*UPDATE* 5D Mark III - February 7, 2012 [CR2]

[waving_odd]

I talking about a 22MP sensor vs a 48MP with the same dynamic range and high ISO performance

Do you mind reminding us which production or rumored FF model(s) that has/have both 22+ in MP resolution and high ISO performance?

Nikon D3x? Its ISO 1600??

Nikon D800 and its rumored ISO 6400??

 

[Kernuak]

22MP is really horrifying news. I will not buy that. Compared to my mk2 I need higher dynamic range, better high ISO performance and higher resolution. I think 48 MP would have been a good choice. Perhaps I will have to consider to switch to D800 if it is really 36MP.
I have no understanding what so ever for those who wants better dynamics and ISO performance but not higher resolution when it's clearly better with the better dynamics, ISO performance AND higher resolution. You must be stupid if you want 2 of 3 improvements instead of all 3.

The specification also doesn't make sense. I don't see the market segment between 1DX and 7D. What we need is high resolution, high dynamic range, high ISO performance and no banding. Of couse also improved AF compared to mk2. High speed isn't important since that area is covered by 1DX and 7D. Anyway with 48MP (or 36MP) there would have been possible to make a crop mode with higher speed.

So how horrified were you when the 1DX was only 18MP?

Very! But would not have bought it anyway because of cost.

Increased resolution results in increased noise. While technology has resulted in improvements with each sensor generation, there are limits. Most people are looking at full frame sensors for cleaner and sharper images. Another disadvantage of 48MP, would be the high pixel density (higher than the 7D). Higher pixel densities result in greater diffraction limitations, due to the effect of Airy Disks, which cause interference with small pixels, resulting in increased affects by diffraction, which softens the image at smaller apertures. Landscape photographers in particular use smaller apertures, so the diffraction limit starts affecting overall image quality, thus negating one of the important reasons for them going with a full frame sensor. Even medium format cameras, with their larger sensors are only around 40MP, so 46MP in a DSLR would be pusing the boundaries.

 

[wickidwombat]

22MP is really horrifying news. I will not buy that. Compared to my mk2 I need higher dynamic range, better high ISO performance and higher resolution. I think 48 MP would have been a good choice. Perhaps I will have to consider to switch to D800 if it is really 36MP.
I have no understanding what so ever for those who wants better dynamics and ISO performance but not higher resolution when it's clearly better with the better dynamics, ISO performance AND higher resolution. You must be stupid if you want 2 of 3 improvements instead of all 3.

The specification also doesn't make sense. I don't see the market segment between 1DX and 7D. What we need is high resolution, high dynamic range, high ISO performance and no banding. Of couse also improved AF compared to mk2. High speed isn't important since that area is covered by 1DX and 7D. Anyway with 48MP (or 36MP) there would have been possible to make a crop mode with higher speed.

DITTO! and perhaps they can make a new 5-500mm f2 Lens as a kit all for $2000

 

[Wrathwilde]

I have no understanding what so ever for those who wants better dynamics and ISO performance but not higher resolution when it's clearly better with the better dynamics, ISO performance AND higher resolution. You must be stupid if you want 2 of 3 improvements instead of all 3.

Donny you're out of your element!

Unless there is some tremendous new breakthrough that negates the laws of physics... you only get to pick two.

If you want amazing DR and ISO performance then you have to sacrifice Resolution.

If you want high ISO and Resolution then your DR is really going to suffer.

If you want DR and Resolution then you're going to have to take a hit on ISO.

Sure we would like all three, but until we can negate the laws of physics in regards to CMOS sensors, you only get to pick two priorities, or a compromising balance of all three, and that's exactly what the 1D X supposedly is, the sweet spot between all three.

I suggest you read up on how CMOS sensors actually work, there are actually physical limits to the photon collecting ability. But the gist is... the smaller the pixel, the less light it will take in, and consequently a lower ISO.

Sensor Technology - http://dpbestflow.org/camera/sensor

Fundamentals of Image Sensor Technology - http://www.cse.wustl.edu/~jain/cse567-11/ftp/imgsens/index.html


See for yourself that higher resolution can cause diffraction that actually lessens the image quality...

 

[bigblue1ca]

You must be stupid if you want 2 of 3 improvements instead of all 3.

Wow, talk about how not to make friends and influence people.

 

[Kahuna]

I have no understanding what so ever for those who wants better dynamics and ISO performance but not higher resolution when it's clearly better with the better dynamics, ISO performance AND higher resolution. You must be stupid if you want 2 of 3 improvements instead of all 3.

...but until we can negate the laws of physics in regards to CMOS sensors...

+1 Thank you Wrathwilde

 

[BDD]

You must be stupid if you want 2 of 3 improvements instead of all 3.

Wow, talk about how not to make friends and influence people.

