That would support my theory that this resolution is beneficial for video and that's why they chose it.
I don't think your theory works either way. HD video is 1920 x 1080. 1920 x 3 = 5760. No matter how you slice it, 18mp doesn't divide by 1920 or 1080 evenly.
-Jack
An "accurate" reproduction of a scene and a good photograph are often two different things.
Of course. I'm just trying to get to a 3:2 shape from 3x extension of HD video.
I don't think your theory works either way. HD video is 1920 x 1080. 1920 x 3 = 5760. No matter how you slice it, 18mp doesn't divide by 1920 or 1080 evenly.
5760 x 3240 = 18.6MP
Not exactly the same as the 18.2 for stills, but maybe some extra pixels are used for the video mode at the right and left edges of the frame.
The stills are 5184 x 3456 BTW.
So the chip would need to be 5760 x 3456.
Not saying this is the case though, just thinking out loud really.
I'm just a newbie when it comes to DSLR. I usually use my phone to get a shot but I want more. I asked a friend (he is a photographer) if which DSLR is better for starters and she said to start with Canon 1100. After I saw this thread, I'd realized that I want this one. Should I wait for this to come out in the market or should I agree with her to start with a Canon 1100?
The camera we are discussing in this thread, the Canon 1D X, is announced but it is not in production yet. When it does come to market it will be desirable, but it will cost around $7000USD. Are you sure that is in your budget?
The Canon T3/1100D is at the opposite end of the dSLR spectrum, and it's designed for entry level.
What is it you want to do with a new camera?
Be as specific as possible about your needs. A camera designed for general photography, family events and everyday snaps, is a very different need that, for instance, a camera designed for indoor sports.
'Wow' and 'Awesome' come to mind. I had not read about the Ethernet port, but the ability to video sync 10 cameras to the same time base, along with the ability to trigger either stills or video from a remote like the Pocket Wizard, along with almost 30 minutes of continuous video (although in multiple files), opens up industrial/ENG/EFP possibilities that did not previously exist at this level.
I'm just a newbie when it comes to DSLR. I usually use my phone to get a shot but I want more. I asked a friend (he is a photographer) if which DSLR is better for starters and she said to start with Canon 1100. After I saw this thread, I'd realized that I want this one. Should I wait for this to come out in the market or should I agree with her to start with a Canon 1100?
I'm just a newbie when it comes to DSLR. I usually use my phone to get a shot but I want more. I asked a friend (he is a photographer) if which DSLR is better for starters and she said to start with Canon 1100. After I saw this thread, I'd realized that I want this one. Should I wait for this to come out in the market or should I agree with her to start with a Canon 1100?
You probably don't want a 1D X. (Unless you have $7000 extra lying around. ) The 1100D is a fine camera. So are the T3i (aka 600D), the T2i (550D), T1i (500D), 40D, 30D, etc. I'd say start with the 1100D and go from there. If you have money left over, you can spend it on lenses
Thanks Nik! A great big geek-meaty burger of info!
3 things stand out for me: multiple exposures - I'd like to have a play!:D More intelligent automated AF, so that shooters who in the case of the 7D remained in perpetual confusion about the AF controls or who left them set on the "least troublesome" setting for everything can benefit from the sophistication in the 1DX while letting the camera make decisions. And more in-camera photo editing doing more of what Lr & Ps etc were used to do, but with camera-lens-shooting situation specific data on the spot.
Interesting to see these guys talking about tech plateauing, as in the case of AF. That's an issue that I've been raising in a number of threads recently.
I'd be always a little nervous with gear like this camera that one little failure somewhere in all this extreme complexity and sophistication would render the whole unit useless, with the only solution being replacement. So, are the warranty conditions keeping pace with the increased likelihood of $9000 having only a 12mo lifespan? dunno
Yeah, stacking up to 9 exposures caught my attention, too , as well as increased number and improved quality of AF points, super crazy high ISO. Basically, they took all the good pieces from 7D (which I like a lot, even though it's only my backup calera), and even improved on them - e.g. custom functions now remember changes you make in the process - huge advantage!
