[Blue]

YES! Some online calculators use the simplified equations which become increasingly inaccurate as distance gets closer than hyper-focal distance. I think someone has confused that fact with "as focus distance approaches hyperfocus distance, the traditional DOF formulas fail", something I've never heard before. Maybe I need to get out more

This online DOF calculator Online Depth of Field Calculator uses the formulas presented at Depth of field - Wikipedia, the free encyclopedia

This is pretty good reading: The DOF equations

And I'm posting this link again, to pertinent Zeiss pdf: Depth of Field and Bokeh

A worthwhile exercise in my experience is to write a spreadsheet macro...



Someone mentioned the relativity of DOF calculations to final print size and viewing distance. (Bravo!!!) It is the Coc number in the DOF equations that accounts for the DOF relativity to sensor size. DOF calculations for different size sensors with the same lens need correct Coc entry to give identical DOF numbers at proper distances -- i.e. that calculation would reveal the settings for: with the same lens on two different sensor sizes, produce 8x12" prints from each camera, with identical framing and identical DOF.

Good post . . . many always overlook that Coc value relative to sensor/film which is "interesting" in and of itself.

---

Last edited by Blue; 05-19-2011 at 07:05 AM.

[jstevewhite]

.


Excellent job, OOF transition and background look very 'real' too, not shopped.

Now, if they only provided that feature in camera as bracketing option...


.

Thanks!

LOL! I wouldn't be too certain that won't show up as an in-camera option soon enough. At least "auto stitching". My iPhone does it, and with AutoStitch Pano, it will produce an 18MP image that's actually of amazingly high quality. Like, a 140 degree pano with my iPhone vertical holds more detail and less distortion, and is nearly the same IQ as a pano crop the same width from my 10-20 Sigma on my K20D (I haven't compared with the K-5 yet).

There are patents for noise reduction technology that uses multiple reads of the sensor; so, for 1/1000 exposure, your sensor might get sampled (electronically) 3x times @1/5000 second. Of course, they can't produce very large MP chips that way right now, but it's a matter of time - three, four, five years - before we could see an APS-c sensor that has an incredible SNR all the way up to ISO ridiculous.

[Blue]

I loved my *istDL, and I got some shots with it that I love, but I can tell the difference between the K-5 cropped to 8x10 and the *ist DL cropped to 8x10 (in subjects with a lot of detail). At 11x14 I can look at a single high-detail image (like one of my bug or flower macros) and tell you which camera it came from. That's my measure of "important difference" in IQ.

If I print full-frame on 8x10, there's no difference, because the *ist DL just makes 300 dpi for 10" - but I can crop the daylights out of the K-5 image and get the same quality of result.

The *isD is a different body than the DL. Secondly, you are assuming that I crop everything. I generally shoot with the intentions of cropping as minimally as possible. I am more likely to frame the insect in the viewfinder and re-size the shot regardless if the shot is made with the K20d or *istD.

I don't think the idea behind a full frame sensor is to crop the crap out of it. With the K-5, you are talking about 16 mp to 6 mp for the *istD. Your overall logic is a bit flawed because anything can be cropped beyond good image quality, even a Hassy or 645D.

[jstevewhite]

The *isD is a different body than the DL. Secondly, you are assuming that I crop everything. I generally shoot with the intentions of cropping as minimally as possible. I am more likely to frame the insect in the viewfinder and re-size the shot regardless if the shot is made with the K20d or *istD.

I don't think the idea behind a full frame sensor is to crop the crap out of it. With the K-5, you are talking about 16 mp to 6 mp for the *istD. Your overall logic is a bit flawed because anything can be cropped beyond good image quality, even a Hassy or 645D.

Woah, there, Big Blue! I'm not gettin' all up in your grill! There was no "logic" to be "flawed", as I made no claim or assertion; I was describing *my* experience with *my* cameras and the relationship between 6mp and 16mp when printed full-frame (that is, without cropping, with big white borders on the sides, and thin ones at top and bottom of a vertical image) or when printed 8x10 borderless (cropped), not about a full-frame (24x36mm) sensor.