OUCH!! Not cool thing to say to any one.

For me I want "high ISO + DR". I would be happy if the MP on the 5D3 is still 21MP. I thought the same way about the D3s when it was first introduced. It only has 12MP but does everything else I want (no I didn't buy it). Now we have the D4 and the 1D-X. 2 great sports/photo-journalism shooters. Which comes out ahead in practice...we'll have to wait and see.

I was so hoping the D800 wouldn't be high MP "focused" but it is. Sacrificing low-light shoot-ability. Plus it's priced too high for both versions IMHO. So my money is on the 5D3...just hoping it borrows the native ISO range of the 1D-X offering 100-51,200 (or a max of 204,xxx). If not...then D3s vs. D4 vs. 1D-X.

 

[LetTheRightLensIn]

Donny you're out of your element!

Unless there is some tremendous new breakthrough that negates the laws of physics... you only get to pick two.

If you want amazing DR and ISO performance then you have to sacrifice Resolution.

If you want high ISO and Resolution then your DR is really going to suffer.

No, no it does not have to.
The 7D actually has BETTER DR per area of sensor than the 5D2 by a little bit....
Some P&S shoot have had better SNR per sensor area than some of the DSLR.

To fix up low ISO DR Canon needs to fix their read noise, less random read noise and a lot less banding noise.

High iso middle gray SNR might get a touch worse with a higher MP sensor than a lower one but with the sorts of difference we talk about here in terms of MP I doubt that it would cost more than 1/3 stop.

See for yourself that higher resolution can cause diffraction that actually lessens the image quality...

That's a terribly misleading demonstration of diffraction, that just shows that the point where you can't maximize what you can best get out of the sensor hits earlier but it implies that you actually do worse in the general sense of a normalized comparison even though you do not do worse from diffraction with the higher MP camera. Shoot the same scene with the same lens at the same f-stop and you never pull in worse detail with the higher MP sensor be it f/1.4 or f/64. It may be that you don't pull in any extra detail once you are at f/64 but you won't do worse and you will pull more detail at lower f-stops and dont; forget that the limit is not a hard limit you slowly gain less and less EXTRA detail the higher you raise the f-stop.

 

[LetTheRightLensIn]

Actually his hints fit these specs pretty well. He was misinterpreted to hint at higher resolution but he did not. And he did not call it a 5D III. If anything he hinted it was a union of the 5D and 7D line.

But didn't he say one of three main specs in that one guy's list were wrong? That guy listed:
6fps
22MP
19 pt AF with Digic 4 assist

Two of those specs don't appear to match this CR2 info.

He has an interesting reputation. He drives people crazy but he is one of the most informed forum participant I've seen on most forums with regard to industry trends and if you know his batting average he is very credible - but he doesn't say as much people read into his posts. He seems to have a special interest in Canon and is up front that he has to be very careful of what he says due to an NDA agreement. He did say the speculated specs seemed close but he believed one detail was wrong. But he did not refer to one specific spec list and there are several variation even in that thread. He did confirm the grip in the pictures was not integrated ;D (duh!) but he was even careful to say that can be derived from the photos to abide by his NDA (and he postulated that the Canon employee may now be the hunted rather than the hunter on the game reserve) ;D.

Yes, most of us are probably read into his hints wrong. If I ignore them and pick among
22MP, 18MP, 32MP, 6fps, 7.5fps, 7D AF, 7D+ AF, 7D++ AF,1DX AF then the most sensible guess to me might be:

32MP, 6fps, with 7D+ AF

Some are counting the zoom box dimensions and size of the mini location finder dot in the box and finding 30MP counts much easier to come up with than 22MP counts and much, much easier than 18MP counts, but that is dicey going.

(if not that, then I would say 18MP, 7.5 fps, 7D++ AF but the above seems more likely to me)

 

[LetTheRightLensIn]

- I also expect we won't see any more than a doubling of the ISO speed, probably up to 12,800, but much more usable and probably software limited just to keep it from eating into the 1D X sales. Hopefully we'll see a nice jump in the DR too. But expect the 1D X to trounce the 5DMK3 when it comes to low light and DR performance, even if Canon has to kneecap the upper ISO select-ability of the 5DMK3 to do so. I don't think Canon will let the selectable ISO be anywhere near the 51,200 of the 1D X, even if the noise levels for the two look identical up to 12,800... the 5DMK3 will just not have anything higher available.

I don't think so, Canon has never once intentionally crippled sensor performance.

You misunderstand me, I'm not talking about crippling sensor performance, I'm sure it will be phenomenal, it's just they will impose a limit to the selectable ISO. Who knows it may even remain exactly where it is for select-ability, just that the resulting images will all be much more useable.