I'm also happy that they finally stopped the infamous MP race and stayed with a perfectly normal 18.1MP this time.
What I kinda didn't get: does the the new DIGIC5 processor delivers true 16 bit image? They mentioned 16 bit in other contexts , but I couldn't find a direct pointer about RAW file resolution.
What I'm not very happy with (like it matters) is that, just like recent PS, a lot of effort (and hence a lot of price) goes into video features. And while I do understand the trend, I personally don't find it that attractive, and would much rather see the same camera (and the same PS) *without* video features, yet hopefully, much lighter and les expensive. I know, I'm speaking like a dyno, cause apparently everybody shooting HD video these days, but it's me, my opinion, my needs, and I stand by them.
It's also a pity one can't use the new GPS and WiFi units similtaneously. I like their new smaller form factor, but a) you have to buy two devices b) you have to carry/store two devices c) you have to choose which one you need more. Granted, the latter is more or less toleable for me - if I shoot studio I probably don't need GPS, and if I shoot on location, I most likely don't have a PC to connect with. So it's OK I guess...
I wonder now: will 5DMkIII get any of those niceties, and if yes, when? Stacking images, improved ISO, smart Custom modes, +/-5EV bracketing, increased number/improved quality of AFs. And - maybe - finally freaking articulated display??? :hide
Yeah, stacking up to 9 exposures caught my attention, too , as well as increased number and improved quality of AF points, super crazy high ISO. Basically, they took all the good pieces from 7D (which I like a lot, even though it's only my backup calera), and even improved on them - e.g. custom functions now remember changes you make in the process - huge advantage!
I'm also happy that they finally stopped the infamous MP race and stayed with a perfectly normal 18.1MP this time.
What I kinda didn't get: does the the new DIGIC5 processor delivers true 16 bit image? They mentioned 16 bit in other contexts , but I couldn't find a direct pointer about RAW file resolution.
What I'm not very happy with (like it matters) is that, just like recent PS, a lot of effort (and hence a lot of price) goes into video features. And while I do understand the trend, I personally don't find it that attractive, and would much rather see the same camera (and the same PS) *without* video features, yet hopefully, much lighter and les expensive. I know, I'm speaking like a dyno, cause apparently everybody shooting HD video these days, but it's me, my opinion, my needs, and I stand by them.
It's also a pity one can't use the new GPS and WiFi units similtaneously. I like their new smaller form factor, but a) you have to buy two devices b) you have to carry/store two devices c) you have to choose which one you need more. Granted, the latter is more or less toleable for me - if I shoot studio I probably don't need GPS, and if I shoot on location, I most likely don't have a PC to connect with. So it's OK I guess...
I wonder now: will 5DMkIII get any of those niceties, and if yes, when? Stacking images, improved ISO, smart Custom modes, +/-5EV bracketing, increased number/improved quality of AFs. And - maybe - finally freaking articulated display??? :hide
Yeah I sympathise on all points, in particular re video.
Re 16bit, I think not all photo editing SW is there yet?
This 1DX demonstrates that Canon is investing in optimising gear-for-purpose. I expect the new model of the 5D2 to be a benchmark setter also.
I doubt very much that video adds a lot of cost here guys. I also think that the 5DIII, Nik, will have the same video capabilities, if not improved
Yes, that point has been made before, and was correct. However, as the info in the article Nik linked shows the 1DX has components exclusively to enhance capture and processing of video which possibly could no longer be called cheap. The level of processing power built into the 1DX is a portent as well. Then there is the matter of the cost of post editing video data. Digital video is experiencing a level of development greater than what dslr tech is now, hardware and soft. This guarantees rising prices of video in/from dslrs which I think will occur over the next 24mo, particularly in top end gear.
Then there is the matter of the cost of post editing video data.
