[Ian Stuart Forsyth]

To simplify, what you're saying is that observing the image at higher magnifications would make existing flaws more visible, which is of course correct. But what are you getting at? That the larger format is necessarily "better"?

Yes as the larger format has to be enlarged less for your final viewing

I keep going back to this comparison just to show you what happens at these pixel densities 16mp for FF and 16mp for cropped
If we look at this you can see even at 16mp we see a hit to the captured resolution with the DFA 100mm WR and the 16mp K5, there is enough of a hit that I could use my old M 80-200 on a FF 16mp camera that would have pixels as large as the ones found in the DS

The DPR image comparison widget I've linked to in a previous posts does exactly that: it compares 24MP APS-C and FF cameras, and the APS-C doesn't lose.

24mp cropped is the goldilocks as you can use a cropped sensor without a AA filter while on FF you still need a AA filter, this in its self should tell you lenses on cropped act as a AA filter

L.E. I guess we both agree that something must be "better" on APS-C, in order to keep up with a same MP count FF.

And that's my main point, but as more FF lenses are updated that gap widens again, or I can simply use a lot of the modern zooms out there and still come out equal and sometimes well ahead of cropped.
And I mean look at the DFA 100mm WR it is no slouch when it comes to resolution, even when we only look at 16mp FF to 16mp cropped we can see a difference even with the outdated 80-200 push pull, what will 36mp cropped and 36mp FF look like

There's a problem though: it's not enough for the FF to get a similar performance as APS-C, it must be better. This translates into lens design, too.

Better how ? Some of the older lenses I have for my D800 where not designed as resolution monsters but because they are sharper when used on FF I still get the resolution I want but I also get other IQ because the lens was designed for other rendering qualities. While other times instead of using several primes on my cropped camera I can now use a single zoom on that FF while still giving me the same or better resolution.

---------- Post added 01-04-2018 at 01:40 AM ----------

It surprises me you don't realize using a single lens on the same camera at different distances produces different subject resolution. That was the part of the test that makes it irrelevant. So, no it doesn't show anything, and since you don't get it, there's not sense in me discussing it with you further.

No they produce different magnifications
If I move the camera back or forth 7m its not going to change this resolution score

http://www.photozone.de/pentax/125-pentax-smc-fa-43mm-f19-limited-review--test-report?start=1
2422 lw/ph @ f/4

Comparative resolution tests, must be taken at the same distance.

I did here

same distance

You do realize that the differences in your APS-c and FF images was 200 lw/ph or about 5%. Such small differences in lw/ph will rarely be visible. It will amount to a little bit of a softer edge on a 60 inch wide print,, so no real functional difference. Most of us consider less than a 20% increase not worth paying for, or paying attention to.

Clearly you missed why I was showing you the test results of the 2 24mp camera

This should be blatantly obvious to you the reason why there is very little difference is because the cropped camera is using the decreased resolution that system ( camera and lens) can capture as a AA filter while the increased resolutions that the FF system ( camera and lens) can capture needs a AA filter.

---

Last edited by Ian Stuart Forsyth; 01-04-2018 at 01:17 AM.

[Kunzite]

Ian, I think you can now stop re-posting the same images.
They're not exactly relevant anyway, as the world moved on from 6MP. An old M-series zoom still doing OK on the center, with a low pixel density sensor...

So... the 24MP cropped cameras will continue to lack an AA filter, and the FF ones will continue to need it. Thus, APS-C 24MP will continue to keep up with 24MP FF, IMO.

"Better how"? We are talking about detail captured in relatively good light. If FF doesn't show a clear difference, then why paying the added cost?

[Alex645]

So... the 24MP cropped cameras will continue to lack an AA filter, and the FF ones will continue to need it. Thus, APS-C 24MP will continue to keep up with 24MP FF, IMO.
If FF doesn't show a clear difference, then why paying the added cost?

AA (anti-aliasing) or OLPF (optical low pass filter) were commonly used on both APS-C and FF cameras to reduce jaggies/stair stepping artifacts in diagonal or curved lines due to pixels arranged in horizontal rows and vertical columns. The latest gen sensors for both APS-C and FF are now eliminating this filter including the K-1.

