Periodic Table of Canon DSLR's

mgates · Dec 6, 2006

BJL said:
mgates said:

In some cases pixel size is different. In some cases not. There are APS sensors with the same pixel size as full frame.

Click to expand...

Virtually none, except the low level entry-level 6MP Sony sensors.

The 10D has the same pixel size as the 1Ds2. The D100 has larger pixels than the 1Ds2. The 1D2 has the same pixel size as the 5D. The D30 has bigger pixels than all of them.

BJL said:
BJL said:

A prime example is the true APS-C format dSLR the 1D2.

Click to expand...

The 1D and 1DMkII are very special cases, and not even in the
class of "APS-C" format DSLR's that I was talking about since its
sensor size is mid-way between the 1.5x and 1.6x "APS-C" formats
and 35mm format.
The 1D series is designed for a very high frame rate, which
requires (up till now) the compromise of relatively low resolution.

None of your subsequent arguments based on the 1DMkII hold when
applied to any other
"APS-C" DSLR's of 8MP and up: Canon EF-S, Nikon, Pentax, Sony. Even
less so to Olympus or Panasonic DSLR's with their even smaller
pixels.

There are various pixel sizes. Some larger some smaller. My point is that MPs are close to meaningless. More MPs trades worse high ISO resolution for more ISO 100 resolution. Over the range of ISO MPs differences even out. Format is the only thing that provides tangible overall IQ benefit.

I prefer less MPs in any given format. High ISO is more important to me than low. Low ISO shooting is easy. A 6MP full frame format camera with a clean ISO would be ideal. A 6MP full frame 5D would have ISO 1600 noise levels at around ISO 10,000. That is useful. More ISO 100 resolution is no big deal. Plus the speed boost in all phases. From shoot to delivery. That body would make more money faster than the 5D.

BJL said:
BJL said:

Any argument in favor of a smaller format misses what most professionals know.

Click to expand...

I am interested in discussing the SLR scene as a whole, not the
tiny fraction of it that consists of professionals using 35mm
DSLRs. (Even a great majority of professionals use smaller digital
formats, from the 1DMkII and D2Xs to mid-level options like the
D200 and 20D/30D.)

Nikon professionals do not have a choice. So it is not a fair question. Sensors are the last priority of smart professionals so that isn't a big deal. If full frame was the same price as small format then everyone would switch. It must have something going for it.

BJL said:
Something else that most professionals also know, and most amateurs
know too: size, weight and cost also matter, especially outside the
studio. Otherwise, 35mm would have been killed of long ago by
medium format.

If an APS chip weighs an ounce then a full frame chip adds one to two ounces.

mgates · Dec 6, 2006

BJL said:
One final comment on the fallacy of trying to decide whether a
smaller or larger format is better: many professionals and serious
amateurs use a mixture of different formats, both with film and
digital. When they use the smaller format, it is not to lower
equipment costs; it is because they judge that the smaller format
does a better job in that situation.

I shoot smaller and larger format dSLRs. But never small for IQ. IQ of full format is always better. For one thing it has small format inside its format.

BJL said:
It seems clear to me that this is true with digital as well as with
film: different situations are best handled by different formats,
and in particular, a lot of telephoto work seems to favor the
smaller DSLR formats (4/3, EF-S, DX) over the larger ones (35mm,
33x44mm, 36x38mm).

Yes. But within 35mm systems larger format is always better.

BJL said:
For example, my photography is mostly nature photography, and
mostly either well into the telephoto range or at apertures within
the reach of any DSLR format due to DOF needs. Within my lens cost
and weight constraints, a smaller DSLR format often does this
better, and almost always does it at least as well as a larger
format would.

You can crop large format 35mm when digital zoom helps.

SandyF · Dec 6, 2006

mgates said:
Foveon says their sensor is 4/3 format.

Have you read one word of the Yamaki interview? Sigma says no, they know their cameras and sensors used far, far better than you do. http://forums.dpreview.com/forums/read.asp?forum=1000&message=21155957

Search Google Yamaki, the link to full interview doesn't work here.Sigma says their cameras are not 4/3. What part of that don't you understand? The #s don't fit Four Thirds Standard. The size is different. The mount is different. The sensor is different. There is a new sensor 2006.

mgates · Dec 6, 2006

SandyF said:
mgates said:

Foveon says their sensor is 4/3 format.

Click to expand...