Irrelevant
I still contend that adding video isn't a huge deal any more to the mfg process .. esp with a camera like this, high frame rate, giant files, full size sensor, they were gonna have dual-Digic V no matter what, eh?
I still contend that adding video isn't a huge deal any more to the mfg process .. esp with a camera like this, high frame rate, giant files, full size sensor, they were gonna have dual-Digic V no matter what, eh?
Typically software comes in suites, and no matter that any particular person might use only a fraction of the tools in a suite the cost of developing the whole suite is passed on to and shared by every purchaser. More still photography editing suites are now including video tools.
Yes, higher processing power is one way that dslr tech can improve, but the upper level of that in a without-video body I think could be significantly less than for with-video, commensurably the cost.
Comments
I think both are somewhat untrue! Apart from the huge increase in processing power, which as Ziggy has explained makes the 1DX more reliable and more responsive, this camera doesn't do any more than what the 5DII and 7D do individually.
But it is clearly a different camera to a simple hybrid of those two cameras, even if such a thing were possible. In Canon's launch media they emphatically draw attention to the 1DX's versatility, and this is something that attracts a number of responders to this thread. But Canon even more emphatically draw attention to the 1DX's advantages for the sports photographer - reliability, responsiveness, speed, high ISO and video. Reach is critical for the sports photographer, but not as critical as for the wildlifers, for reasons such as the relative closeness of the subject, and the relative predictability of the subject's position within a known limited space. Therefore much cropping is also less likely to be needed. Sporters' shots are mainly action, they need fast shutter speeds-high ISO and fps which the 1DX has, while wildlifers' shots have a lower proportion of action. Wildlifers typically don't use video, whereas for sporters it's a high value addition to their reporting tools. In addition, and yet another difference, is that sporters typically shoot in a much greater range of light levels than wildlifers, from daylight to twilight to artificial, therefore bigger pixels and high ISO are very desirable.
Why are the wildlifers bellyaching that this camera is not for them?! It is clearly NOT MEANT for them! Does that make Canon mad? Has Canon abandoned wildlifers? I don't see the logic of such ideas. Nor of a belief that a camera which will take the 1DIV another notch is no longer a possibility because of the 1DX. Seems like hysterical "sibling rivalry" to me!
Neil
http://www.behance.net/brosepix
http://cweb.canon.jp/camera/eosd/1dx/samples/index.html
I am surprised they didn't provide better photos at high resolution.
the video looks amazing tho at the ISO settings they've used
The part where the boxer was lying down and breathing heavily was shot at ISO 25,600!! looks pretty good!
and some of the photos where shot at ISO 51,200!! pretty clean to me! then again, i would like to see the full res
My images | My blog | My free course
I get 15.7mp by my math, did I do something wrong?
1080 x 3 = 3240 vertical pixels
3240 x 3 / 2 = 4860 horizontal pixels
3240 x 4860 = 15.7mp
1DX is 3456 x 5184, so I don't see how that fits your theory. (interestingly that is the same pixel dimensions as the 7D)
An "accurate" reproduction of a scene and a good photograph are often two different things.
No you have to divide twice because the crop factor applies to both vertical and horizontal dimensions. That is, a FF sensor is 1.3x wider and 1.3x taller than an APS-H sensor.
So, 18mp/1.3/1.3 = 10.65mp. That's just a smidge more than the 1DIII.
This is also why the 5DII has the same pixel size as the 20D. 21/1.6/1.6 = 8.2mp.
An "accurate" reproduction of a scene and a good photograph are often two different things.
It also seems to me, subjectively at least, that this is true at ISO=100 as well, but it is not clear to me why this should be true. I have done similar comparisons between my 5D2 and my 1D4 as well, with less obvious differences, though I still think the 5D2 gives me better IQ. Because of this, I put up with a lot of the 5D2's limitations as a sports camera and use it all the time to shoot pro sports events, along with my 1D4.