The larger the sensor, the less magnification is required on both the light hitting the sensor and the final image and will also typically improve low light capability, resolution, etc. At a certain viewing distance or image size, there is a clear difference between FF and APS-C. A bulk of the added cost is due to the larger sensor and bigger buffer, larger shutter curtain, etc, needed to support the larger format.

This is not addressed specifically to any poster, but just to help clarify for others reading this thread from confusion.

[Rondec]

Full frame is "better" in certain specific ways. Those ways usually have less to do with resolution (although some full frame sensors have more resolution) and more to do with dynamic range and noise at a given iso. I do think that as you go up in pixel density, getting truly sharp images at the pixel level becomes more difficult. I can't imagine trying to get a 50 megapixel APS-C image to be pixel sharp. That doesn't mean that if you were able to do it that it wouldn't beat a 36 megapixel full frame image with regard to resolution -- it probably would -- it is just it would require good stabilization or higher shutter speeds than we are used to using.

But I don't have a dog in this fight. I shoot both with a K3 and K-1 and find them to be enjoyable to use. I prefer the K-1 for landscape shots, mostly because of the dynamic range I mentioned above. Otherwise for family photos, etc I view them as pretty equal.

[Kunzite]

@Alex645: to my knowledge, camera makers are still using AA filters on 24MP FF sensors.
I'm well aware that my K-1 has no AA filter - but I was specifically talking about 24MP sensors, not 36.

@Rondec: I do not believe in "pixel-level sharpness" as an end goal; I'd rather have softer pixels, if it results in a better image.
But increasing the APS-C's resolution in order to keep up with FF is not my point. On the contrary, what I'm saying is that FF can and should (and did, in some products) increase its resolution, to the point where it can safely dispense of AA filters and clearly differentiate from APS-C (even in good light).

[normhead]

This should be blatantly obvious to you the reason why there is very little difference is because the cropped camera is using the decreased resolution that system ( camera and lens) can capture as a AA filter while the increased resolutions that the FF system ( camera and lens) can capture needs a AA filter.

That's not blatantly obvious to anyone but you.

I did here

You compared a 100 macro to an M 80-200 F4.5 (ambitiously rated 8.5) against a DFA 100 macro rated 9.63? Why didn't you use the 80-200 for both? It didn't prove your "point" that way? I really don't need a lens test to know which of those two images is going to be sharpest. Once again, you handicap the larger sensor, this time by using an inferior lens.

You are making me want to start a thread on testing methodology.

Yes as the larger format has to be enlarged less for your final viewing

As I have pointed out many times, both images are usually reduced in digital, not enlarged, unless doing really big prints. Also the above statement is only true if you are talking about a large format using the full frame. For birds and distant subjects where cropping is necessary, or for any image where the photographer doesn't have a long enough lens to fill the frame as desired APS-c MP actually gives you a "bigger" as in more MP and More lw/ph on the subject than 36 MP FF, forget about 24 MP FF. So this statement is problematic. And anyone in the habit of taking both an APS_c 24 MP K-3 and a 36 MP K-1 out shooting wildlife know that. It would actually be the larger format image that would have to be enlarged more, because of fewer pixels on the subject. The old film concept of how much is it enlarged is largely irrelevant in digital. The only thins relevant in digital is "how many lw/ph does you sensor/lens combination give you on your subject" and that is roughly equivalent to the number of MP are crammed into the area the subject occupies. The old film era thing where you actually had to worry about grain being enlarged enough to be easily visible is simply irrelevant with digital. The images are enlarged by complete different and not really comparable processes.

As I pointed out above in the article I linked to, lens sharpness comparisons are only valid only if the weakest lens cannot resolve the image detail captured in the frame.