And yes Canon has crippled performance of cameras by not including a robust firmware. See the following link. The 300D has the same sensor as the 10D, but the 300D was firmware limited to 1600 ISO and the 10D to 3200 ISO. A firmware hack was all that was needed to reclaim the 3200 ISO on the 300D, and other functions that had been left out.

http://www.bobatkins.com/photography/digital/digital_rebel_firmware_hack.html

OK true yeah they have crippled it in that way. Although that really only cripples jpg and video since you do better pushing RAW from lower ISO than using those top settings anyway (granted, it makes reading the histogram and trickier and in camera image review looks dark and nasty when you under expose and then push in post).

 

[Wrathwilde]

Donny you're out of your element!

Unless there is some tremendous new breakthrough that negates the laws of physics... you only get to pick two.

If you want amazing DR and ISO performance then you have to sacrifice Resolution.

If you want high ISO and Resolution then your DR is really going to suffer.

No, no it does not have to.
The 7D actually has BETTER DR per area of sensor than the 5D2 by a little bit....

From what I've read on DxOMark the 5DII beats the 7D in DR at every ISO level with a final score of 11.9 vs the 7D's 11.7.

What did the 7D give up to achieve it's higher pixel count per square mm while maintaining a high DR, oh yeah ISO quality. Look at the DxOMark Sports Score, this is their criteria...

"Sports Score is based on Low-Light ISO performance (values in ISO index). Low-Light ISO indicates the highest ISO sensitivity to which your camera can be set while maintaining a high quality, low-noise image (based on a Signal-to-Noise-Ratio [SNR] of 30dB, a dynamic range of 9EVs and a color depth of 18bits)."

The 5DII reached 1815 ISO before falling below that quality threshold. The 7D only reached 854 ISO before falling below the same quality threshold.

So yes, the 7D did manage to get close to the 5DII's DR levels and upped the resolution per square mm, but it came at the expense of ISO performance.

I will admit the diffraction limit example wasn't the best... but it was the best I could find in the limited time I had before heading off to watch the Pro Bowl.

Cheers,
Wrathwilde

Also where are you getting the info that some P&S have a better signal to noise than DSLR? I haven't run across that yet. Not saying that it's not true, but what good is it if the rest of image quality isn't there?

 

[Wrathwilde]

The 7D outperforms the 5Dmk2 at high ISO image quality per area sensor.
A 7D sensor scaled to FF at 46MP would have better image performance than 5Dmk2 in every aspect.

You keep saying that, where's your proof? I think it's clear that the 7D gave up ISO performance to hit the DR numbers it did. DxO labs is probably the most respected independent lab out there, and their tests show that the 7D is not even close to the 5DII in ISO quality.

Again, the 5DII hits 1815 ISO before falling below acceptable quality.

The 7D is only able to hit 854 ISO before falling below acceptable quality.

The 7D does a decent job, but it in no way "outperforms" the 5DII in ISO quality.

So show me the proof, or I write you off as a troll.

And, no, at best a 46mp FF sensor based on the 7D would perform at the level of the 7D, possibly more noise issues cropping up from the larger size. It would still be subpar to the image quality of the 5DII in every category. The trade off might be ok for you, but some of us actually want better quality at high ISO, and a 7D equivalent FF sensor doesn't cut it.

Sensor Scores 5DII 7D Bold type denotes winner in each category

Over all Score 79 66

Color Depth (bits) 23.7 22

Dynamic Range 11.9 11.7

Low Light ISO 1815 854

Edit - to include other DxOMark scores for the 5DII and 7D

 

[briansquibb]

The 7D outperforms the 5Dmk2 at high ISO image quality per area sensor.
A 7D sensor scaled to FF at 46MP would have better image performance than 5Dmk2 in every aspect.

I must write to Canon and tell them that they made a mistake in reducing the mpd from 21 to 18 as it will reduce the performance below that of the 5DII.

I will tell them that they should be looking to a 46 mps ff and all their high iso/low noise issues will go away

 

[Wrathwilde]

I must write to Canon and tell them that they made a mistake in reducing the mpd from 21 to 18 as it will reduce the performance below that of the 5DII.

I will tell them that they should be looking to a 46 mps ff and all their high iso/low noise issues will go away

I applaud you sir, a good laugh was had. But why limit it to 46 mps ff, obviously more is better, the more pixels you have... the more light you can gather, right? So I'm thinking tell Canon to shoot for the moon - 1,800MP Full Frame, if the 1D X can hit 204,000 ISO at 18MP, then a 1,800MP full frame sensor should be able to hit 20,400,000 ISO. The DR would be amazing too, because everybody knows that image quality goes up across the board when you increase the megapixels.

The trade offs in MP/ISO/DR up until now have just been a conspiracy to increase sales by leap frogging the competition. Canon, Sony and Nikon are all sitting on the perfect High Megapixel, HDR, million ISO sensor... with no noise, all these tradeoffs are just inside jokes they're playing on their customers.

