[Blue]

Disagree about there being a "fundamental" relationship between CoC and sensor size. Yes, the sensor (and frame size) does affect assumptions about the magnification factor for final viewing, but CoC ultimately is a subjective measure, and not everyone view images at the same magnification factor or viewing distance (or have the same visual acuity).

There is no "constant" that would ever be appropriate for all cameras. If your point is that all these discussions about sensor sizes and image quality have therefore missed the point, then I agree.

What is the CoC of the *istD, K20d, and K-5?

[pinholecam]

I would be curious to see what percentage of total camera sales for Nikon or Canon are their Full Frame models. I just can't buy into the survival of Pentax being based on a Full Frame camera unless I could see statistical sales data that indicated Canon/Nikon's Full Frame sales were 25% or more of their total sales. In any event, Pentax has the 645D.

I'm guessing that most camera buyers have no clue about sensor sizes. I have a friend with the 5D MkII. She has no clue whatsoever about the camera or sensor. She just had money to burn and was given a good sales spiel. Of course she doesn't really like the camera because it's so big. She does really like my EP2. So anecdotally, I would say Full Frame isn't the issue with the average consumer.

Also does Pentax have the professional support for people who actually make their living with photography and therefore have no issue spending that kind of money on camera demand? I don't think so.

Not being critical but I wonder how many of the Full Frame advocates have the K5. I own the K5 and good Pentax lenses and I'm not clamoring for a Full Frame camera. If someone isn't willing to purchase the best ASPC sensor camera Pentax makes, why should make Pentax make the assumption that someone will shell out $2500 or more for a Full Frame body?

Maybe Ricoh will develop one for Pentax. I'm doubtful, but if they do and it's $1500 or less I'll buy one. If not I'll get a GXR which is a really great camera and covers all the focal lengths I need at 28 and 50mm.

That said, I do think the Q is too expensive. I'm remain skeptical that it will be successful.

I think a large % will be eagerly waiting by ring side to comment on it being too expensive; loaded with bugs; performance not as good as their *istD; etc

Having the K5 now, other than a potentially larger/brighter viewfinder for Manual Focus and less DOF, I'm not even sure what a FF will bring to the table in terms of improving on the pleasure of picture taking.
On the k5, DR, ISO, resolution is already very good indeed. The only improvement I need from it is just better and more accurate AF and better PTTL.

[falconeye]

Disagree about there being a "fundamental" relationship between CoC and sensor size. [...]
There is no "constant" that would ever be appropriate for all cameras.

This (your) opinion may explain why (if) your point of view in DoF discussions differs from other's.

IMHO, you are:

- right that the "constant" is subjective and is a constant for a given person only. No constant will ever be appropriate for all individuals. So, there is a "Tham constant" as there is a "Zeiss constant" (1/1730) or a "Lumo constant" (1/2203) to be used in the Zeiss formula which, as a formula, is strict. The constant does actually dependent on the purpose of an image, e.g., if it is for print (an ad), gallery wall, paper print-out etc. (what will be size/viewing-distance of the printed/displayed image). Also note that the constant always applies to the image circle after cropping. So, with cropping in mind, the constant will be another one, to be divided by the intended crop factor during post processing before printing. At the end of the day, this constant expresses an individual eye's capability for angular resolution and wide angle perception. But there is a minimum for such a constant where smaller fractions make no sense whatever be the individual person or purpose. I believe my factor (1/2203) to be pretty close to such a limit.

- wrong in assuming that such constant would be camera-dependent. It is not. Your opinion may differ on that last point. But IMHO this is a fact and not an opinion. The only exception I can imagine would be a camera which delivers special purpose images only.

BTW, all online DoF calculators assume the constant to be ~1/1400 which results in CoC values of 30µm for full frame and 20µm for APS-C cameras like the K-5 or *istD. The Zeiss constant delivers 17µm for APS-C, the "Lumo" constant 13µm.

[falconeye]

Not being critical but I wonder how many of the Full Frame advocates have the K5. I own the K5 and good Pentax lenses and I'm not clamoring for a Full Frame camera. If someone isn't willing to purchase the best ASPC sensor camera Pentax makes, why should make Pentax make the assumption that someone will shell out $2500 or more for a Full Frame body?