Have you read one word of the Yamaki interview?

Was he the one who went to jail for stealing $7 million from his own company? Correct me if I'm wrong. I think he also said Foveon's advantage is more silicon per pixel than the 5D. Has he seen his own camera's low light high ISO images?

Sigma is a rare exception to the Format is IQ rule of thumb. They tried to push the number of MPs too far for their small format. Foveon high ISO degrades too fast. It is enough to make smaller formats more attractive across the full ISO range.

If professional photographers had advised Foveon. We would tell them to use full frame silicon for 14MPs. High MPs in a small format is more like a point and shoot concept. Maybe they are trying to invent a new niche. Point and shoot MP-to-silicon ratios in a dSLR body. To be successful they will need to price their cameras and lenses lower than other entry level dSLRs. The D40 will hit $400.

SandyF said:
Sigma says no,

Foveon says 4/3 format. Foveon knows more about Foveon sensors.

SandyF · Dec 6, 2006

mgates said:
SandyF said:

mgates said:

Foveon says their sensor is 4/3 format.

Click to expand...

Have you read one word of the Yamaki interview?

Click to expand...

Was he the one who went to jail for stealing $7 million from his
own company? Correct me if I'm wrong.

You're wrong. On this and on 99% of what you write about Sigma and Foveon, give or take a 0.00% point or two.

light_reaction · Dec 6, 2006

mgates said:
The 10D has the same pixel size as the 1Ds2. The D100 has larger
pixels than the 1Ds2. The 1D2 has the same pixel size as the 5D.
The D30 has bigger pixels than all of them.

I think you are mixing up pixels and photodetectors here. Its not the same. You need three photodetectors for one RGB pixel. Does it matter if they are next to each other or stacked? No.

mgates said:
There are various pixel sizes. Some larger some smaller. My point
is that MPs are close to meaningless. More MPs trades worse high
ISO resolution for more ISO 100 resolution.

All else (e.g.sensor design) being equal.

Over the range of ISO

mgates said:
MPs differences even out. Format is the only thing that provides
tangible overall IQ benefit.

see above.

mgates said:
I prefer less MPs in any given format. High ISO is more important
to me than low.

This is your best argument. Its more important to YOU. Not "every smart professional".
How many ISO 1600 prints do people like Reichmann sell? Iso 800?ISo 400?
How many Gurskys are shot at ISO > 100?

O.
--
http://www.flickr.com/photos/ollivr/

mgates · Dec 6, 2006

SandyF said:
mgates said:

SandyF said:

mgates said:

Foveon says their sensor is 4/3 format.

Click to expand...

Have you read one word of the Yamaki interview?

Click to expand...

Was he the one who went to jail for stealing $7 million from his
own company? Correct me if I'm wrong.

Click to expand...

You're wrong. On this and on 99% of what you write about Sigma and
Foveon, give or take a 0.00% point or two.

Is this a different Yamaki fellow?
http://www.longislandpress.com/?cp=162&show=article&a_id=9542

But who was the one who said their 14MP small format sensor has more silicon per pixel than a full frame dSLR? Something like that. Do you have the quote I'm thinking of?

mgates · Dec 6, 2006

Tiffles said:
mgates said:

The 10D has the same pixel size as the 1Ds2. The D100 has larger
pixels than the 1Ds2. The 1D2 has the same pixel size as the 5D.
The D30 has bigger pixels than all of them.

Click to expand...

I think you are mixing up pixels and photodetectors here. Its not
the same. You need three photodetectors for one RGB pixel. Does it
matter if they are next to each other or stacked? No.

It matters. All else equal a horizontal layout uses three times as much silicon. Higher fabrication cost. A one to two stop light gathering advantage. Less noise at all ISOs.

light_reaction said:
light_reaction said:

There are various pixel sizes. Some larger some smaller. My point
is that MPs are close to meaningless. More MPs trades worse high
ISO resolution for more ISO 100 resolution.

Click to expand...

All else (e.g.sensor design) being equal.

Over the range of ISO

light_reaction said:

MPs differences even out. Format is the only thing that provides
tangible overall IQ benefit.

Click to expand...

see above.

light_reaction said:

I prefer less MPs in any given format. High ISO is more important
to me than low.

Click to expand...

This is your best argument. Its more important to YOU. Not "every
smart professional".
How many ISO 1600 prints do people like Reichmann sell? Iso 800?ISo
400?
How many Gurskys are shot at ISO > 100?

