Are all the complaints about APS-C low light performance exaggerated?

[MissionMan]

These aren’t typical shots I would shoot at high ISO and this wasn’t a planned test. I pulled out my camera for a couple of product shots of the X100F and I must have bumped my ISO to 25600.

The First shot was well overexposed, largely because when using an AD600 Pro, the minimum power at 1/128 was still too powerful for f/16 @ ISO25600.



I switched down the ISO to 200 and the result was this under exposed image at f/16.



The purpose of this isn’t to show my lazy stock photography skills, unlike shooting with a model, I am sometimes lazy when it comes to camera setup because I can easily change the settings to what I want, and as a result within 2 or 3 frames it’s pretty much perfect. The purpose of this article was to show 2 things that people constantly complain about…dynamic range and high ISO noise.

Let’s take shot number one which was overexposed. JPG has noise reduction built in which is obviously fairly good, so if I drop the exposure by 2.5 stops and correct the white balance, I am left with this, a perfectly good stock image…taken at a ridiculous ISO25600.


If I did some basic editing to the raw file, 20 noise reduction in Lightroom, set the white balance to flash, some small vignette, set the film simulation to Class Chrome, I get this. It’s not spectacular, but it’s still pretty decent, when you consider it is an ISO25600 file.


I’d argue that if I posted this on instagram, most people wouldn’t actually know it was ISO25600.

Obviously this is one extreme, but lets take another. My next photo taken at ISO200 isn’t worth using for stock, but let’s lift the exposure on the jpg by 5 stops and apply 20 noise reduction, then this is what we get. It’s not horrendous but obviously expecting dynamic range in a JPG is unreasonable so it isn’t great either. You can see the discoloration in the noise:


What people often forget however is even with the discoloration of the noise, you could switch to black and white and still get a nice image.


Playing with the raw file gives us much better results. This is with some basic vignette, 4 stops of lifted exposure, and 20 noise reduction.



This isn't an attempt to sell you on the virtues of APSC, but it’s clear that many of the people who feel they need full frame because of high ISO noise, dynamic range and any other reason they might have, are probably kidding themselves unless they are really pushing the limits. There is nothing wrong with APSC. The image quality of these is better than full frame was 5 years ago, and most photographers were producing spectacular images then.

[swifty]

Hi MM. I agree with your conclusion but I don't think your post is a demonstration of this.
It is more of a demonstration of exposure and how your particular camera's sensor handles noise with respect to in-camera ISO settings and post processing.

What you have to remember with respect to flash exposure is that the flash duration becomes your effective shutter speed (provided your ambient exposure settings contributes effective nothing to the final image).
So for your images, the exposure between the 25600 ISO shot and 200 ISO shot has not changed at all.
It is at f16 and I'll have to guess a typical flash duration, lets say 1/1000s (flash duration often varies between something like 1/800s and 1/8000s).
Now to get your desired image brightness you pulled back the 25600 ISO image 2.5 stops or pushed up the ISO 200 image 4 stops so it roughly equates to between ISO 3200 and 6400. Let's use 6400 for argument sake.

So what you're demonstrating is a scenario equivalent to a scene where you'd use ISO 6400, f16 and 1/1000s to get your desired image brightness, which isn't a particularly low light scene.
It would be the same as f5.6, 1/500s and ISO 400 for example.

So provided your overexposed image with ISO 25600 didn't clip any highlights, it should yield more or less the same if not better results than the lifted ISO 200 image after PP.
All this on images where there is plenty of signal, so results should be good regardless of what ISO you shot it at. This is a demonstration of why good SNR is important.

If you want to reveal some of the difference between sensor sizes in low light, you need to demonstrate using a scenario with low light levels where shot noise predominates and where there's very little signal.
If your strobe is already at its minimum power, you'd just need to move it much further away from your setup or use a continuous light source at a low enough level.