Stanley, you make some valid points. I too do have a K-5 with good lenses, so I qualify to answer to your question

You are right in that the K-5 has a stunning image quality which only few full frames top (e.g., not the 5DmkII IMO). But ...

The difference between an average and top-notch image from the K-5 can be frustrating sometimes (in IQ I mean). That's because it is so easy (for the K-5) to be sup-optimum for something, like AF precision, shake, moving subject etc.

And that's my point: Tolerances in a FF are higher, AF inaccuracies typically are lower (and explainable so), shake can be smaller, esp. with a brighter and larger lens, etc. AF accuracy is my main argument in favour of FF actually. Lens quality my second. Brighter VF my third (although not strictly depending on the sensor size). Potential DR my fourth (where the K-5 shines, so not an argument for now). And high ISO performance my fifth (funny enough, it typically comes first in discussions). And possible DoF my sixth, if at all.

As a curiosity, some brilliant Canon FF L lenses are cheaper (and at least as good) than the equivalent Pentax DA* lenses, like the 70-200/4L vs. DA*50-135/2.8. So, lower lens cost can be an argument for FF as well (and explainable so). Not so for Nikon and Sony though. BTW, this is why Sony fails in the FF segment: lenses too expensive.

I agree in a last point Stanley makes: Current FF cameras are too bulky. They're not Pentax to start with (nobody currebtly beats Pentax when it comes to compact SLRs -- from MX over *istDs to K-r/K-5). Note that Pentax includes the SR mechanism and WR seals in their bodies. And then FF cameras are made bulky to justify the price premium. But this is Pentax' chance: A compact FF 35MP (5µm) SLR for $1999 MSRP (there is no technical reason why it should be more expensive). All DA lenses would perform like on a K-7 and if the image circle is better than APS-C, better.

And I don't see how this camera (if executed right) could fail on the market.

[falconeye]

As this is a thread about the Q ...


I wonder if Ricoh has influenced the Q project as we all assume. After all, the Q would most likely have gotten a port to attach the Ricoh VF-2 EVF if Ricoh exercised any influence, I guess. BTW, it now is a pitty the Q lacks such an EVF option.

[jsherman999]

stanley, you make some valid points. I too do have a k-5 with good lenses, so i qualify to answer to your question

you are right in that the k-5 has a stunning image quality which only few full frames top (e.g., not the 5dmkii imo). But ...

The difference between an average and top-notch image from the k-5 can be frustrating sometimes (in iq i mean). That's because it is so easy (for the k-5) to be sup-optimum for something, like af precision, shake, moving subject etc.

And that's my point: Tolerances in a ff are higher, af inaccuracies typically are lower (and explainable so), shake can be smaller, esp. With a brighter and larger lens, etc. Af accuracy is my main argument in favour of ff actually. Lens quality my second. Brighter vf my third (although not strictly depending on the sensor size). Potential dr my fourth (where the k-5 shines, so not an argument for now). And high iso performance my fifth (funny enough, it typically comes first in discussions). And possible dof my sixth, if at all.

As a curiosity, some brilliant canon ff l lenses are cheaper (and at least as good) than the equivalent pentax da* lenses, like the 70-200/4l vs. Da*50-135/2.8. So, lower lens cost can be an argument for ff as well (and explainable so). Not so for nikon and sony though. Btw, this is why sony fails in the ff segment: Lenses too expensive.

I agree in a last point stanley makes: Current ff cameras are too bulky. They're not pentax to start with (nobody currebtly beats pentax when it comes to compact slrs -- from mx over *istds to k-r/k-5). Note that pentax includes the sr mechanism and wr seals in their bodies. And then ff cameras are made bulky to justify the price premium. But this is pentax' chance: A compact ff 35mp (5µm) slr for $1999 msrp (there is no technical reason why it should be more expensive). All da lenses would perform like on a k-7 and if the image circle is better than aps-c, better.

And i don't see how this camera (if executed right) could fail on the market.

__++__

.

[Christine Tham]

- wrong in assuming that such constant would be camera-dependent. It is not. Your opinion may differ on that last point. But IMHO this is a fact and not an opinion. The only exception I can imagine would be a camera which delivers special purpose images only.

Please don't claim your unsubstantiated assertion as "fact."

In my "opinion", the CoC can also be constrained by factors such as lens resolution and pixel density, both of which are camera dependent.