ISO 100 would look better too. The MP race is about marketing. More than usable IQ. MPs in a given format don't matter much. More MPs only trade more ISO 100 resolution for less high ISO resolution. Larger format increases IQ across the ISO range.

Press Correspondent · Dec 6, 2006

Jay Turberville said:
In order to make meaningful comparisons, the proper standard is to
compare a situation where you are trying to take the same picture.
That means that the FoV, DoF and shutter speed are all the same.

Jay Turberville said:
there is a HUGE overlap where equivalent images can be taken
and the large format has no intrinsic or practical advantage.

This is true. If we keep the the same distance and reduce both the format and lens size twice, then the FoV and visual proportions (like the nose size in portraits) would remain the same. To keep the same DoF we would need to increase the aperture by approximately 2 stops that would automatically keep the same amount of light and noise per (smaller) pixel, assuming the same total number of pixels (and adjusting ISO accordingly). The diffraction effects also would remain the same in relation to the pixel size.

So the true limitation of a smaller format is the maximum aperture of the lens. For example, the crop factor of Fuji F30 (named the best compact of the year) is 4.5 (or 4.7 considering the frame proportions) relative to 35mm. This implies a 4.5-stop (!) aperture difference for the same DoF. Its maximum aperture of f/2.8 is equivalent to f/13 in 35mm. Imagine yourself limited to the maximum of f/13... Of course, this is hardly an issue with EF-S where the aperture difference is only a stop and a quarter and high speed lenses are available.

Press Correspondent · Dec 6, 2006

With all due respect to both your arguments, please move them to a Sigma/Foveon thread. My thread is on Canon DSLR sensors. I would appreciate it. Thanks! - Alex

Press Correspondent · Dec 6, 2006

Press Correspondent said:
BJL said:

I don't see a disagreement here.
Larger vs. smaller or better vs. quicker is a choice one
makes based on the specific need.

Click to expand...

BJL said:
It seems clear to me that this is true with digital as well as with
film: different situations are best handled by different formats

Like I said, I don't see a disagreement here

Press Correspondent · Dec 6, 2006

BJL said:
Then, all that the larger photo-sites and sensors do is spread the
same light over a larger sensor area, getting the same total signal
(photons and photo-electrons) but more noise from sources like dark
current, and so somewhat worse overall S/N ratio.

Agreed. A larger frme of the same total pixel count has a disadvantage only when the needed DoF requires closing down the aperture and the amount of light becomes insufficient for low noise.

BJL said:
Larger sensors are only clearly better in certain cases, and a
worse choice in others.

Exactly my point. Although I would phrase it rather this way. There seems to be an overall practically optimal range of sensor sizes. Larger sensors within this range are better in some cases (IQ, etc.), smaller sensors within the same range are better in other cases (DoF, etc.). Both 35mm and EF-S seem to be within this range. Going beyond this optimal range in either way, sensors face various limitations and loose their advantages.

BJL said:
b) Where a larger format allows use of aperture diameters larger
than is feasible in a smaller format, due to the aperture ratio
required in the smaller format being too low for good optical
quality.

What benefit does the larger diameter provide if light is spread over a larger area anyway?

BJL said:
For some reason, medium format was never used like this, and did
not even bother to offer the bigger aperture lenses needed to have
a speed advantage over 35mm: does anyone understand why?

Weight? Why would medium format lenses have a speed advantage over 35mm? Bigger diameter does not imply bigger aperture ratio. Am I missing anything?

BJL said:
maybe it explains why 35mm digital is not popular for sports/action
photography: remember, the 1DMkII is not 35mm format.

Frame rate?

Jamie H. · Dec 6, 2006

All information about upcoming models - which is most of what is on this "list" - can pretty much be ignored. What youre left with is unimpressive to say the least.

--

http://jmhphoto.net/gallery

bearcat · Dec 7, 2006

Actually, the 5D was announced in August of 05

http://www.dpreview.com/reviews/specs/Canon/canon_eos5d.asp

I tend to think that the 5D successor will be announced at either next PMA or next summer in 2007. There is a rumor that seems to have some steam of there being two 5D successors:

1. similar to current model in terms of sensor and features, but priced lower
2. higher FPS, higher MP, etc for more $

If that isn't the case, then I would expect the next 5D to be more of a token upgrade. Maybe a 5DN with dust protection and a number of small improvements for a somewhat better price. Nothing too earth-shattering.

