5D Mark III Information [CR1]

[dgb]

I LOVE LOVE LOVE the idea of less megapixels and better iso

Even though the one has sweet FA to to with the other...

That's kind of the point of all this talk about MPs , one has everything to do with the other.

 

[AprilForever]

We can make this thread even 20 pages and there will still be NO CAMERA announced... :-\
And we can argue about the Good, the Bad and the Ugly in MPs and ISOs and landscapes but there are many gusto's and still NO CAMERA... :-\
So, better discuss more real things which make sense and have a teaching impulse, Canon will do their job some day and deliver a low light DSLR AND a High MP DSLR so we can all have one of each .

My path to 7D High ISO ultra performance is to shoot at 6400, noise process it, and downres it, and glory at the wonders....

But I do marvel at how a CR1 managed to get 11 pages in like 2 days...

 

[Fleetie]

We can make this thread even 20 pages and there will still be NO CAMERA announced... :-\
And we can argue about the Good, the Bad and the Ugly in MPs and ISOs and landscapes but there are many gusto's and still NO CAMERA... :-\
So, better discuss more real things which make sense and have a teaching impulse, Canon will do their job some day and deliver a low light DSLR AND a High MP DSLR so we can all have one of each .

My path to 7D High ISO ultra performance is to shoot at 6400, noise process it, and downres it, and glory at the wonders....

But I do marvel at how a CR1 managed to get 11 pages in like 2 days...

I'd like to give this a go. Normally, it's firm POLICY for me not to touch pics in post, ever.

But I do sometimes want to use high-ISO.

However, I run Linux, and only have (The) Gimp installed. I'd like to do 2x2 pixel binning, for a 4.5MPix
result from my 7D pics.

Anyone know how to do that binning in Gimp?

Thanks.

Martin

 

[LifeAfter]

Hi all

I don't know why a 5DII with digic 4 can have 21mp and 3.9/fps (14 bit a/d), and the 5DII would have 18mp digic 5 (perhaps) with 5 or 6 fps (while 7d has 18mp, 8fps two digics)...

I don't know if i'm clear with my English (sorry), but if it's true (the rumor) than it should have a hell of a DR and NOISELESS ISO, it should really be a great step forward, otherwise there isn't any logic to all of this

thank you

 

[Radiating]

i personaly dont really mind more megapixels however seeing the lenses at the moment cant deliver enough detail for the 18+ mp sensors its a good point to stick to 18MP sensors for a year and perfect those.
maby if they find a way to imporve lenses further there is a point in adding more megapixels for this cheapass cropfreak.

I've heard this before and find it hard to believe. Do you have any data to back up that claim? The top shelf lenses today have (give or take) been around for decades in one way or another and have always been resolving high enough for film. How could they be not good enough for 18 or 21 or whatever MP sensors? There are certainly many issues today that could see improvement. Lens quality doesn't seem to be one of them really to me (other than the fact that we're pretty much married to AF these days and that lenses aren't that haptically appealing anymore).

There is a reason by the way why Leica decided to develop M-series cameras that take all of their old lenses. Seems to work really great.

All you have to do is look at MTF data taken from test charts. The simple fact that we can even get an MTF figure by shooting a test chart with a camera means that the lens at some point cannot out resolve the sensor. MTF is actually a measure of resolution, just like megapixels, and you can convert between the two without much difficulty. If you look at the MTF data at the mid frame and use that as a marker for the whole lens you will find that many Canon zoom lenses can only resolve 10-18 megapixels at their sweet spot a kit lens generally resolves 10 and an L lens or some high end non-L lenes resolves 16.5-18. Many Canon prime lenses also only resolve 18 megapixels. ONLY a very select few super expensive lenses resolve more and are probably good for 30-40 MP, which we can figure out by looking at APS-C test charts.

The sharpness of a good lens is usually relatively consistent over the image circle, except at the very corners. The sharpness tends to vary less than 30% across the vast majority of the frame, so there are benefits to having more mexapixels to some of the image, but the majority of it will not benefit. Around 24 megapixels you see less than around 90% of the frame gaining from additional resolution.

Canon would not benefit from going up in megapixels until it introduces a new series of primes and normal zooms for full frame. They have the technology to make these high resolution lenses, but simply do not have products capable of doing the job.

I was personally hoping that Canon would introduce a much sharper line of lenses along with pushing the megapixels higher but obviously that's not been the case.

With the current state of the Canon lens line up I think that an 18 MP camera with much better image quality in a 5D III is the best solution.

 

[jbwise01]

It’s quite clear with demonstrated ISO capabilities of the c300 are quite important. It seams that the rumor alludes to these High ISO performance features to be a major part of the 5D MIII.