Cheers,
Wrathwilde

 

[Loswr]

A 7D sensor scaled to FF at 46MP would have better image performance than 5Dmk2 in every aspect.

Fine. The prototype 120 MP APS-H sensor scaled to medium format at 495 MP will outperform a 7D sensor scaled to FF at 46 MP. It will probably outperform any other imaginary sensor, too. Who cares?

 

[briansquibb]

A 7D sensor scaled to FF at 46MP would have better image performance than 5Dmk2 in every aspect.

Fine. The prototype 120 MP APS-H sensor scaled to medium format at 495 MP will outperform a 7D sensor scaled to FF at 46 MP. It will probably outperform any other imaginary sensor, too. Who cares?

Wow - is that the Feb 7 announcement ?? ??? ??? ???

 

[LetTheRightLensIn]

Donny you're out of your element!

Unless there is some tremendous new breakthrough that negates the laws of physics... you only get to pick two.

If you want amazing DR and ISO performance then you have to sacrifice Resolution.

If you want high ISO and Resolution then your DR is really going to suffer.

No, no it does not have to.
The 7D actually has BETTER DR per area of sensor than the 5D2 by a little bit....

From what I've read on DxOMark the 5DII beats the 7D in DR at every ISO level with a final score of 11.9 vs the 7D's 11.7.

What did the 7D give up to achieve it's higher pixel count per square mm while maintaining a high DR, oh yeah ISO quality. Look at the DxOMark Sports Score, this is their criteria...

"Sports Score is based on Low-Light ISO performance (values in ISO index). Low-Light ISO indicates the highest ISO sensitivity to which your camera can be set while maintaining a high quality, low-noise image (based on a Signal-to-Noise-Ratio [SNR] of 30dB, a dynamic range of 9EVs and a color depth of 18bits)."

The 5DII reached 1815 ISO before falling below that quality threshold. The 7D only reached 854 ISO before falling below the same quality threshold.

So yes, the 7D did manage to get close to the 5DII's DR levels and upped the resolution per square mm, but it came at the expense of ISO performance.

I will admit the diffraction limit example wasn't the best... but it was the best I could find in the limited time I had before heading off to watch the Pro Bowl.

Cheers,
Wrathwilde

Also where are you getting the info that some P&S have a better signal to noise than DSLR? I haven't run across that yet. Not saying that it's not true, but what good is it if the rest of image quality isn't there?

per area
not talking per entire sensor

all the DSLR blow away P&S per entire sensor area, but some P&S are better per area, the thing is they have wayyyyyy less total area to collect light

And sure the 5D2, if you shoot with the same settings, has less noise than the 7D across the entire frame in total.

dxo does appear to show the 5D2 still having a bit better SNR per area than the 7D although some other measurements showed the 7D a trace ahead

dxo shows them tied for DR per area at low ISO, some other sites had 7D a trace ahead

But that just shows that even a 48MP FF could easily have the same DR and nearly the same SNR, never mind a 30MP FF.

 

[briansquibb]

And sure the 5D2, if you shoot with the same settings, has less noise than the 7D across the entire frame in total.

dxo does appear to show the 5D2 still having a bit better SNR per area than the 7D although some other measurements showed the 7D a trace ahead

dxo shows them tied for DR per area at low ISO, some other sites had 7D a trace ahead

But that just shows that even a 48MP FF could easily have the same DR and nearly the same SNR, never mind a 30MP FF.

As iso under 400 is not important - who cares that the 7D marginally beats the 5D.

The important measurement is at higher iso when noise starts creeping in and the IQ starts to fall - then you see the 7D fall away

 

[Loswr]

per area
not talking per entire sensor

all the DSLR blow away P&S per entire sensor area, but some P&S are better per area, the thing is they have wayyyyyy less total area to collect light

But again...so what? We're taking pictures - the summed output of all the pixles - not looking at the output of individual pixels. Some of the detectors I use in the lab have read noise and DR that completely blow away any CCD or CMOS sensor - but those are PMTs, essentially one giant pixel. So, who cares?

You do a good job of refuting your own argument - if a P&S is better than a dSLR per unit area but that 'advantage' is negated by the smaller area, the same logic holds when comparing the 7D to the 5DII - the larger sensor area trumps the putative better per-pixel performance.

Unless you can wave a magic wand and 'grow' the 7D sensor to FF size, the 5DII is the winner.

 

[K-amps]

Well done, indeed. You two are about the 56th and 57th, respectively, to post about 'discovering' this on CR alone.

ROFL! ;D

I gave almost everyone on this thread a +1 applaud, especially lensrightin dude who gets a lot of smites for saying his mind, and to everyone one else as well who is excited about the 5d3, it's been a fun thread to read. Thanks guys!

You get one too John for your wit