Just my thoughts. Thanks for putting together your table. The hard thing now, I think, is that the future of these cameras will become less linear. They will be working on different things like improved DR to differentiate them, whereas a couple of years ago the most visible changes to the average consumer were increased megapixels.

Press Correspondent said:
The approximate release dates are implied in the table. The usual
period between the 1.6x models is 18 months. Although, I believe
that EOS 40D will be an exception and should be released next
spring or only 12 months after EOS 30D. The reason is that EOS 30D
cannot successfully compete with EOS 400D or Nikon D80.

The period between the pro models can be at least 24 months. This
gives the expected dates of the next few cameras:

Press Correspondent said:
EOS 5D Mark II (16.5 mp) - Spring 2008

BJL · Dec 7, 2006

mgates said:
You can crop large format 35mm when digital zoom helps.

Not with the larger pixel sizes that are essentially universal in formats larger than EF-S and DX: have I not already pointed out how low pixel counts are if one crops from the 5D (or even 1DsMkII) to EF-S frame size?

You seem to committed to a vision of DSLRs in larger formats like 35mm with the same pixel sizes as in EF-S, DX and 4/3 cameras, but I see no evidence of likelihood of that happening, and have repeatedly stated optical facts that make it likely that smaller formats will continue to offer higher absolute resolution in lp/mm.

BJL · Dec 7, 2006

My claim is that at any given time , sensor designs in EF-S and mainstream "APS-C" and smaller formats have and will probably continue to have smaller pixel sizes that sensors in larger formats like 35mm.

None of your examples below are relevant: The 10D, D100 and D30 are discontinued products with sensors designs older than that of the 1Ds2, and the 1D2 is not of mainstream "APS-C" format, and is also a special case, sacrificing resolution for high frame rate. Much of this I already said in the post to which you are replying, so I will not say more.

mgates said:
The 10D has the same pixel size as the 1Ds2. The D100 has larger pixels than the 1Ds2
The 1D2 has the same pixel size as the 5D.
The D30 has bigger pixels than all of them.

mgates said:
mgates said:

Something else that most professionals also know, and most amateurs know too: size, weight and cost also matter, especially outside the studio. Otherwise, 35mm would have been killed of long ago by medium format.

Click to expand...

If an APS chip weighs an ounce then a full frame chip adds one to two ounces.

I think you know that I am talking about total size and weight including the longer focal length lenses needed with larger pixels and larger formats.

Jay Turberville · Dec 7, 2006

Press Correspondent said:
Press Correspondent said:

there is a HUGE overlap where equivalent images can be taken
and the large format has no intrinsic or practical advantage.

Click to expand...

This is true. If we keep the the same distance and reduce both the
format and lens size twice, then the FoV and visual proportions
(like the nose size in portraits) would remain the same. To keep
the same DoF we would need to increase the aperture by
approximately 2 stops that would automatically keep the same amount
of light and noise per (smaller) pixel, assuming the same total
number of pixels (and adjusting ISO accordingly). The diffraction
effects also would remain the same in relation to the pixel size.

Right. The main problem with this analysis and my generalization is that sensor technologies are typically not equal. There can be signifcant variation.

Press Correspondent said:
So the true limitation of a smaller format is the maximum aperture
of the lens. For example, the crop factor of Fuji F30 (named the
best compact of the year) is 4.5 (or 4.7 considering the frame
proportions) relative to 35mm. This implies a 4.5-stop (!) aperture
difference for the same DoF. Its maximum aperture of f/2.8 is
equivalent to f/13 in 35mm. Imagine yourself limited to the maximum
of f/13... Of course, this is hardly an issue with EF-S where the
aperture difference is only a stop and a quarter and high speed
lenses are available.

Well, yes and no. The reality is that most people shoot with kit lenses that have maximum apertures between f/3.5-f/5.6. Add your stop and change for the smaller format and you have the typical equivalent range of f/5.6 - f/9 or so. Certainly better than the F30, but still not great. Of course, you did make the point about the option of faster lenses, but these are typically very expensive options. So we shouldn't be surprised if the F30 (which you can get for well below $300 U.S. if you shop well) doesn't quite measure up. Just about every fast lens other than a 50mm prime will be more expensive than the entire Fujifilm F30 camera.