Why doesn't everyone understand that improving the ISO performance is a HUG HUGE accomplishment?

High ISO performance is not just about shooting in low light. Let relax and remember exactly what ISO is all about… speed

Higher ISO means faster shutter speeds and larger DOF which increase the crispness and clarity of a photo under challenging conditions. High MP are surely the next step, but what MOST people complaining about megapixels don’t get is that increasing the megapixel count is NOT as important or as useful as increasing the ISO capabilities, and for that reason the MP jump will not come after the5D III.

 

[DavidRiesenberg]

For a lot of people high ISO means nothing. Landscape and studio photographers, barring extremely fringe situations will never use an ISO setting above the base or L value.

 

[torger]

With the current state of the Canon lens line up I think that an 18 MP camera with much better image quality in a 5D III is the best solution.

It depends on what you expect. A 7D shows how the center portion will look of a 46 megapixel full-frame sensor. You will see dropoff in the corners, especially at large apertures of course, but at say f/8 you'll see good performance from many current lenses.

Here's a few example of current lenses that I think would work well with a high resolution sensor (in the 36 - 48 mp range):
TS-E 17, TS-E 24 II, 50/1.4, 85/1.8, TS-E 90, 100/2, 135/2, 70-200/2.8 II, 400/5.6, + high end tele lenses. These will provide good corner-to-corner performance at f/8, noticably less sharp when pixel-peeping yes, but no extreme differences. In short DoF photography corner performance is usually not very important. At 36 megapixels the pixel pitch is only 25% smaller than current 5Dmk2, so it is not extreme resolution increase we are talking about. If lenses were really bad at 36 megapixels, it would clearly show already at 21.

Wide angle and standard zooms will probably show the weakest performance (70-200 is a much better performing zoom range), and won't be the choice for a high res photographer. On the wide end we have the new TS-Es which are great, but I'd like to see a new TS-E 45mm, the current is not too good from a high res perspective.

Getting the most out of high res sensor does require expensive lenses on the wide end and on the tele end, but in the range 50 - 100 sharp is cheap.

It is not necessary to limit the sensor to what the worst lenses can provide. There will be binning modes, sRAW etc, for hand-held, high ISO, and less sharp lenses, you won't have to use max resolution when you don't need it.

There's also another aspect - "sharp pixels" (=aliased, jaggies) don't enlarge well. Is is good if the sensor has so high resolution that you get fairly soft pixels which can be stretched around in PP and enlarging without showing any artifacts. Low-res images without AA filter (i e Sigma) is the worst for enlarging.

 

[jbwise01]

For a lot of people high ISO means nothing. Landscape and studio photographers, barring extremely fringe situations will never use an ISO setting above the base or L value.

I would argue that for the majority of the 5D MIII market, the ISO improvements would be more welcome than increase in MP. Personally, I would rather be able to shoot at ISO 8000 without worrying about noise, that would give me a lot of flexibilty and increase the usablility of my L zooms. Shooting wide open usually is out of the sweet spot for most lenses in terms of sharpness.

Would I like more MP? Sure, but if I had to choose between the MP and ISO i would choose ISO, everyone is different, but i think there are a lot less large print photos coming out of the curretn 5D cameras than these landscape photographers suggest.

 

[DavidRiesenberg]

Personally, my biggest fear is that Canon would segment out the place that the 5D currently holds. When it came out, it offered a top of the line resolution, low light capability, video features in DSLR form and all that at a reasonable price. This combination is what made the 5D such a hit with an extremely wide variety of users. Now, more than three years later, taking into account the extremely fast pace of progress in electronics, from a technical point of view I don't think it is outrageous to expect a camera that improves on all of the capabilities of the 5DII and maintains the same price.

On the other hand, from a marketing point of view, I can see Canon wanting to segment out the 5D into an ISO camera, a high MP camera and video camera. And while it depends on exactly how much they would differentiate the products, it would be a disappointment to me if I find myself in the position of having to buy two different bodies to get the features and quality that three years of development could have brought to a single 5D body from a technical point of view.

 

[spaceheat]

i guess canon engineers know it better then forum members and thats why they choose 18MP.

I doubt very much that it was Canon engineers that chose 18MP.

The 1DX appears to have been created to fit a market and in doing so, I'll wager that Canon's marketing arm did their home work and talked to lots of photographers, for whom the 1DX is aimed at, to find out what they want in a camera.

I remember filling out a survey for Canon this past year that covered these exact topics. One of the questions was something like... What improvement would you like to see most in future camera bodies... Increased Resolution or Increased ISO and Dynamic Range? You had to pick one or the other. You could not answer "both". They also asked some questions about mirrorless systems. For the record, I answered more dynamic range and ISO... even though I only really care about the Dynamic Range. I think the ISO performance of the current generation is fine.