So in thinking about the 1Dx (especially since I am at the top of the waiting list at Calumet) and what it has to offer, I am not sure that there is much advantage in the ISO range I work at (up to 6400 in some stadiums, which is OK on the 5D2 and 1D4). My lenses give me enough reach for almost all shots, save perhaps those in the far outfield, and the bump in FPS I'm sure could be useful but not often for the way I shoot.
I'm sure there are pixel peepers out there who have done all these studies. I have seen a few, but not sure how well they translate into overall IQ. Maybe I should just rent the 1Dx and take it for a test spin before plunking down the $$$...
The differences between the Canon 1D MKIII and the Canon 5D MKII, with regard to random/high-ISO sensor noise are more than just FF versus crop, although that certainly plays a part.
The total concept of "random sensor noise", and the noise level that we see at the in-camera image level (RAW, JPG, TIFF mostly) has to do with the sensor itself, the sensor design (CMOS, CCD, LiveMOS, etc.), distance from sensor to LNA (low noise amplifier), internal shielding, LNA performance, signal bias voltage, analog to digital converter (ADC) accuracy and efficiency, image bit depth, image processor, demosaicing algorithm (type and accuracy), image file storage type, etc.
Trying to separate any of these underlying issues from the total system in a camera is beyond the abilities of most camera users, so it's best to assume that we should limit ourselves to the "output" efficiencies of each camera and the accumulative effect of the system in total.
Assuming that all other parts of the system in a camera are equal, and looking at just the pixel density as the only variable, it's easy to understand the reason why more pixel density results in more random sensor noise (and conversely lower pixel density results in less random sensor noise).
Lower pixel density tends to equalize the random nature of sensor noise because you are taking a single sample of a larger site. That single data access point is accumulating data from a larger sample and the larger sample is using a natural type of normalization (averaging) of the randomness of the media. The same thing occurs with longer time duration sampling (longer time exposures) and/or more samples normalized together (multiple short exposures combined into a single image). All of these mechanisms and techniques will normalize/average the input data resulting in less randomness of the data, thus less random sensor noise.
When you sample a smaller pixel site there is less opportunity for averaging the data, and a greater likelihood that random noise values will have a greater influence on the measured values at that photosite. You can compensate with either longer exposures and/or multiple exposures combined (when appropriate).
Software noise reduction, including the camera's internal firmware noise reduction, uses various different algorithms to try to achieve similar results. (Basically the best software tries to separate random "and" predictable noise from image detail, and then the software uses normalization against the noise component.)
More modern cameras even seem to use a type of image masking, biased towards shadow tones, to help reduce the noise that's more visible in the shadows (both chrominance and luminance noise). This level of processing occurs before the RAW file is generated and the user has no control over the masking or degree of NR, making the resulting RAWs much better with regard to both visible and measurable noise and operationally transparent to the user.
Moderator of the Cameras and Accessories forums
It's interesting that the 1D3 has the largest pixel size (7.2uM) but the worst noise of my three bodies. I guess this is where the 'all other things being equal' assumption fails. Random noise (and I assume you are referring to thermal noise, not shot noise, as the latter is not likely to be the case) can be introduced anywhere in the system. true pixel-level thermal noise is relatively easy to clean up as it appears as pixel-level aberrations. Banding and noisy patches are the hardest to fix, as large amounts of the picture are affected. I believe the latter is due to cross-talk in the electronics, or other thermal properties in the route between the pixels and the A-to-D conversion. In comparing the noise on my 5D2 to my 1D3, I find roughly the same amount of pixel noise on the former, but not much noise from the other components. Part of this might be due to the relatively lower cross talk in the FF sensor circuitry because of the larger spacing. In any case, part of what I incorporated into the denoising code I wrote was the ability to apply noise reduction in a weighted fashion to various parts of the sensor, which takes into account the inhomogeneous distribution of noise on most sensors. IOW, more noise reduction is applied to sensor sectors that are inherently more noisy.
