[ElJamoquio]

If you want narrower depth of field than you get with a 35mm f1.4 or a 20mm f1.8 on APS-C, it is fine to go with full frame.

What if you're fine with the DOF of an APS-C F/2.8 zoom, but don't want to spend the extra money for APS-C?

Isn't this Dr. Cameraman amateur blog post just the same as billions of others who want to claim this or that by picking just the data they want to see?
There is so many aspects to the theory alone and then even more to practical aspects that you can get any result you plan for.

Let's assume for now Dxomark would be knowing what they do and their perceptual MPix score was meaningful.
Then you pick two equivalent combos (recent generation sensors with same 24 Mpx plus equivalent lenses):

• D600 plus Nikkor 300/2.8 - price point at amazon $1,535 + $5,800 =$7,325 scores 20 MPx
• D5300 plus Nikkor 200/2 - price point at amazon $796 + $5,800 =$6,596 scores 18 MPx

And I hope all agree that none of the two lenses is a decidely poor performer, same as the D600 should not be a one off poor performer against the D5300.
20 vs 18 MPx on a 3:2 sensor means 3651 versus 3464 pixels picture height as resolution is most often expressed in something rated against picture height. That is 5,3% difference.

So Dr. APSC fanboy now could write a post about the investment of full $800 budget stretch resulting in a completely meaningless +5,3% resolution gain. Or:
"Practical FF resolution gain is a sad invisible +5% for the cost of loosing all of a 35/1.8 DX plus 18-105 DX plus 50/1.8 plus 70-300 as your lens arsenal just for this"
Cost advantage? Nil. Even worse.
Resolution advantage? None. Not even Mr. Pixelpeeper will find the 5% with his nose pressed against the screen of any real life picture.
And all this only by picking some different data point. No magic involved.

I would stay away from this level of cheap propaganda. The bloggers usually only do it for the click-count.

Better just buy a camera that does what you want and lenses you like within your budget and be happy with it.
Today on the market there are pretty much no really "poor" cameras or lenses any longer unless you really want to print very large scale stuff.

[Rondec]

What if you're fine with the DOF of an APS-C F/2.8 zoom, but don't want to spend the extra money for APS-C?

I don't have an answer for you. I shoot Pentax because I like Pentax glass and quite a bit of my glass is actually full frame compatible (all my primes except for the DA 15 limited). The DA * zooms are OK (the 50-135 is really good), but if I went full frame, I would want f2.8 zooms. My experience is that camera companies just put more effort into their f2.8 zooms than their slower zooms.

Anyway, if Pentax offers full frame in the future, I will probably buy, but not because of a 30 percent increase in resolution. And not for less depth of field. And not because I will save money (I have all the lenses I need currently and if I buy full frame I will need to buy a 2500 dollar body and two 1600 dollar zooms). I guess I would buy it because I secretly want it and that's probably enough.

[ElJamoquio]

Cost advantage? Nil. Even worse.
Resolution advantage? None. Not even Mr. Pixelpeeper will find the 5% with his nose pressed against the screen of any real life picture.
And all this only by picking some different data point. No magic involved.

You're right. If you
1) limit yourself to very long lenses
2) ignore higher megapixel cameras available for larger sensors (or don't want to pay the extra money for a camera to pair with a $5k lens)
3) ignore that teleconverters can improve the situation more on larger sensor cameras (less error)
4) ignore the AF improvements that most practical FF cameras bring

then sometimes APS-C cameras are as good as FF cameras!

An example in my mind would be a 24MP APS-C 600mm F/4. Although personally I'd rather have the D800E.
Another example also is the dirt cheap combos - the $100 APS-C lens on the D5300. It's less than half as expensive as the D610+kit lens.

I think most people who read about photography on the internet to influence their opinion are between those two extremes though.

[nomadkng]

Class A - I'm not being obtuse or argumentative. I've been here long enough to have read Falc and understand.

Yet on a resolution test, have we used the exact same sensor technology except for a larger slice of the wafer, the same lens on a camera using the same RAW capture algo and precisely the same processing in some processor - or the exact same in-camera jpeg engine - at the exact same true ISO - i.e. is everything the same except an APSc sensor in a camera instead of a FF sensor in an identical camera?

or have we NOT controlled for all variables other than sensor, and the rest is mathematically inferred (by Falconeye)?

I merely contend we cannot know precisely how much different FF is from APSc even empirically, and we cannot know from tests whther such difference matters in real-world use,other than for several well-explained conditions. Mine is a rather sarcastic rant against all this sniveling about sensor sizes.

Just out of curiosity, since several FX Nikons have a DX mode, couldn't one compare FX mode vs DX mode images made with the same lens? It would be the same sensor, etc with the only variable being the field of view and number of photosites utilized?

You could even do the opposite, and use a DX lens in both modes and only compare the non vignetted image portions. Why hasn't anyone thought of this? Or would such logical testing invalidate the "FF is the holy grail of photography" theory?

[Parallax]

Just out of curiosity, since several FX Nikons have a DX mode, couldn't one compare FX mode vs DX mode images made with the same lens? It would be the same sensor, etc with the only variable being the field of view and number of photosites utilized?

You could even do the opposite, and use a DX lens in both modes and only compare the non vignetted image portions. Why hasn't anyone thought of this?

Perhaps because it's too obvious and too logical?

[monochrome]

Perhaps because it's too obvious and too logical?