An image from a 24x16mm sensor *must be cropped* to print 8x10 borderless. I made no assumptions about you "cropping the crap out of your images". (same for 24x36mm, also, incidentally)

When you're shooting Hassy, you almost always have to crop, as clients don't buy 6x6cm images, they buy ~6x4.5cm images for a single page bleed, or 6x
~3.75cm for a two page bleed, and they EXPECT to have room to drop in text. You commonly shoot for multiple possible crops. Some shoot multiple crops, but the final result still out-resolves the final media either way, and you can end up where the AD likes *that* image, but *this* crop. I did, however, print many 10x10in images for gallery display.

Appropriate cropping limits are controlled by your output media. If you want to make borderless 8x10s @ 300dpi, you can't do it with a 6MP 3:2 sensor. You can print 10x6.5 images @300dpi. With 16mp, you can print ~16x10inch @300dpi. Which means I can crop down to about half the frame and make a full-quality 300dpi image from the 16mp K-5 sensor. I usually only do this with things I can't get closer to, like birds (since I've only got a 500mm and no blind to hide in); bugs I shoot full-frame.


There are many brands that are just as tasty without all the caffeine, you know

[Blue]

Woah, there, Big Blue! I'm not gettin' all up in your grill! There was no "logic" to be "flawed", as I made no claim or assertion; I was describing *my* experience with *my* cameras and the relationship between 6mp and 16mp when printed full-frame (that is, without cropping, with big white borders on the sides, and thin ones at top and bottom of a vertical image) or when printed 8x10 borderless (cropped), not about a full-frame (24x36mm) sensor.

An image from a 24x16mm sensor *must be cropped* to print 8x10 borderless. I made no assumptions about you "cropping the crap out of your images". (same for 24x36mm, also, incidentally)

When you're shooting Hassy, you almost always have to crop, as clients don't buy 6x6cm images, they buy ~6x4.5cm images for a single page bleed, or 6x
~3.75cm for a two page bleed, and they EXPECT to have room to drop in text. You commonly shoot for multiple possible crops. Some shoot multiple crops, but the final result still out-resolves the final media either way, and you can end up where the AD likes *that* image, but *this* crop. I did, however, print many 10x10in images for gallery display.

Appropriate cropping limits are controlled by your output media. If you want to make borderless 8x10s @ 300dpi, you can't do it with a 6MP 3:2 sensor. You can print 10x6.5 images @300dpi. With 16mp, you can print ~16x10inch @300dpi. Which means I can crop down to about half the frame and make a full-quality 300dpi image from the 16mp K-5 sensor. I usually only do this with things I can't get closer to, like birds (since I've only got a 500mm and no blind to hide in); bugs I shoot full-frame.


There are many brands that are just as tasty without all the caffeine, you know

Again you are making assumptions about caffeine. I just called it the way I saw it. If you are shooting insects completely framed on the images, you shouldn't be cropping the snot out of them. I was merely suggesting that cropping should be a tool, not a crutch.

[jstevewhite]

Again you are making assumptions about caffeine. I just called it the way I saw it. If you are shooting insects completely framed on the images, you shouldn't be cropping the snot out of them. I was merely suggesting that cropping should be a tool, not a crutch.

Forgive my caffeinated assumptions... I'm not certain insects have mucos membranes, but I completely agree with your last line.

[Blue]

Forgive my caffeinated assumptions... I'm not certain insects have mucos membranes, but I completely agree with your last line.

I guess you have never encountered spittle bugs.

[jstevewhite]

I guess you have never encountered spittle bugs.

...aaaand hope I never do. Unless they're really colorful.

More on-topic... Macro is a place where APS-c excels and pixel density is king. No matter how you cut it, though, I love the color that comes out of these Pentax cameras and lenses:

[Marc Sabatella]

Not *all*; I only beat that particular horse so much because it really is *less* DOF, all things being equal, not "more control over DOF" as it's been characterized.