I could make the argument that my Signa 70-300 is just as good as my DFA 100 macro simply by finding a subject close enough to the camera with course enough detail that both lenses resolve all necessary detail. That doesn't however prove the Sigma 70-300 will win that comparison in all cases. It only proves that I devised a test to show a circumstance where they are roughly equal. But the fact that I can do that says nothing about the overall quality of either lens. It just would show I can cherry pick specific circumstances to support an otherwise erroneous assumption.

I recently did 3 12x24 inch prints taken with a K-1, with no enlargement, or reduction in size at 300 DPI. The same image done with a APS-c image would have been 20 inches. The enlargement ratios are completely different in digital, because you aren't magnifying grain. My larger format image is only 20% "larger" than my APS-c format image, not 100% more enlarged as it would be with film and the so called "enlargement ratios" need to be adjusted accordingly. And a 20% resolution increase is just on the cusp of what might be noticeable in every print. That won't be a lock until you achieve a 50% increase.

The crazy thing is a Panasonic 1 inch FZ1000 image taken in good light, even though it's only 20 MP out resolves my K-3 even though it's sensor is less than half the size and it only has 20 MP. Film era enlargement analogies are completely obsolete. Especially since it's easier to manufacture a really high quality lens for a small a sensor than it is for large one. If a large pixel gives a you a values of 167 in the RGB channls and small sensor give you the same values in a picture in the same position of the sensor array, there is absolutely no large pixel advantage. Magnification of the size of the sensor is irrelevant. They are just numbers. There is no way you can even tell which is which.

But this is old news for any of us who actually took formal training in photgraphy. My old practice used to be, if you want a nice clean looking high contrast black and white image, use 32 ASA (super fine grain) copy film. A 35mm image taken with copy film could be enlarged to a larger size than a 645 image taken with 400 ASA film, and still look better. Even in the old days there was more to it than just format size. And I personally shot many rolls of 32 ASA copy film both in studio and out walking around, just for that reason.

In both film and digital there is a scale, under exposed, correctly exposed, over exposed. If you can hit "correctly exposed" with fine grain film or a digital small pixel sensor, it's the same as a correctly exposed large sensor of the same MP in digital. It doesn't matter in digital how big the capturing pixel was. It's not enlarged the same way it was with film. What people don't seem to understand is, if the small sensor image is correctly exposed at the edge of its functionality, is the same as a larger sensor image also correctly exposed. The fact that you could get a much more over-exposed image with the large sensor camera is irrelevant. Over exposure is bad, and arguing one could over expose and the other can't is irrelevant.

In poor light, larger formats are better, in good light, pixel size, grain size, lens quality and probably a few other factors I didn't think of can flip the tables. It's all about recognizing the circumstances you need to exploit the system you have. Defintinately not about blanket statements like "larger formats are better."

You are almost writing a text book case here in how to shoot tests to prove an erroneous point, rather than to discover the truth. You should work in marketing.

[Rondec]

@Alex645: to my knowledge, camera makers are still using AA filters on 24MP FF sensors.
I'm well aware that my K-1 has no AA filter - but I was specifically talking about 24MP sensors, not 36.

@Rondec: I do not believe in "pixel-level sharpness" as an end goal; I'd rather have softer pixels, if it results in a better image.
But increasing the APS-C's resolution in order to keep up with FF is not my point. On the contrary, what I'm saying is that FF can and should (and did, in some products) increase its resolution, to the point where it can safely dispense of AA filters and clearly differentiate from APS-C (even in good light).

I guess I just see people comparing APS-C and full frame and am not sure what the big deal is about. It makes sense that full frame is about a stop better with regard to performance, although some of the newer APS-C cameras are probably as good with respect to DR and SNR compared to Canon's 6D II, at least at low iso settings.

The problem with full frame (and Ian might argue with me on this) is that long lenses can get pretty expensive. APS-C doesn't really make your lenses longer, but if you shoot wildlife and can't afford a 600mm lens and are mostly cropping to APS-C anyway, what's the point of paying for the extra sensor area that you aren't using?

[Kunzite]

@Rondec:
Well, I never said that APS-C is unnecessary
Just like FF's advantage over APS-C can be the image quality, APS-C can have its own advantages and thus finding a place in our bags.