I am guessing that this survey led to their decision to take the path that they have.

I personally would prefer to see them develop a medium to large format solution for everyone crying for increased resolution.

 

[Picsfor]

But I do marvel at how a CR1 managed to get 11 pages in like 2 days...

The answer is quite simple - and has been for quite some time.

The 5D2, apparently, seems to have been the most popular and well received digital camera of all time ???

When every thing gets measured against a camera as it does the 5D2, you can expect any upgrade/ future path for such a camera is going to evoke some fairly strong feeling.

• The problem is, the camera is so good killed off a whole pro line for Canon.
• Nikon have never come close to producing anything that matches it, and often roll out the D3s as "there you go" (does any one ever mention the D3X?)
• The video industry discovered they could turn out professional quality staff at a fraction of the price
• The large print/ landscape market found a camera that could achieve their needs without paying MF & LF prices

11 pages in 2 days? I think we've had it lightly. All we've had is opinions and techno babble - we haven't even got to ranting and abuse yet! :-X :-X

 

[Radiating]

With the current state of the Canon lens line up I think that an 18 MP camera with much better image quality in a 5D III is the best solution.

It depends on what you expect. A 7D shows how the center portion will look of a 46 megapixel full-frame sensor. You will see dropoff in the corners, especially at large apertures of course, but at say f/8 you'll see good performance from many current lenses.

Here's a few example of current lenses that I think would work well with a high resolution sensor (in the 36 - 48 mp range):
TS-E 17, TS-E 24 II, 50/1.4, 85/1.8, TS-E 90, 100/2, 135/2, 70-200/2.8 II, 400/5.6, + high end tele lenses. These will provide good corner-to-corner performance at f/8, noticably less sharp when pixel-peeping yes, but no extreme differences. In short DoF photography corner performance is usually not very important. At 36 megapixels the pixel pitch is only 25% smaller than current 5Dmk2, so it is not extreme resolution increase we are talking about. If lenses were really bad at 36 megapixels, it would clearly show already at 21.

25% is actually a huge difference. Keep in mind we're talking about area here, the pixels would be 42% smaller. Lenses are already struggeling at 21 megapixels. If you make the pixels 42% smaller the problems will be severe.

That list of high resolution Canon leneses is more or less comprehensive, the 400mm f/5.6 does not belong there, and the TS-E 17 50mm f/1.4 and 85mm f/1.8 are limited around 24 MP based on APS-C tests. There are a few other rarer lenses missing from the list, but yes those are the ones one would have to work with, most of which would have to be stopped down. Realistically with a 36mp camera you'd just see etremely marginal benefit outside of the 24mm TS-E, 70-200mm, and the supertelephotos, which isn't very good at all.

There are definitely some exceptions to the limited resolution of Canon's current lenses and even some cheap lenses the could eek out marginal benefit from a few more MP, but the reasonable conclusion is that the benefit of more resolution is just too marginal at this point. On the other hand more DR and iso has a much much more wide reaching benefit.

Wide angle and standard zooms will probably show the weakest performance (70-200 is a much better performing zoom range), and won't be the choice for a high res photographer. On the wide end we have the new TS-Es which are great, but I'd like to see a new TS-E 45mm, the current is not too good from a high res perspective.

Getting the most out of high res sensor does require expensive lenses on the wide end and on the tele end, but in the range 50 - 100 sharp is cheap.

It is not necessary to limit the sensor to what the worst lenses can provide. There will be binning modes, sRAW etc, for hand-held, high ISO, and less sharp lenses, you won't have to use max resolution when you don't need it.

There's also another aspect - "sharp pixels" (=aliased, jaggies) don't enlarge well. Is is good if the sensor has so high resolution that you get fairly soft pixels which can be stretched around in PP and enlarging without showing any artifacts. Low-res images without AA filter (i e Sigma) is the worst for enlarging.

I think the biggest benefit of going to 36mp would be a 10% or so bump in resolution accross the board by the presumed removal of the AA filter. There are definitely many side benefits to more megapixels.


The thing is just that even most landscape photographers would have very marginal benefit from a 36mp camera, and going with an 18mp one that has way more DR and iso is much better for everyone.

I personally want Canon to release a 36mp camera even without the right lenses for it because it will drive innovation but if I worked at Canon I would make their 18 mp sensor a priority and the 36mp one a niche product.

 

[sarangiman]

" Canon zoom lenses can only resolve 10-18 megapixels at their sweet spot a kit lens generally resolves 10 and an L lens or some high end non-L lenes resolves 16.5-18. Many Canon prime lenses also only resolve 18 megapixels."

Where are you getting these numbers?