And by the way, not ALL online DOF calculators assume a relationship between sensor size and CoC. My DOF calculator allows input of CoC (which is an individual subjective measure) directly.
https://spreadsheets.google.com/spreadsheet/ccc?key=0Ahto5-ljsK44dDFYazFpVmh...NRV3c&hl=en_US

[falconeye]

Please don't claim your unsubstantiated assertion as "fact."

Why not?
Things such as facts do exist. They are rare and difficult to prove. But do exist. And to the best of my knowledge, the before-mentioned "assertion" is a fact. In modern times though, there is somebody for every fact who denies it. So, facts may have become difficult to recognize.

This is where I try to help by calling facts what they are: facts. I don't do that often because, as I said, facts are rare. Because you like to challenge me: I will not provide prove that things I call fact are facts. So, it is my opinion only that a fact is a fact. It is this opinion I wanted to share on this forum.

If you are interested to understand why I call it a fact, I am ready to do so by PM. But IMHO it shouldn't go into this thread.

BTW, your DoF calculator doesn't provide means of input for image diagonal or Zeiss constant. So, direct input of coc is the only other choice left. As such, your DoF calculator is very good. IMO though, an even better version would replace coc by input fields for image diagonal (or sensor dimensions) and Zeiss constant. By allowing the Zeiss constant to be input, you preserve all freedom your calculator currently offers.

In my "opinion", the CoC can also be constrained by factors such as lens resolution and pixel density

It can't.

But I think I understand what you try to say here. That what you define to be acceptable softness near the focal plane depends on the sharpness in the focal plane. This is a valid point when defining an images's potential for pixel-peeping and cropping and I do actually agree. But ... the corresponding measure is not called coc and the resulting depth is not called DoF. You need to invent new terms for what you try to express. Or otherwise, you create confusion and nothing else.

DoF (Depth of Field) expresses one thing and only one thing: The human eye's inability to perceive softness off the focal plane when watching an image from a distance allowing to perceive the entire image (the distance a human allowed to move chooses to watch an image, like in a gallery). You can't abuse DoF to mean something else you want to say. You need to give it its own name then.

[Mistral75]

I would be curious to see what percentage of total camera sales for Nikon or Canon are their Full Frame models. (...).

Between 3 and 5 per cent. in units, around 10 per cent. in turnover.

[fuent104]

"But I think I understand what you try to say here. That what you define to be acceptable softness near the focal plane depends on the sharpness in the focal plane. This is a valid point when defining an images's potential for pixel-peeping and cropping and I do actually agree. But ... the corresponding measure is not called coc and the resulting depth is not called DoF. You need to invent new terms for what you try to express. Or otherwise, you create confusion and nothing else."

Falconeye, what you are describing here almost sounds like a definition of depth of focus. This, of course, can be measured. In motion pictures, which are enlarged hugely, depth of focus can really become an issue of the equipment is not maintained properly. This often surfaces in lower budget films, as the production circumstances are perfect for limited depth of focus. Wide aperture + wide angle lens = perfect recipe for soft images if the film plane isn't exactly where it needs to be (i.e. depth of focus is shallow and the film plane is outside of the acceptable area).

[falconeye]

Falconeye, what you are describing here almost sounds like a definition of depth of focus

"Almost" is the problem here.

I recommend Depth of field and diffraction for a further read.
With special mention of the section how coc is defined (0.01" on an 8x10" print) which translates (when doing the math) to a Zeiss constant of 1/1280.

This article contains a link to an excellent source of information ( http://www.stanford.edu/class/ee368b/Handouts/09-HumanPerception.pdf ) and concludes that this constant doesn't actually apply to a person with 20/20 vision who can resolve 3x as well. The Zeiss constant then would read 1/4250. So, anything between 1/1280 and 1/4250 can be justified. For more than 8MP this means that coc is always larger than a pixel. In the 1930's, industry was satisfied with the lower standard which like Christine I think is too low for today's standards.

Like I said above, I personally use 1/2203 which seems to be the current consensus for "sharp enough" (Full HD spec).

I agree with Christine in one point: if a camera doesn't actually even resolve the coc within the focal plane (which is possible for a small Zeiss constant and old technology) then the coc should be made larger. Which introduces a camera dependency. But only for old or bad cameras paired with small values for the Zeiss constant. Rather than changing the Zeiss constant, one would change camera or lens though. Because then it doesn't obviously satisfy a person's criteria for acceptable sharpness.

