Canon 5DmkIII

[fyngyrz]

Here's what I want.

About 8...10 MP. Why? Because large sensels produce larger signals and this will improve the signal to noise ratio, hence allow higher ISOs, a'la the D3s *except* that Canon handles black better than Nikon and also doesn't foul up my images with noise reduction I can't turn off. Also, because an image of this size transfers faster, takes up less space both in the camera and on disk, processes faster, loads faster, saves faster, and also means the camera can shoot faster, and ALSO means that the available lenses will do an excellent job because the size of the individual pixels pushes diffraction further up the f-stops and doesn't require such tight focus to be pixel-accurate, either.

The *only* reason I don't own a 5DmkII is because of the 21 MP. That's ridiculous. My 50D, at 15 MP, is already making images that are far too large. My 40D was better; but the 50D's ability to shoot at ISO 12800 has me hooked. But I can't see going to 21 MP for ISO 25k (still, if it was in my hands... the wallet might be in danger.)

Other than low noise, high ISO, and hopefully an end to megapixel madness, I don't really care about things like AF - give me one good center point and I'd be happy; I *really* don't care about video; I'd like approximately the 50D feature set and that'd be fine.

Having said that, I'm sure they'll give me a plastic body camera with 50 MP, ISO 6400, and every video mode known in NTSC, PAL, SECAM and HD. And I'll have to stick with my 50D or break down and buy a 5DmkII.

 

[Grendel]

Here's what I want.

About 8...10 MP. Why? Because large sensels produce larger signals and this will improve the signal to noise ratio, hence allow higher ISOs, a'la the D3s *except* that Canon handles black better than Nikon and also doesn't foul up my images with noise reduction I can't turn off. Also, because an image of this size transfers faster, takes up less space both in the camera and on disk, processes faster, loads faster, saves faster, and also means the camera can shoot faster, and ALSO means that the available lenses will do an excellent job because the size of the individual pixels pushes diffraction further up the f-stops and doesn't require such tight focus to be pixel-accurate, either.

Well, you could buy a used 1Ds -- 11MP FF, cheaper than a 5DII. Max. ISO of 1250 tho. Sometimes less isn't more

The *only* reason I don't own a 5DmkII is because of the 21 MP. That's ridiculous. My 50D, at 15 MP, is already making images that are far too large. My 40D was better; but the 50D's ability to shoot at ISO 12800 has me hooked.

Set a 5DII to sRAW2 -- volia, 10.8MP... Keep in mind tho that a single photosite of the 21MP sensor is bigger than a single photosite of the 50D or even 40D sensors...

 

[fyngyrz]

Set a 5DII to sRAW2 -- volia, 10.8MP... Keep in mind tho that a single photosite of the 21MP sensor is bigger than a single photosite of the 50D or even 40D sensors...

I'd be more likely to take a full res shot and then reduce it. That gives some improvement on the noise too. That's probably what'll happen. Still end up with ridiculous RAWs, but noise and ISO are critical to my needs, so I don't see a a lot of choice. Unless Canon puts out a competitor to the Nikon D3s... now that would be a camera I'd willingly plunk down some money for!

 

[Macadameane]

My interest in DSLRs is largely video, so I would love a low MP count also, but the photographer in me says 1920 x 1080 just isn't enough resolution.

I don't mind having a high MP count so long as its not too expensive and that the noise levels stay low. The end result is what's important. On the flip side, a low res camera count probably greatly reduce the rolling shutter effect in video.

 

[pwp]

Problem solved with a 1DMkIV. Good video function plus the option of medium mRAW or even small sRAW if you don't want the full size files. I've often switched to mRAW on jobs depending on what the client's planned output is. If it's guaranteed 100% web use, the full size RAW is superfluous. Maybe the 5DIII will have this option too.

pwp

 

[dahmooser]

professionally speaking, what I'd like to see on the Video side is:

Ability to smoothly control zoom, and have auto focus follow - even if this requires a separate "Grip" for lack of a better word.

BUT, MOST importantly, I'd like to see Canon use a compression codec that is less lossy/noisy. The current codec used is a great use of H.264, and a good quality bit rate, but when you put it up against professional shot material, it looks only mildly better than DNxHD36 (36 mbit) which is what we use to edit with in Avid. The issues arise in really dark or really bright areas (not unlike still images), the black fall apart, or in Keys, when shooting Blue or Green Screened VFX.

If they can do something akin to Red (though at a smaller resolution) or HDCAM SR quality, I'd be far happier, and far more likely to use this camera for real work, rather than a video novelty. Yes, I know this is asking for a lot, and Red specifically uses proprietary CF cards capable of maintain a 24fps at 4480x1920, which is pretty damn impressive.