Things that make me go, "Hmmmmm."

(There's probably some scientific reason someone can give why that isn't a scientific test, so the output isn't scientifically reliable).

You're right. If you
1) limit yourself to very long lenses
2) ignore higher megapixel cameras available for larger sensors (or don't want to pay the extra money for a camera to pair with a $5k lens)
3) ignore that teleconverters can improve the situation more on larger sensor cameras (less error)
4) ignore the AF improvements that most practical FF cameras bring

Actually I wanted to point out that all it needs is:

• Pick that tiny data point out of the whole ocean of data that I personally like (the combos chosen)
• Choose a way of measuring or expressing things which shrink the disliked figure (single dimension resolution sounds less)
• Express the money difference in a dramatic way by turning it into an impressive number of cheapo lenses
• Write a heavily biased headline around all of this using negative terms such as "sad invisible", "just", "loosing"

I do fully acknowledge that the same lens will typically resolve more on a larger DSLR sensor when having the same pixelcount.
I think it is helpful to understand that there is to be expected some better resolution from a larger sensor - I do like to understand these things so I know how they work and what I have to tweak, but:
I just don't think it's really that much to fuzz about. I would consider it to be about +20% in reality with the best lenses. And that is not really that much when I look at actual pictures with non-test-chart stuff on it.
When having reached that equipment level where this actually comes into play I would rather look at the softness quality of bokeh, the color rendition, the feeling of the lens in my hands when using it and similar things than count pixels when making purchasing decisions.
But that is just my personal view as amateur.

[Na Horuk]

lol or take a photo with a film camera, then take scissors and cut the edges off. According to the theory, the bit you cut off will immediately deteriorate and look worse than it did before you cut the corners.. and f1.2 is not f1.2, focal length is not a lens property, DoF depends on sensor size (not aperture), and A does not equal A... not only are all theories of photography wrong, but so is the law of identity. At least, when it comes to FF/APSC science

[ElJamoquio]

Actually I wanted to point out that all it needs is:

• Pick that tiny data point out of the whole ocean of data that I personally like (the combos chosen)
• Choose a way of measuring or expressing things which shrink the disliked figure (single dimension resolution sounds less)
• Express the money difference in a dramatic way by turning it into an impressive number of cheapo lenses
• Write a heavily biased headline around all of this using negative terms such as "sad invisible", "just", "loosing"

You think a normal prime is a 'tiny data point'? I'm sorry, but I disagree completely, that's a very important lens. Especially the $100-$800 ones, but all the lenses in FF were better than all the lenses in APS-C.

When having reached that equipment level where this actually comes into play I would rather look at the softness quality of bokeh, the color rendition, the feeling of the lens in my hands when using it and similar things than count pixels when making purchasing decisions.

No argument there.

[jsherman999]

Just out of curiosity, since several FX Nikons have a DX mode, couldn't one compare FX mode vs DX mode images made with the same lens?

Yes, they could, and have. What exactly would you like to know?

It would be the same sensor, etc with the only variable being the field of view and number of photosites utilized?

Yes, but also DOF would change once the resulting crop was resized - change from what it was as a % of PH in the uncropped FF image to be equal to what you would have got if you had shot with an aps-c lens in the first place.

Basically, DX mode on the D800 gives you almost exactly the same image you would have if you shot the scene with a D7000 or a K5 natively. About a stop more noise, a bit less DR noticeable if you're shooting up from base ISO, about 1,5x 'tighter' FOV (or 'magnification' when you display at the same size.)

You could even do the opposite, and use a DX lens in both modes and only compare the non vignetted image portions. Why hasn't anyone thought of this? Or would such logical testing invalidate the "FF is the holy grail of photography" theory?

Yes, you could, but what would you want to test by doing that exactly?

.

[Kunzite]

The practical results (the OP) agrees with the theory. So in practice, there is no difference in this case.

Yet when I'm looking at DPReview's studio samples, K-3 vs. D610 (the areas where the K-3 is focused properly, of course - the image suffers from poor alignment) I don't see any 30-40% difference. Double checking with Imaging Resource, carefully looking at the fabric patches - same story. It's hard to see a difference at all.
Maybe when you stop looking and start measuring instead, you can find that apparently large difference.

Of course, doing high ISO and the story changes; but we weren't talking about high ISO here.
And just in case: I believe Ricoh Imaging should make a Pentax FF DSLR, and I would most likely buy it.

[carrrlangas]

I just finished reading the Lumolab paper on "Camera equivalence" LumoLabs -- Camera Equivalence -- Whitepaper
I suggest anyone to read it if with doubts about how to proceed on a format comparison. I conclude that the Dr.Camera analysis is not valid but he´s onto something with using available lenses for each system... Which is the most cost effective solution?

[Parallax]

If there is more than one variable, the results are meaningless. You can't put a bikini on a pig, coveralls on a super model then declare as a result of the comparison that coveralls are sexier garments than bikinis.

[cali92rs]

If there is more than one variable, the results are meaningless. You can't put a bikini on a pig, coveralls on a super model then declare as a result of the comparison that coveralls are sexier garments than bikinis.

There are more than one variable in every single system in the universe. The trick is to make reasonable assumptions which yield meaningful results. And living in a complex world, there are often more than one variable. Sorry, just because something is complex does not make it invalid.

And if someone does not understand the analysis, that doesnt make it invalid either.
I don't understand the standard model for particle physiscs. That doesnt make it invalid.