I think perhaps the issue here is that you mean something entirely different by "less DOF" than everyone is assuming. Your wording makes it sound like every single shot taken on APS-C will have more DOF than even the deepest DOF shot on FF. This is, I hope you'll agree, completely false. What is actually the case is that for most of the shots that exist to be taken, APS-C and FF are capable of getting exactly the same DOF (as well as framing and everything else), simply by choosing the correct focal lengths and apertures. The difference is only at the extremes. At the wide open end of the scale, FF provides the *option* of shallower DOF at any focus distance, assuming your FF lens goes to the same maximum f-stop as your APS-C lens. And at the fully stopped down and/or hyerfocal distance end of the scale, APS-C provides the *option* of more DOF, assuming your FF lens doesn't have an extra f-stop at that end, and/or assuming your hyperfocal calculations are more accurate than the ones that use the usual DOF formulas (which as I have observed say the hyperfocal distances are the *same* for otherwise equivalent focal lengths and apertures). I'm still open to the possibility that your calculations are more accurate; right now I'm simply unconvinced.

But assuming your calculations are more correct than Atkins', really, you are talking about "DOF control", just as I (and others) are. For *most* shots, you can get the *same* DOF with either system. FF allows you to squeeze some extra shallow DOF at one end of the scale should you so wish, and APS-C (if your calculations are correct) allows you to squeeze some extra deep DOF at the other end of the scale should you so wish, but there is more overlap than difference.

So - again taking your calculations at face value - it would not be correct to say that either system gives you more DOF control overall. They both allow the *same* DOF control in the heart of the aperture range and focus distance range, but FF provides additional options at one extreme and APS-C may possibly provide additional options at the other extreme.

Hyperbole about image destruction ( the LTDs are often still sharper at f16 APS-c than many other commonly used lenses are in the first couple of stops, so it *is* hyperbole ) notwithstanding, you don't get *more* DOF with FF without *changing the standards you're measuring by*.

I would challenge the assertions that the limited are sharper at f/16 than other commonly used lenses would be in the first couple of stops, unless you mean, consumer zooms shot wide open, or literally the first two stops of some sub-f/2 primes. Of course, since it's the sub-f/2 primes that one might turn to to get the extra shallow DOF that gives FF its advantage at that end of the scale, I would agree that at face value, it seems one could similarly say that getting that extra shallow DOF on FF comes at a price optically, just as getting the extra deep DOF on APS-C does.

Except that when you consider many of the common cases, this just isn't so. A typical FF portrait lens shot at f/2.4 is likely to be quite sharp already - certainly sharper than a typical APS-C portrait lens shot at the equivalent aperture of f/1.7. So even opening up your FF portrait lens to f/2 or more - going to DOF places you would not be able to get to on APS-C at all - your image is probably still as sharp as your APS-C image was at its maximum aperture. Obviously, this depends on the specific lenses involved, but the point is there is no basic physical principle at play that limits the IQ of a lens at large aperture.

However, diffraction *is* a physical principle that limits IQ at the other end of the DOF scale. The effect is quite real, no matter how sharp your lens is, and this is not taken into account by the simplified/idealized hyperfocal distance formula you have been relying on.

Here's a way of putting into plain English. DOF can be described as the zone in which points in the actual scene will be resolved to circles no larger than the CoC for the sensor size, print size, viewing distance, and visual acuity of the viewer. You've been using a typical value of 0.03mm for CoC on FF, which corresponds to roughly 0.02mm on APS-C. So you're saying that a point on an image is sufficiently sharp to you if it can resolve to a circle no larger than 0.02mm. Well, if my understanding is correct, diffraction at f/22 on APS-C is already such that *no* point - not even the points right at the plane of focus - can possibly resolve to a circle that small. And long before you reach that aperture, diffraction is going to take its toll on sharpness, such that the sharpest points in an f/16 image are already less sharp than the sharpest points in an f/11 image. Which in turn may mean that points toward the front or rear of the zone of acceptable sharpness that were allegedly more in focus at f/16 than at f/11 because of ordinary DOF calculations will actually be *less* sharp due to diffraction. No, I haven't worked through all the math. But it's pretty clear that your hyperfocal formula is blowing this off entirely.