[fuent104]

Interesting. I have to admit, I have a hard time grasping these numbers in anything other than an academic sense.

[Christine Tham]

Oh dear. This discussion has yet again turned into a "nerdy" semantic discussion based on numbers, rather than photography.

Rather than respond to an "accusation" that I am somehow arbitrarily "redefining" what DOF and CoC means, I would prefer to refer people to the following, which I believe is completely consistent with how I deal with those concepts:
Depth of field - Wikipedia, the free encyclopedia

I actually believe that CoC can, and should be empirically derived (since it is unique to each individual, and photography is about taking actual photographs). It's easy enough to use something like a Spyder Lenscal to measure apparent DOF, and from that solve the DOF equation for CoC. That would then give a CoC that is appropriate for the individual, camera, lens and viewing conditions.

In this way, there no need to rely on simplistic and naive formulae such as "Zeiss" or "Lumo" (there is no corresponding "Tham" because "Tham" does not believe such formulae is appropriate).

[Edit: the other advantage of deriving the CoC empirically rather than based on assumptions is that: if you don't have the actual camera in your hands, you can't calculate DOF since you don't know what the CoC is. Maybe that will prevent people on internet forums from making wild speculations on the image quality of cameras that haven't been released yet. Idealistic of me perhaps, but I can only live in hope.]

[jsherman999]

Oh dear. This discussion has yet again turned into a "nerdy" semantic discussion based on numbers, rather than photography.

Oh dear. IIRC, you brought up the CoC issue initially in an attempt to imagine more DOF control potential in that tiny Q sensor.

[Edit: the other advantage of deriving the CoC empirically rather than based on assumptions is that: if you don't have the actual camera in your hands, you can't calculate DOF since you don't know what the CoC is. Maybe that will prevent people on internet forums from making wild speculations on the image quality of cameras that haven't been released yet. Idealistic of me perhaps, but I can only live in hope.]

The CoC is not going to be derived from any new variables introduced by the Q, and we know what happens to DOF control on larger tiny sensors, like 1/1.7.... What new revelation are we supposed to be waiting for?


.

[Blue]

Oh dear. This discussion has yet again turned into a "nerdy" semantic discussion based on numbers, rather than photography.

Rather than respond to an "accusation" that I am somehow arbitrarily "redefining" what DOF and CoC means, I would prefer to refer people to the following, which I believe is completely consistent with how I deal with those concepts:
Depth of field - Wikipedia, the free encyclopedia

I actually believe that CoC can, and should be empirically derived (since it is unique to each individual, and photography is about taking actual photographs). It's easy enough to use something like a Spyder Lenscal to measure apparent DOF, and from that solve the DOF equation for CoC. That would then give a CoC that is appropriate for the individual, camera, lens and viewing conditions.

In this way, there no need to rely on simplistic and naive formulae such as "Zeiss" or "Lumo" (there is no corresponding "Tham" because "Tham" does not believe such formulae is appropriate).

[Edit: the other advantage of deriving the CoC empirically rather than based on assumptions is that: if you don't have the actual camera in your hands, you can't calculate DOF since you don't know what the CoC is. Maybe that will prevent people on internet forums from making wild speculations on the image quality of cameras that haven't been released yet. Idealistic of me perhaps, but I can only live in hope.]

The problem with using the Spyder Lenscal to calculate DOF is that you would endup making estimations on near and far focus. One of those would be needed to solve for H in order to use the hyperfocal formula to solve for CoC.

Depth of Field Equations

In the end, I doubt the CoC will be any more accurate than those listed here.

Circles of Confusion for Digital Cameras

Canon Powershot G12 .007
Canon 60D (aps-c 1.6x) .019
Canon 5D Mk II .03
Canon Powershot S95 .006
Nikon P7000 .006
Nikon D70s (aps-c 1.5x) .020
Nikon D7000 .020
Nikon D700 .030
Olympus e-p1 (m4/3) .015
Olympus e-420 (4/3) .015
Panasonic LX5 (1/1.63") .006
Panasonic GF2 (m4/3) .015
Pentax *istD (aps-c 1.5x) .02
Pentax K-5 .02
Pentax W90 .005
Pentax Q .005