Another factor to consider is the effect of noise on AF. We all know that the speed and accuracy of AF is proportional to light level: low light is slower (or nonexistent), bright light is faster. If the 1Dx follows this pattern (and they claim it does) then overall IQ might actually be better in tight crops that wildlife shooters routinely do than in a sensor with more pixels. Obviously the advantage is inversely proportional to light levels; but in my experience, my crops are far more often limited by small AF errors than absolute resolution. YMMV.
An "accurate" reproduction of a scene and a good photograph are often two different things.
That's what I am hoping... I am glad to see that Canon broke this trend, and am very excited to see what this means ITO IQ and other matters.
Me too. A lot of people seem to have not realized or have lost sight of that. (my frustrations are also fueled by the bench racers in the dpreview forums)
Furthermore, if you graph pixel density vs noise for the same sensor generation, I believe generally there is an inverse relationship. i.e., D7000 vs 7D, D700 vs 5DII, etc.
An "accurate" reproduction of a scene and a good photograph are often two different things.
Signal to noise ratio is greater per unit area with larger pixel size. Initial resolution is relatively low, but binning NR threshold can be relaxed, maintaining a relatively high resolution, and local contrast and DR remain relatively high.
Lens CoC affects local contrast, too.
Overall, given equal Av & Tv & focus & ISO & lens & stability of gear, for cropping purposes larger pixels give better IQ, providing that in camera algorithms are taking the greatest advantage of signal over noise.
SNR can be a mathematical value, or a subjective value. Seems to me that what you guys are talking about is a subjective value, ie eyeballing. In that case, post processing then becomes a relevant factor as well, eg how the RAW was converted. For instance, Lr's defaults enhance without you ever touching a slider, and what Lr does might not be comparable to what ACR does, etc. On top of that there is the quality of the display you are using to make judgements about IQ. In short, any variation in any post processing and viewing factor will affect subjective IQ assessments.
Viewing prints is a different viewing environment again, and has it's own factors affecting IQ.
A very complex dynamic, and so subjective rating of IQ has a very limited relationship to the underlying tech.
Neil
http://www.behance.net/brosepix
Video is 16x9, not 3x2. Obviously in video mode the rectangle that is recorded is a different shape.
It doesn't seem to be clear exactly which pixels are used for the video image, but it might even be possible that some pixels outside the 3x2 area are utilised. I hadn't realised it was the same as the 7D, but it is interesting that that is a camera designed with video in mind as well. That would support my theory that this resolution is beneficial for video and that's why they chose it.
Of course. I'm just trying to get to a 3:2 shape from 3x extension of HD video.
I don't think your theory works either way. HD video is 1920 x 1080. 1920 x 3 = 5760. No matter how you slice it, 18mp doesn't divide by 1920 or 1080 evenly.
An "accurate" reproduction of a scene and a good photograph are often two different things.
5760 x 3240 = 18.6MP
Not exactly the same as the 18.2 for stills, but maybe some extra pixels are used for the video mode at the right and left edges of the frame.
The stills are 5184 x 3456 BTW.
So the chip would need to be 5760 x 3456.
Not saying this is the case though, just thinking out loud really.
"You miss 100% of the shots you don't take" - Wayne Gretzky
That's 16:9. What does that have to do with the 7D being 18mp?
That would be 5:3.
Canon DSLR chips are only 3:2. There is no connection between a 18mp 3:2 sensor and HD video.
An "accurate" reproduction of a scene and a good photograph are often two different things.
The camera we are discussing in this thread, the Canon 1D X, is announced but it is not in production yet. When it does come to market it will be desirable, but it will cost around $7000USD. Are you sure that is in your budget?
The Canon T3/1100D is at the opposite end of the dSLR spectrum, and it's designed for entry level.
What is it you want to do with a new camera?
Be as specific as possible about your needs. A camera designed for general photography, family events and everyday snaps, is a very different need that, for instance, a camera designed for indoor sports.
Moderator of the Cameras and Accessories forums
'Wow' and 'Awesome' come to mind. I had not read about the Ethernet port, but the ability to video sync 10 cameras to the same time base, along with the ability to trigger either stills or video from a remote like the Pocket Wizard, along with almost 30 minutes of continuous video (although in multiple files), opens up industrial/ENG/EFP possibilities that did not previously exist at this level.