BTW, the typical compact 35mm with a telescoping zoom lens had a very slow lens of around f/10. This is just one of the prices that must be paid for having a small and compact camera. It is largely dictated by the reality of optics as you outlined above. A small sensor needs the same entrance pupil diameters to deliver the same picture. So a 9mm "normal" on an F30 still needs a 27.7 mm entrance pupil to do the same thing that a 50mm f/1.8 lens can do on a 135 format camera. It would need to be a 9mm f/0.35 - which may not even be possible and almost surely isn't practical. And even if ti was, it would pretty much trash the goal of being compact.

And finally, the F30 sensor appears to also be a very good example of how we don't necesarily get the same technology in all formats. It seems to have a clear advantage that isn't found in sensors from other makers. So while it is good to understand the relationships of light collection, aperture, DoF and so forth from the standpoint of equal sensor capabilities - it is also important to understand the limitations.

--
Jay Turberville
http://www.jayandwanda.com

BJL · Dec 7, 2006

Press Correspondent said:
If various lens aberrations are somewhat proportional to diffraction and diffraction depends only on the aperture, but not on the format, then the overall lens quality should not improve simply because the lens with all its elements was reduced in size.

Lens aberrations are not vaguely proportional to diffraction; aberrations and diffraction follow opposing trends with aberrations effects increasing with larger apertures ,while diffraction increases as the aperture gets smaller (higher f-stop).

I explained a previous post why a linear downsizing of a lens design from one format given one for a smaller format of the same FOV, same minimum f-stop and aberrations effects at equal f-stop also scaled down in proportion to the format size, giving a smaller length scale that can be resolved at a given f-stop.

I strongly suggest that you look around at the hundreds of MTF graphs and lens resolution tests for good lenses in various formats from 4/3 to EF-S to DX to 35mm to medium format: the overall trend of increasing resolution as format size decreases is very clear, though of course it is possible to select examples where a (mid-priced wide-ranging) zoom lens for a smaller format fails to have higher resolution.

Press Correspondent said:
Most full frame lenses don't do too well in the frame corners and will do much better with 1.3x.

This complaint is greatly over-stated: good 35mm format telephoto lenses in particular have little fall of in image quality in the corners. And surely when there are such problems, the best solution would be occasional cropping according to the needs of the lens and the composition. Slightly soft corners are irrelevant in many images, where than part of the image is OOF anyway. It makes little sense to impose a fixed crop to slightly more than half of the image area on every image taken with every lens.

Press Correspondent said:
Press Correspondent said:

hampering wide angle coverage

Click to expand...

Not really. 1.3x has a much wider coverage than 1.6x with full frame lenses.

But 1.6x has EF-S lenses for wide angle coverage: it is ridiculous to judge its wide angle options under the constraint of using 35mm format lenses.

Press Correspondent said:
Many people avoid EF-S lenses even with 1.6x bodies

A few do, but I would bet that the vast majority of EF-S body users use EF-S lenses for standard zooms and wide angles.

Press Correspondent · Dec 7, 2006

Jamie H. said:
All information about upcoming models - which is most of what is on
this "list" - can pretty much be ignored. What youre left with is
unimpressive to say the least.

Thanks for your comments, Jamie! The purpose of this table is to notice and reveal the pattern and use it to envision new models. For example, all 1.6x Canon sensors are on a straight line of a 20-pixel-per-mm linear increase with the statistical margin of error for the next predicted sensor of only 3%. There is also a megapixel pattern as well as some others. The thread also has generated many interesting and educational discussions. Many people find this interesting and enjoyable, but no one has to. I do not see you offering anything more comprehensive, so if this does not interest you or you are unable to notice the periodic patterns, please feel free to ignore it and move on. - Alex

Dan Tong · Dec 7, 2006

Your table is very well done and fun to think about.

Don't pay attention to the nitpickers and naysayers. Some people are incapable of not being critical and belittling in general.

Great Job,

Dan

Periodic Table of Canon DSLR's

Leading Member

Leading Member

Forum Pro

Leading Member

Forum Pro

Veteran Member

Leading Member

Leading Member

Senior Member

Senior Member

Senior Member

Senior Member

Senior Member

Leading Member

Veteran Member

Veteran Member

Forum Pro

Veteran Member

Senior Member

Leading Member

Similar threads

Keyboard shortcuts