However all of this actually pans out, though, there is another aspect to it that keeps getting glossed over. That is the fact that every FF sensor has an APS-C within it. If it truly does work out that shooting a 35mm lens at f/16 on APS-C gives you more DOF (and/or a shorter hyperfocal distance) than shooting a 50mm lens at f/22 on FF, then simply shoot the 35mm lens at f/16 on FF and crop in PP. You don't need a degree i math to see that this is going to come out *exactly* the same the same as shooting the 35mm lens on APS-C in the first place.

Needless to say, the converse is not true - there is no equivalently simple way using an APS-C camera to emulate the extra shallow DOF that FF provides at the wide open end. Yes, you can use heavy duty PP techniques to mimic the effect given sufficient inclination and time, but that's hardly the same as a simple crop. Heck, most FF cameras these days can do the crop for you in the camera.

I've also pointed out that with the same MP, crop sensors provide greater detail at the same focal length

While this may or may not be true, I don't see the relevance, and I guess that's why I haven't commented. Who compares different formats at the same focal length? What would the point of that be? And who compares sensors of the different sizes and then saddles the larger sensor with only the same number of pixels as the smaller one? There is no advantage in doing so, and you miss out on the very real potential advantage of the larger sensor: the ability to get higher resolution images out of the same basic sensor technology.

And of course, my observation above about cropping an image from FF to get *exactly* the same image as with an APS-C camera depends on the FF sensor being the same basic technology as well - meaning the same pixel size, just more of them.

[Marc Sabatella]

Absolutely, but to achieve that you have to drop back to a lower effective pixel count; you might as well use the larger sensor sites and a lower MP sensor. This is where many people get the "megapixels don't matter" mantra, but it's really "megapixels *alone* don't matter". I'm just saying that you can *get* more detail with the FF by adding more pixels, but you have to give up some of that detail to get the DR back, or give up the DR to get the detail. Trade-offs, not a "slam dunk".

??? If I have an FF sensor with pixels of a given size, then physical cut it smaller to create an APS-C sensor (or crop the image in firmware), I now have to sensors of different sizes using the same technology - the perfect point of comparison if we're trying to analyze the effect of sensor *size*, which is what we're trying to do here.

In such a comparison, the FF camera has more pixels, and thus the potential for more detail (assuming it's matched to a lens capable of resolving that much detail). In what way do I have to give up some of that detail to get DR?

The point of that part of the exercise was to illustrate, anyway, that megapixels sometimes *do* matter.

True. Interesting how popular notions change over time. At one time, it was considered a given that more MP = better; no one thought about sensor size at all. Then for a while people got the notion that more pixels was worse unless you increased sensor size, because more pixels means smaller pixels and people somehow got the notion that lots of smaller pixels meant more noise than fewer larger ones (probably from comparing images at "100%" rather than by comparing at similar physical print/display sizes). Then once people finally accepted that sensor size really was the determining factor for noise, the conventional wisdom became that once you got past 6MP - or maybe 10MP - additional pixels didn't matter *at all*. Now, I think the tide has turned back to people realizing that sometimes more pixels really are better.

[jstevewhite]

I think perhaps the issue here is that you mean something entirely different by "less DOF" than everyone is assuming. Your wording makes it sound like every single shot taken on APS-C will have more DOF than even the deepest DOF shot on FF. This is, I hope you'll agree, completely false.

Indeed, and never made that claim. Sorry if somehow it sounded like it. The point is that, for a given FOV, there are two DOF scales, one for FF, one for APS-c, and they're offset; the FF DOF doesn't extend beyond the APS-c range in both directions, as has been asserted.