As I understand it, the resolving power of a system is roughly:

1/R(system) = 1/R(sensor) + 1/R(lens)

Hence, increasing the resolution of the sensor without increasing the resolving power of the lens will typically still lead to an increase in resolution of the system until you've reached the point where the resolving power of the lens is severely limiting. If one does a simple plot of this equation (see below), that ends up being when the resolution of one component is an order of magnitude greater than the resolution of the other component. At this point, increasing the resolving power of the higher resolution component leads to severely diminishing returns. See plot below:

So what I'm saying is I seriously doubt Canon's lenses resolve anywhere near 10-fold less than their sensors... which means resolution increases in sensors even without new glass will still lead to increases in MTF.

I hope people remember that just b/c a pixel looks soft on a higher resolution sensor doesn't mean that the increased sampling resolution is useless or detrimental. A higher resolution sensor will always, barring other unforeseen variables, produce at least as sharp as an image as a lower resolution sensor once you've appropriately scaled both images to the same size.

That being said, I'm more in favor of higher dynamic range sensors & I'm hoping that the tradeoff will be worthwhile in the 1Dx.
Rishi

 

[torger]

That list of high resolution Canon leneses is more or less comprehensive, the 400mm f/5.6 does not belong there, and the TS-E 17 50mm f/1.4 and 85mm f/1.8 are limited around 24 MP based on APS-C tests. There are a few other rarer lenses missing from the list, but yes those are the ones one would have to work with, most of which would have to be stopped down.

How do you define "limited"? 50mm/1.4 limited to 24 MP means that it cannot do more than about 9 megapixels APS-C, that seems to me as a low number. I think 12 megapixels is more reasonable as the breakpoint "of diminishing returns" for APS-C, yielding 31 megapixels fullframe. However, I think one can push farther past this point, perhaps unnecessary to go as far as 7D has gone (46 megapixels fullframe) which I still think is not extreme (I'd say Sony 24 mp APS-C is over the top though), but as said some softness at pixel level is beneficial for post-processing and enlarging. 36 seems perfectly reasonable. I could go higher than that, say 40 and a nice 10 megapixel bin mode for handheld use.

I think it is not really good if the sensor is a limiter in the system. I rather have lens and diffraction being the clear limiters.

 

[torger]

To clarify the pp argument. I've been upscaling images some for prints. From that I've come to the conclusion that upsizing software with spline algorithms etc (photozoom etc) does not yield pleasing closeup results, and is also totally unnecessary if the input file has properly outresolved diffraction and/or lens. Then you can upscale to any size you want with normal bicubic scaling, never any artifacts. This is also beneficial in lens distortion correction, rotation and deconvolution sharpening.

Therefore I think it is healthy to have a high resolution sensor, which is pressing somewhat past what the lens can deliver. Based on APS-C measurment results extraploation I would think 36 MP is on the lower end of that scale, ~42 or so being optimal compromise and more than 48 would be unnecessary. Binning modes may benefit from a bit higher resolution though, 48 would give a nice 1/4 12 mp bin mode, I think 12 mp is ideal for hand-held photography (ok I know I said 10 mp in my previous post, but I just changed my mind... ).

 

[wickidwombat]

What I find clear from this conversation is that two 5D sized cameras are needed. And not because they are actually needed IMO, but because that's what people seem to be demanding.

One group wants 36mp FF images, and are saying they care about that more than bleeding edge ISO performance, which became possible by keeping the pixel size larger which requires lower mp total. This same group doesn't seem to care about video features either. Makes perfect sense then to just give them what they want... Maybe this can be done cheaper, perhaps not. I suspect there is very little if any actual cost savings here but perhaps there is a way.

Smaller form factor 5D sized body with ISO improvements and improved generation 2 video, in FF, and you have the other group covered nicely and I suspect this is also what all the 'new' customers that gathered to the 5D out of no-where during the video boom and others as well have come to expect out of a next-gen replacement.

Between the two, I'll take the second one. I don't care if it comes out in early 2012 or not for another year, that bit really doesn't make or break anything for me.

Problem solved, close the conversation. lol.

we can only dream, i would take 1 of each if that was the case and thats why they wont do it unfortunately because that will canabalise 1Dx sales

 

[AprilForever]

Canon could indeed make a baby 1DX, but must not make it a 5DIII or a 7DII. And they had better not merge the 5D-7D line into it!

They could call it the 9D. Or a 9DX. Or even 5Dx and 5Ds.

 

[Blaze]

Man, I would love a baby 1DX. Put the 1DX sensor in the body of a 7D (same AF, same FPS, same size and build quality) for less than $4k and I'd buy it in a heartbeat.

 

[Jettatore]

That's what I'm talking about Blaze, with video and FF stills for me though. And my money is reserved precisely for that. They can call it anything.