Moderator of the Cameras and Accessories forums
You probably don't want a 1D X. (Unless you have $7000 extra lying around.
I apologize if this is an old news, but just in case:
http://cpn.canon-europe.com/content/education/technical/eos_1d_x_explained.do
3 things stand out for me: multiple exposures - I'd like to have a play!:D More intelligent automated AF, so that shooters who in the case of the 7D remained in perpetual confusion about the AF controls or who left them set on the "least troublesome" setting for everything can benefit from the sophistication in the 1DX while letting the camera make decisions. And more in-camera photo editing doing more of what Lr & Ps etc were used to do, but with camera-lens-shooting situation specific data on the spot.
Interesting to see these guys talking about tech plateauing, as in the case of AF. That's an issue that I've been raising in a number of threads recently.
I'd be always a little nervous with gear like this camera that one little failure somewhere in all this extreme complexity and sophistication would render the whole unit useless, with the only solution being replacement. So, are the warranty conditions keeping pace with the increased likelihood of $9000 having only a 12mo lifespan?
Neil
http://www.behance.net/brosepix
I'm also happy that they finally stopped the infamous MP race and stayed with a perfectly normal 18.1MP this time.
What I kinda didn't get: does the the new DIGIC5 processor delivers true 16 bit image? They mentioned 16 bit in other contexts , but I couldn't find a direct pointer about RAW file resolution.
What I'm not very happy with (like it matters) is that, just like recent PS, a lot of effort (and hence a lot of price) goes into video features. And while I do understand the trend, I personally don't find it that attractive, and would much rather see the same camera (and the same PS) *without* video features, yet hopefully, much lighter and les expensive. I know, I'm speaking like a dyno, cause apparently everybody shooting HD video these days, but it's me, my opinion, my needs, and I stand by them.
It's also a pity one can't use the new GPS and WiFi units similtaneously. I like their new smaller form factor, but a) you have to buy two devices b) you have to carry/store two devices c) you have to choose which one you need more. Granted, the latter is more or less toleable for me - if I shoot studio I probably don't need GPS, and if I shoot on location, I most likely don't have a PC to connect with. So it's OK I guess...
I wonder now: will 5DMkIII get any of those niceties, and if yes, when?
Yeah I sympathise on all points, in particular re video.
Re 16bit, I think not all photo editing SW is there yet?
This 1DX demonstrates that Canon is investing in optimising gear-for-purpose. I expect the new model of the 5D2 to be a benchmark setter also.
Neil
http://www.behance.net/brosepix
Portfolio • Workshops • Facebook • Twitter
Yes, that point has been made before, and was correct. However, as the info in the article Nik linked shows the 1DX has components exclusively to enhance capture and processing of video which possibly could no longer be called cheap. The level of processing power built into the 1DX is a portent as well. Then there is the matter of the cost of post editing video data. Digital video is experiencing a level of development greater than what dslr tech is now, hardware and soft. This guarantees rising prices of video in/from dslrs which I think will occur over the next 24mo, particularly in top end gear.
Neil
http://www.behance.net/brosepix
Irrelevant
I still contend that adding video isn't a huge deal any more to the mfg process .. esp with a camera like this, high frame rate, giant files, full size sensor, they were gonna have dual-Digic V no matter what, eh?
Portfolio • Workshops • Facebook • Twitter
Typically software comes in suites, and no matter that any particular person might use only a fraction of the tools in a suite the cost of developing the whole suite is passed on to and shared by every purchaser. More still photography editing suites are now including video tools.
Yes, higher processing power is one way that dslr tech can improve, but the upper level of that in a without-video body I think could be significantly less than for with-video, commensurably the cost.
Neil
http://www.behance.net/brosepix
For me, I love the video capabilities of today's cameras!
Portfolio • Workshops • Facebook • Twitter