But assuming your calculations are more correct than Atkins', really, you are talking about "DOF control", just as I (and others) are. For *most* shots, you can get the *same* DOF with either system. FF allows you to squeeze some extra shallow DOF at one end of the scale should you so wish, and APS-C (if your calculations are correct) allows you to squeeze some extra deep DOF at the other end of the scale should you so wish, but there is more overlap than difference.

So - again taking your calculations at face value - it would not be correct to say that either system gives you more DOF control overall. They both allow the *same* DOF control in the heart of the aperture range and focus distance range, but FF provides additional options at one extreme and APS-C may possibly provide additional options at the other extreme.

This is exactly point I was trying to make all along.

I would challenge the assertions that the limited are sharper at f/16 than other commonly used lenses would be in the first couple of stops, unless you mean, consumer zooms shot wide open, or literally the first two stops of some sub-f/2 primes. Of course, since it's the sub-f/2 primes that one might turn to to get the extra shallow DOF that gives FF its advantage at that end of the scale, I would agree that at face value, it seems one could similarly say that getting that extra shallow DOF on FF comes at a price optically, just as getting the extra deep DOF on APS-C does.

Many of the sub-f/2 primes start out at resolutions well below what they'll produce at f/16, and very close to what they'll produce at f/22. f22 from my FA35 makes an acceptable 8x10 for most folks (due in part to the contrast it starts with). For web display - where, let's face it, the vast majority of photographers show their work - even that is razor sharp after the image is re-sized to 1024x768 - or even twice that, which would be an 8x6 inch print @300dpi, but is full screen for a 100dpi device like a monitor.

Except that when you consider many of the common cases, this just isn't so. A typical FF portrait lens shot at f/2.4 is likely to be quite sharp already - certainly sharper than a typical APS-C portrait lens shot at the equivalent aperture of f/1.7. So even opening up your FF portrait lens to f/2 or more - going to DOF places you would not be able to get to on APS-C at all - your image is probably still as sharp as your APS-C image was at its maximum aperture. Obviously, this depends on the specific lenses involved, but the point is there is no basic physical principle at play that limits the IQ of a lens at large aperture.

You're once again depending on the limitations of available lenses, not anything inherent to the format. Because everyone persists in the ridiculous assertion that pros *must* use FF, most "pro" lenses are aimed at FF, not designed for APS-c. From everything I understand, a 35mm f1.0 for APS-c image circle wouldn't be any more difficult or expensive to build than a 50mm f1.2 for FF. Emacs said this wasn't the case, and I'm waiting for his supporting information, as everything I've ever read said as image circle decreases, it becomes easier to build sharp, fast lenses. My Rodenstock 5.6 250mm was considered a good lens ($2400 at the time), but it's resolving power was very low compared to a 250mm Hasselblad lens($2400 at the time), which was low compared to an equivalent Leitz ($2400 at the time) 35mm lens.

However, diffraction *is* a physical principle that limits IQ at the other end of the DOF scale. The effect is quite real, no matter how sharp your lens is, and this is not taken into account by the simplified/idealized hyperfocal distance formula you have been relying on.

The required COC changes as a function of size and viewing distance. It's not a set value. The general guideline is "an acceptably sharp print at 8x10 inches, viewed from about 15 inches away" or some near variation of that number, and makes an assumption about the discernment of the human eye being trained upon that print.

The fact is that APS-c can make an acceptably sharp print at 8x10 up to f22. Bob Atkins explains this on this page. f22 is not the wholesale image destruction many hint that it is. This comes from being pixel-peepers. I *am* a pixel-peeper, but I also understand the relationship between what I see at 100% and what I'll see in an 8x10 print.