[Rondec]

What are we arguing about? I think we believe the same things.

APS-C tends to be cheaper, have faster frame rates, smaller size, and still offer really good quality these days (who needs more than a 24 megapixel image anyway?). It probably does offer some extra reach over full frame cameras, particularly if you are comparing it to entry level full frame.

Full frame offers ability to shoot with shallower depth of field and gives better quality images in poor light and difficult conditions. As you say, there is room for both in your bag (clearly I still shoot with both). People tend to be a little bit too attached to one point of view or the other -- either full frame is the best or, only professionals need something more than what is offered by APS-C.

Better to shoot with what you can afford and learn to use it well.

[Kunzite]

I guess we're arguing about which one of us is agreeing more with the other?

(I was trying to state my agreement for a while, sorry, perhaps I should make it more obvious...)

[normhead]

I guess we're arguing about which one of us is agreeing more with the other?

(I was trying to state my agreement for a while, sorry, perhaps I should make it more obvious...)

I volunteer to be an independent 3rd party judge to add up who's agreeing more than the other. Sometimes these things are important.

[Alex645]

@Alex645: to my knowledge, camera makers are still using AA filters on 24MP FF sensors.
I'm well aware that my K-1 has no AA filter - but I was specifically talking about 24MP sensors, not 36.

....and to my knowledge, only the Nikon D610 (2013) and Nikon D750 (2014) are the only currently available 24MP FF DSLRs on the market, and yes, they both have AA filters. So I would agree that up until 2014 camera makers were still using AA filters on 24MP FF sensors.

What other 24MP FF DSLRs today are using an AA or OLPF filter?

[Kunzite]

Sony makes non-DSLR ILCs with 24MP FF sensors, AFAIK with AA filters; and Canon is using AA filters even with it's 50MP 5DS (not with the R variant, of course).
This is a question of "how many MP for a given format before you can safely remove the AA filter", and nobody said "24 for FF"

[Nicolas06]

....and to my knowledge, only the N
ikon D610 (2013) and Nikon D750 (2014) are the only currently available 24MP FF DSLRs on the market, and yes, they both have AA filters. So I would agree that up until 2014 camera makers were still using AA filters on 24MP FF sensors.

What other 24MP FF DSLRs today are using an AA or OLPF filter?

Nikon FF Camera with low pass filter D610, D750, D5, Df for Nikon for the recent one. Most of the older got one two D600, D700, D800 (not D800 E).
Canon: All Canon FF DSLR except the 5ds R
Sony: A7/A7 II, A9, A7s, A7sII...
Leica: No idea but I'd bet they have quite a few with a low pass filter...

Interrestingly the top of line of all FF manufactuers except Pentax has a filter like the D5, Sony A9, Canon 1DX mark II... And Pentax has a way to simulate it if you need it.

[photoptimist]

Sony makes non-DSLR ILCs with 24MP FF sensors, AFAIK with AA filters; and Canon is using AA filters even with it's 50MP 5DS (not with the R variant, of course).
This is a question of "how many MP for a given format before you can safely remove the AA filter", and nobody said "24 for FF"

It depends on how good (or bad) the lenses are.

A Bayer filter sensor has pretty nasty sampling performance in both the red and blue channels -- throwing away 75% of the image information in those color bands. In effect, a 50 MPix Bayer sensor is really only a 12.5 MPix sensor in red and blue with huge gaps between the light-sensitive elements. If the lens has any ability at all to resolve any contrast at 60 LPM, there's a risk of moire without some sort of OLPF on even a 50 MPix FF (or 22 MPix APS-C). (Some of the best lenses can resolve to 160 LPM)


That said, for most scenes (which don't have moire-suseptible details), most lenses (which are not the best of the best), and most shooting situations (some combination of aperture or shutter speed leaves a pixel or two of blur), then an OLPF does more harm than good.

(It's another reason that more recent Pentax cameras are superior to Canikon. Pentax uses the SR system to induce OLPF when the photographer needs it.)