The reason I am thinking this way, is that strategically, canon has a real opportunity to get a piece of the high end market, just like Red has. We're shooting red on a $300M 3d film!! and just a few years ago, Red didn't even exist. Just goes to show.....

 

[Macadameane]

That all sounds great dahmooser. This is a little aside of the point, but I also hope there are more players in the VDSLR rigs market too. Yes Zacuto is nice, but $1000 for some pipes with clamps on them? Follow Focus should be cheaper too.

 

[jc88]

Video is unimportant.

Massive megapickles are unimportant (look at the Nikon range).

AF system is super important: this is where Canon lags behind Nikon. Compare the 5DmkII to the D700 and it's pretty obvious which is the better camera on almost all fronts. The D3s is just amazing as well.

 

[Flake]

"About 8...10 MP. Why? Because large sensels produce larger signals and this will improve the signal to noise ratio, hence allow higher ISOs, a'la the D3s"

I'm afraid that is an urban myth, it's only true at the most basic of levels.

Imagine two people buy an acre of land, one of them builds a skyscraper on it, the other a garden shed, by the logic you're using you would view the two as exactly the same, simply because they each have an acre of land.
You don't know how large the pixel sites are, and as they grow larger they become closer to each other.

Imagine that the two sites have no fences, people from each site are free to pass into each others land, but it's further to go from the shed to the boundary. Imagine a simple wire fence, everyone can see into the neighbouring plot and some are able to climb the fence into it. Now imagine a huge concrete wall no one can see it, and no one can pass through. This is isolation, and it's very important, because you can have small plots where little can pass through and cause noise.

So to bring it back to sensors, although you are given pixel densities, you are not told pixel sizes, without additional isolation there is an optimum size, but again you don't know what that is. Once isolation is introduced pixels themselves can grow larger & therefore closer, and still maintain low noise. Of course on top of that there are micro lenses to focus light into the pixel sites themselves and the design of these makes a difference too. There is a complicated mathematical formula (Cos 4) for working out light fall off at wide angle.

Hopefully I've been able to show a little that it's much more complicated than the simplistic urban myth would have us believe, and I haven't mentioned the signal processors which have a huge effect on noise.

 

[Inst]

Not exactly; for certain matters like dynamic range (who cares if you have more pixels to reduce shadow effective noise if you can't read the shadow anyways due to inadequate sensitivity?) and tonality larger pixels tend to be better. But regarding noise, smaller pixels have zero correlation with SnR; the overall picture quality is determined by the total data received on camera, and if your pixels are noisier on an individual level, by increasing the total number of pixels you also increase signal so the total picture noise does not chance.

Canon, right now, is inferior to Nikon as Nikon simply has better sensor technology than Canon. They've figured out how to jam another half stop worth of quantum efficiency into their sensors, and Canon can't quite catch them. I suppose this may be why Canon may opt for a square sensor on 1Ds4; by adding another 50% onto the sensor, they can immediately compensate for, well, the half stop quantum efficiency advantage Nikon has.

 

[dahmooser]

Thanks for the insightful replies.

I'm not a big fan of Video on an SLR, though, my simple point was, if you're going to do it, make it kick ass.

I was quite intrigued by the post stating that Nikon has better sensor technology. I currently own a 40D (and I do love it), but my (lucky) girlfriend owns a Nikon D3, and aside from it being a full 35mm sensor, I just think the images that are shot on that seem so much sharper, more in focus, something. It's an intangible that I've been unable to properly express. Now, I realized I am comparing an $8K camera to a $1K camera, but, my point is, it's unbelievably crisp where you want it to be.

I am not a pro (in ANY camera area, still or video), but, this issues fascinates me. I sometimes get buyers remorse over having chosen canon, then again, her camera weighs so much, and I must say, the user interface, seriously needs some improvement. It's not the kind of camera you can pick up and use...

Granted, I have seen some great 1ds images, too....

I am patiently saving my hard earned dollars for the successor to the 5D Mark II, so I truly hope it hits the mark!

 

[Aputure]

If the 50D's ability to shoot 12800 has you hooked, and you can tolerate it, then I think you could tolerate a few more megapixels in a new 5D. The 5D2 is already performing much much better than the 50D. So I think there could be a middle ground, such as a 16 or 18mp sensor, but we all know that won't happen. I'm afraid the next round will likely be around the 30mp mark. The 5D2 is probably exactly what you want, just turn the resolution settings down a notch. I shoot my 50D at 8mp all the time!

Here's what I want.