Here's a way of putting into plain English. DOF can be described as the zone in which points in the actual scene will be resolved to circles no larger than the CoC for the sensor size, print size, viewing distance, and visual acuity of the viewer. You've been using a typical value of 0.03mm for CoC on FF, which corresponds to roughly 0.02mm on APS-C. So you're saying that a point on an image is sufficiently sharp to you if it can resolve to a circle no larger than 0.02mm. Well, if my understanding is correct, diffraction at f/22 on APS-C is already such that *no* point - not even the points right at the plane of focus - can possibly resolve to a circle that small. And long before you reach that aperture, diffraction is going to take its toll on sharpness, such that the sharpest points in an f/16 image are already less sharp than the sharpest points in an f/11 image. Which in turn may mean that points toward the front or rear of the zone of acceptable sharpness that were allegedly more in focus at f/16 than at f/11 because of ordinary DOF calculations will actually be *less* sharp due to diffraction. No, I haven't worked through all the math. But it's pretty clear that your hyperfocal formula is blowing this off entirely.

It's not blowing it off; conversely, you're making much more of it than is rational. f22 is not as bad as you make it out. The k-5 with a well corrected lens will make an acceptably sharp 8x10 from an exposure at f22. Period. Since most lenses don't stop down any more...

However all of this actually pans out, though, there is another aspect to it that keeps getting glossed over. That is the fact that every FF sensor has an APS-C within it. If it truly does work out that shooting a 35mm lens at f/16 on APS-C gives you more DOF (and/or a shorter hyperfocal distance) than shooting a 50mm lens at f/22 on FF, then simply shoot the 35mm lens at f/16 on FF and crop in PP. You don't need a degree i math to see that this is going to come out *exactly* the same the same as shooting the 35mm lens on APS-C in the first place.

No, it wasn't glossed over. On at least two occasions I said "But FF can always change lenses and crop." (meaning to get the same FOV and DOF). But on the other one, the APS-c can always open up a stop and a third, right? Oh, wait, there's no 35mm f1.0 for K-5s, is there? But guess what? There's no FULL FRAME PENTAX to put the 50mm f1.4 on! (if lens limitations are fair game, then so are camera bodies, no?)

Needless to say, the converse is not true - there is no equivalently simple way using an APS-C camera to emulate the extra shallow DOF that FF provides at the wide open end. Yes, you can use heavy duty PP techniques to mimic the effect given sufficient inclination and time, but that's hardly the same as a simple crop. Heck, most FF cameras these days can do the crop for you in the camera.

Not true! Stitching is simple and can produce resolutions (and DOF) that you can't make on a FF camera without stitching yourself. Notice I didn't say *trivial*; but with the right software and half an hour's practice, it's simple.

If the lenses existed, I wouldn't *have* to emulate it; I could just produce it by opening up my 35mm prime to f1.0. Of course, there aren't any; but there aren't any FF Pentaxes to put the 50mm 1.4 on anyway, so the point is moot. *shrug*. If we're hung up on equipment availability, I can say without reservation that Pentax APS-c is better than Pentax FF, since the format that exists MUST be better than the format that doesn't. Anselm's APS-c proof, anyone? LOL!

While this may or may not be true, I don't see the relevance, and I guess that's why I haven't commented. Who compares different formats at the same focal length? What would the point of that be? And who compares sensors of the different sizes and then saddles the larger sensor with only the same number of pixels as the smaller one?

Quick! What's the resolution of the redoubtable D3s? Are you suggesting that no pros should be using them? Or that the K-5 is better just because it has more PIXELs?

Photography happens in the real world. The assertion is often made that FF is superior to APS-c, but this is literally only true if all else is equal - and it's not. Not at all. The K-5 outperforms and out-resolves most of the last generation of FF, and the ones it doesn't out-resolve it out-DRs. Not saying it's superior, I'm just saying that by any reasonable measure the K-5 is *in the hunt* as far as IQ goes, FF or no.

There is no advantage in doing so, and you miss out on the very real potential advantage of the larger sensor: the ability to get higher resolution images out of the same basic sensor technology.

And of course, my observation above about cropping an image from FF to get *exactly* the same image as with an APS-C camera depends on the FF sensor being the same basic technology as well - meaning the same pixel size, just more of them.

This is addressed to the noise limitation of FF. Larger sensor sites means lower noise, and this is one of the talking points of FF proponents. This is the origin of the myth that megapixels don't matter, as there is a min/max beyond which adding pixels decreases detail by increasing noise. Fortunately, neither format we're discussing is approaching that limit.