About 8...10 MP. Why? Because large sensels produce larger signals and this will improve the signal to noise ratio, hence allow higher ISOs, a'la the D3s *except* that Canon handles black better than Nikon and also doesn't foul up my images with noise reduction I can't turn off. Also, because an image of this size transfers faster, takes up less space both in the camera and on disk, processes faster, loads faster, saves faster, and also means the camera can shoot faster, and ALSO means that the available lenses will do an excellent job because the size of the individual pixels pushes diffraction further up the f-stops and doesn't require such tight focus to be pixel-accurate, either.

The *only* reason I don't own a 5DmkII is because of the 21 MP. That's ridiculous. My 50D, at 15 MP, is already making images that are far too large. My 40D was better; but the 50D's ability to shoot at ISO 12800 has me hooked. But I can't see going to 21 MP for ISO 25k (still, if it was in my hands... the wallet might be in danger.)

Other than low noise, high ISO, and hopefully an end to megapixel madness, I don't really care about things like AF - give me one good center point and I'd be happy; I *really* don't care about video; I'd like approximately the 50D feature set and that'd be fine.

Having said that, I'm sure they'll give me a plastic body camera with 50 MP, ISO 6400, and every video mode known in NTSC, PAL, SECAM and HD. And I'll have to stick with my 50D or break down and buy a 5DmkII.

 

[Inst]

The difference is that the D3 is a full-frame camera and a 40D is a APS-C camera. The full-frame camera has a sensor about 2.56 times as big as your APS-C, meaning it has that much more surface area to collect light.

Generally speaking, the disadvantages of smaller pixels (less capable in gathering light, resulting in more noise, reduced dynamic range, reduced tonality) should be canceled out simply by having more of them, but if your sensor is smaller, you can't cram more pixels onto the sensor.

And as far as sharpness goes, smaller pixels demand more than larger pixels, simply because the light differentiation they're capable is greater than larger pixels. I wish people would make adapters to condense light onto an APS-C or DX camera, but apparently, first, this may be unworkable due to optical limitations (CA is supposed to get outrageous as you approach f/1.0 and a light condenser would effectively make a f/1.8 nifty fifty into a f/1.1 lens), second, you would need an adapter for every single focal length, third, there would need to be high-quality glass to render all the resolution, and fourth, Kodak is sitting on the patent for such a device.

You also should look up diffraction-limited aperture, which is caused by having low pixel-pitch.

I trust this explanation of how diffraction works.

http://photography-on-the.net/forum/showthread.php?t=747761

After you hit DLA, you need to start using sharpening tools to restore sharpness; you won't be able to get everything back, but you should be able to recover more resolution than you're just getting from the raw image.

 

[Flake]

Canon, right now, is inferior to Nikon as Nikon simply has better sensor technology than Canon.

What a stupid crass thing to write!! anyone who knows anything about this indusrty knows that Kikon has been buying its sensors from Sony for years! when it came to the high MP sensor for the D3x they had to use the Sony version. Thanks to inferior signal processing from the Expeed processor it's one of the slowest cameras on the market today. If you can cut through the bare faced lies which claim it's capable of 5fps to match the 1Ds MkIII (who says the photo press isn't biased!) but neglect to say that it can only do that be reducing the bit depth and sensor to Dx.

The D3 Sensor was a result of Nikons first venture into sensor production on their own, and largely it has lagged behind the likes of Sony & Canon with it's low MP count, they were lucky that people decided that they want a low light low noise camera that can do high Iso. It was an accident, it wasn't by design.

The problem for Nikon is that Sony are pulling out of FF development and that leaves Nikon in a real tight spot, thy currently have Two FF sensors in three camera bodies, that is unsustainable, both Sony & Canon only have one FF sensor in production, so either the future D3 will have a drop down to crop sensor or the High MP FF sensor will be dropped altogether. If even the biggest & best funded camera companies cannot afford to develop two sensors, Nikon has no chance.

 

[Inst]

I'm not exactly sure what the relationship between Sony and Nikon is on the sensor front; they certainly share a fab, but I think Nikon does their own sensor designs.

Regarding the D3, the market knows what they want, and they picked the D3, which combines high-dynamic range and very low-noise. At the time, it was far better than Canon's equivalent offering, given that the 1D3 was on APS-H, and that the 1Ds3 didn't use gapless sensors.

I don't see why you're fetishizing mega-pixels. It mainly matters at ISO 100 shooting, where noise is low enough that the chief limitation on image quality is the amount of detail your sensor can resolve. And even then, loss of dynamic range caused by small pixels is still a consideration. Around maybe ISO 400/800, noise becomes the chief consideration in image quality, so more megapixels is meaningless; only the quantum efficiency of the sensor matters. Canon with the 7D outperformed the D300s for that reason; it had the same noise performance, about the same if not better AF performance, but also had more megapixels. Unfortunately, it could not exceed it in dynamic range, but that was only a one-stop / half-stop consideration.

In any case, the fact remains. The D3s sensor is half a stop better than the technology on the 1D4, even when the 1D4 is scaled up. The D7000 sensor has yet to be benchmarked, but to start with, it's only 10% or so less MP than the 7D, and aside from that, from preliminary reports it appears to have better high ISO performance AND better dynamic range.

 

[Flake]

The reason people changed from the 1D to the Nikon D3 was mostly down to the auto focus issues which Canon didn't do a quick enough or good enough job at resolving.

Fetishising mega pixels? how strange you are! There is no attempt to 'fetishise' anything, small pixels? How large are the pixels on the D3 then? trouble is no one knows except Nikon, it's possible to work out the density and the area, but not the size of the pixel chosen to place there. "only the quantum efficiency of the sensor matters" This is plainly not true as the image processor has at least if not more influence over noise levels.

You believe that the D3 is a half of one stop better than the D4 but so what? There's more to a camera than dynamic range and where the noise floor is, as you said yourself it doesn't become an issue until Iso 800, I'd go at least a stop further, but it's very rare I shoot at these higher Isos and most times Dynamic range isn't an issue either!

You see, to be very much in favour of Nikons products, and I have to ask the question if you've actually come to the right forum? Perhaps Nikon Rumours might be better for you?

 

[kubelik]

let's strive to keep things civil and not personally directed here.

it is fair to bring up things that canon can do better - let's face it, there's lots of them.

but in this particular case, the concern of the original poster that the 5DII's files are too large ... I'd have to say I doubt they're much more of a memory/processing hog than the 50D files you're already working with. and, as the OP himself observed, he can downsample the 5DII's images into even nicer, sharper, lower-res images as part of his workflow.

the question I have is, what's the real complaint? let's face it, no one is going back to 8-10 MP, not even Nikon at this point. I know people hate using noise reduction because it destroys detail ... but an 8 MP image is clean and lacks the detail anyway ... so you're really not ending up with anything different.

I do think Flake is getting at an important point, however, which is that sensor density doesn't mean anything unless we know more about the sensor. I have owned both the 30D and 5DII, which have identical pixel densities and were released fairly soon after one another. I can tell you the images look nothing alike, even when you are only using the central 3600 x 2400 pixels of the 5DII. the 30D looks, to be honest, fairly crappy at 1600 or higher. the 5DII looks fairly usable up to 3200.

 

[carlsanford]

The article states that a 5d Mk III may be announced in March. How long do you think it would take for bodies to actually be available for sale? Anyone have some knowledge as to how Canon has handled its announcements in the past?

I really want to upgrade now and get a Mk II, but I could wait a bit longer. However, if it may not actually arrive on the market until closer to Christmas, then I may buy now.

 

[epsiloneri]

I currently own a 40D (and I do love it), but my (lucky) girlfriend owns a Nikon D3, and aside from it being a full 35mm sensor, I just think the images that are shot on that seem so much sharper, more in focus, something.

That is most likely due to the lenses she has, aside from the FF advantage. Lenses are usually more important than detectors in producing a sharp image. I also assume you compared images obtained from both cameras by the same photographer, because image quality problems can also be due to what's behind the camera (I'm not saying it is in your case).

 

[epsiloneri]

Imagine two people buy an acre of land, one of them builds a skyscraper on it, the other a garden shed, by the logic you're using you would view the two as exactly the same, simply because they each have an acre of land.
You don't know how large the pixel sites are, and as they grow larger they become closer to each other.

Imagine that the two sites have no fences, people from each site are free to pass into each others land, but it's further to go from the shed to the boundary. Imagine a simple wire fence, everyone can see into the neighbouring plot and some are able to climb the fence into it. Now imagine a huge concrete wall no one can see it, and no one can pass through. This is isolation, and it's very important, because you can have small plots where little can pass through and cause noise.

Haha, flake... that was one of the most incomprehensible similes I've ever seen And it had no cars in it. But I agree that larger pixels don't necessarily translate to higher S/N. What mostly matters is the number of photons captured. Whether you capture them on one or four pixels doesn't matter. The exception would be in the very low S/N domain, where the noise becomes read-out dominated. Also, since the well capacity scales closely to the area of a pixel (not volume!), the dynamic range does not change much with pixel size either (the well capacity will be smaller for smaller pixels, but that is exactly made up by there being more of those pixels for an equal area).