However, all else being equal, lower noise==greater available dynamic range (provided your sensor records enough bits to capture it). So the amazing D3s still has higher dynamic range and better high-iso performance than the K-5; but the 5DmkII does *not*. You can, as has been pointed out, resample and mix pixels to recover dynamic range, but then your effective pixel count drops. To add one stop by resampling, you have to cut your resolution by 30% or so. And AFAIK, you can't recover more than the base DR of your sensor that way, anyway.

And as to "Who compares different formats at the same focal length?" I have the answer. People who shoot high magnification. Dangerous or skittish wildlife (higher pixel density wins) and Macro (same thing). 1:1 doesn't care what format you're using, and pixel density is king (provided you can illuminate the subject above the noise floor).

I've never argued that APS-c was superior to FF; I'm arguing against the misguided idea that it's reasonable to assert that "Full frame is objectively better!" without first answering the question "For what?". Also, in the wider scheme of photography (p&s to 8x10 view cameras) the differences between FF and APS-c are very, very small, and limited to edge cases like razor thin DOF or high magnification macro.

[jsherman999]

But assuming your calculations are more correct than Atkins', really, you are talking about "DOF control", just as I (and others) are. For *most* shots, you can get the *same* DOF with either system. FF allows you to squeeze some extra shallow DOF at one end of the scale should you so wish, and APS-C (if your calculations are correct) allows you to squeeze some extra deep DOF at the other end of the scale should you so wish, but there is more overlap than difference.

So - again taking your calculations at face value - it would not be correct to say that either system gives you more DOF control overall. They both allow the *same* DOF control in the heart of the aperture range and focus distance range, but FF provides additional options at one extreme and APS-C may possibly provide additional options at the other extreme.

This is exactly point I was trying to make all along.

Isn't that a pretty expansive definition of 'extreme', though, if it starts at 85mm and wider on FF?

Breaking it down, using Nikon:

85mm f/1.4 (or 1.8): No 55mm f/1 (but there is a MF 55 f/1.2! My beloved Cosina!)
50mm f/1.4D (or even my 1.8G): no 35 f/1.0 (or 35 f/1.2) on aps-c
35mm f/1.4G (or even the 1.8G, which I shoot on FF): no 24 f/.09 (or 24 f/1.2) on aps-c
24mm f/1.4G (or even 2.8D): no 16mm f/.09 (or 16mm f/1.8) on aps-c
20mm f/2.8D: no 13mm f/1.8 on aps-c
14mm f/2.8D: no 9mm f/1.8 on aps-c (but this is getting extreme)

(& if you're willing to shoot MF, there's not a 35mm to match the 50mm f/1.2)

If you want to just approximate the DOF capability at the same FOV, you can with some combos, but the FF to aps-c matchup greatly favors FF in price (and lens size!) in those cases:

tiny $110 50mm f/1.8D on FF vs. huge $1750 35mm f/1.4G on aps-c
tiny $199 35mm f/1.8G on FF vs. huge $2000 24mm f/1.4G on aps-c

One example approximation at the longer end: (this single lens/body combo was instrumental in my D700 purchase)

$500 180 f/2.8 AF-N vs $1400 135 f/2 or $1000 105 f/2 on aps-c

Once you get up to 200mm f/2.8 and higher, it starts to favor aps-c in price and size - but there is no practical DOF/FOV combo you can't get to up there with FF, it just gets extremely prohibitive and stops making sense for anything but very low-light telephoto.

You're once again depending on the limitations of available lenses, not anything inherent to the format.

These discussions often eventually lead to this point.

"Well, OK, OK, granted, everything granted. But it's not inherent to the format! It's a lens limitation on aps-c!"

Absolutely correct.

Same argument can be made for the P&S sensor size - if they were able to make the right lenses for that sensor, you'd see fantastic DOF control there also.

Because everyone persists in the ridiculous assertion that pros *must* use FF, most "pro" lenses are aimed at FF, not designed for APS-c. From everything I understand, a 35mm f1.0 for APS-c image circle wouldn't be any more difficult or expensive to build than a 50mm f1.2 for FF. Emacs said this wasn't the case, and I'm waiting for his supporting information, as everything I've ever read said as image circle decreases, it becomes easier to build sharp, fast lenses.

Conspiracy theory time? Maybe the Nikon/Canon Illuminati forced Sigma to keep that 30mm at f/1.4 instead of the f/1.0 we know they could have achieved!


.

[jstevewhite]

Isn't that a pretty expansive definition of 'extreme', though, if it starts at 85mm and wider on FF?

Wrong definition of extreme. He didn't mean extreme as in "of a character or kind farthest removed from the ordinary or average: extreme measures." He meant extreme as in "farthest from the center or middle; outermost; endmost: the extreme limits of a town"; my 50mm is f1.4 at one extreme, and f22 at the other extreme.

These discussions often eventually lead to this point.

"Well, OK, OK, granted, everything granted. But it's not inherent to the format! It's a lens limitation on aps-c!"

I *didn't* say "everything granted"; I was pointing out that his specific example was depending on the lack of lens availability, not couching an entire argument on it.

Absolutely correct.

Same argument can be made for the P&S sensor size - if they were able to make the right lenses for that sensor, you'd see fantastic DOF control there also.

Ok.. what's the point that you're trying to make? "Fantastic DOF control" != "Fantastic Image Quality" (which has been ANOTHER point I've been trying to make).

Conspiracy theory time? Maybe the Nikon/Canon Illuminati forced Sigma to keep that 30mm at f/1.4 instead of the f/1.0 we know they could have achieved!

.

Uh... Price? How much is a Canon 50mm f1.2 AF? $1600? How much is the Sigma 30mm f1.4? $450? I said I think they could build a 30mm f1.0 for APS-c for the price of a 50mm 1.2... I still think that. But the money is in FF.

[jsherman999]

Wrong definition of extreme. He didn't mean extreme as in "of a character or kind farthest removed from the ordinary or average: extreme measures." He meant extreme as in "farthest from the center or middle; outermost; endmost: the extreme limits of a town"; my 50mm is f1.4 at one extreme, and f22 at the other extreme.

I'll let Marc address what he meant, but there is a misconception out there that FF only gives advantages like that at the extremely wide end.... I see that you were addressing the aperture range argument there, on second reading.

Ok.. what's the point that you're trying to make? "Fantastic DOF control" != "Fantastic Image Quality" (which has been ANOTHER point I've been trying to make).

Basically that lack of lens availability for a format makes it a defacto format limitation, especially when it becomes technically impossible to design certain lenses for a given format that would have a reasonable combination of size/price/IQ.

Uh... Price? How much is a Canon 50mm f1.2 AF? $1600? How much is the Sigma 30mm f1.4? $450? I said I think they could build a 30mm f1.0 for APS-c for the price of a 50mm 1.2... I still think that. But the money is in FF.

Sigma, for example, sells a lot of telephoto lenses to aps-c shooters for more than $1600. Nikon has had great success selling a pedestrian 17-55 2.8 to aps-c shooters for $1500 a pop.

You don't think that Sigma could sell an unheard-of-before, stunning 30mm f/1.0 for $1600 if they wanted to? It would probably be sold out everywhere for months after introduction.

If Sigma (or anyone) could make a 30mm f/1.0 for aps-c that was smaller than a canned ham, had corners that wasn't like looking through a clown-tunnel, and cost $1600 - it would have happened by now.

I think you challenged someone to prove that it couldn't be done - isn't the onus on you to prove that it could? Sigma and everyone else is waiting to hear about it, I think.


.

[twitch]

jsherman999 is intent on completely wrecking by bank account with good logic here.

Conventional wisdom is that we should be spending money on lenses rather than bodies, but actually it looks like spending on a FF body means significant savings in lenses <200mm. So for budgets over $3k it does seem as though FF is cheaper < 200mm FL.