A typical low light scenario might be f1.4, 1/60s using maybe ISO 6400 and above to produce a desirable image brightness.
Compare to f16, 1/1000 ISO 6400, that's (7+4) 11 stops less light than your demonstration.

PS, I'll re-check my math because I did it quite quickly but I believe my premise is correct.

[ameerat42]

^^ Well reasoned. Although [MM's] shots are interesting - the camera does not seem to be a slouch in
dim conditions - it is not a test with what you [MM] are trying to compare. I suppose if you had two
late model cameras side-by-side, one a FF and the other an APSC, and tried to get all other conditions
equal, you could then run a series of comparison shots...

It would be hard to control everything, though, because apart from the photography, you'd have to get
the PP of the images "pretty equal" and also a valid yardstick for comparison of the final images.

[MissionMan]

^^ Well reasoned. Although [MM's] shots are interesting - the camera does not seem to be a slouch in
dim conditions - it is not a test with what you [MM] are trying to compare. I suppose if you had two
late model cameras side-by-side, one a FF and the other an APSC, and tried to get all other conditions
equal, you could then run a series of comparison shots...

It would be hard to control everything, though, because apart from the photography, you'd have to get
the PP of the images "pretty equal" and also a valid yardstick for comparison of the final images.

I don't think there is any question that full frame is better, or that medium format is better than full frame as an example, it's more the question of when is it going to be enough that people stop considering high ISO to be a constraint" and go back to what's real which is the photography itself? It's a little like the question of megapixels. Is it when we have clean ISO100 like photos at ISO51200 or 102400? Or are we constantly going to be chasing an elephant? I think people get tied to specs, pixel peeping and worry about image sharpness at levels where it is unlikely to make a difference. Having 5 stops of image recovery is great, but the amount of times that having 5 stops makes the difference between a good photo and a spectacular photo are relatively minuscule. Cameras already have 5 stops of image recovery so what next? 10 stops? And when will you actually need that?

[Tannin]

Good, thoughtful posting, MissionMan. As you say, there comes a point where further improvements achieve nothing. Similarly, working in the opposite direction, there comes a point where skimping on the gear costs you more in quality than you are willing to pay. It's all about finding the level of technical excellence which suite what you are doing.

So where is this point? It all depends on the task. The one place where full frame really proves its worth in my photographic world is the rainforest. (Or any other place with similar lighting challenges.) Birding in those conditions, the difference between two cameras of roughly equal ability otherwise, but one full frame, the other one not, is the difference between good shots and throwaways. Most of the other stuff I do, it really doesn't make a huge difference.

Your mileage will vary, of course.

[MissionMan]

but one full frame, the other one not, is the difference between good shots and throwaways.

How so? There is about 1 stop difference, maybe 1.5 depending on the generation of sensor. Are you saying 1 stop is a show stopper? That said, some sensors are worse than others, for example Nikon’s APSC was better than Canon’s full frame fairly recently so that would essentially mean the Canon full frame could not achieve a good photo under the given circumstances?

[Tannin]

Noise is primarily a function of the total number of photons captured, which (for any given lens and lighting) is a function of the size of the sensor.

Is one stop a show stopper? Of course it is! Blimey, it's the difference between 800 ISO and 400 ISO, or 1600 and 800. A one stop difference means that one camera is twice as good as the other one.

Where noise is a factor (and in bird work that is practically all the time because you can't use slow shutter speeds - as a general rule, you want to be over 1/1000th - and you have limited aperture (generally f/4 for pro-grade lenses, f/5.6 or even f/6.3 for amatuer ones), a larger sensor can make a huge difference.

Now in a lot of bird work, the light is half-decent and you are further away from the subject than you would like. You can use a crop camera and crop a bit, or you can use a full frame camera and crop a lot. Result: it's a bit of a wash. I generally prefer to use the crop camera, but there is no compelling reason either way.

But in poor light (rainforest is the obvious example) you are less limited by focal length because birds are highly visual and will let you get much closer in bad light than they do in bright light. This means that you can often fill the frame even with a full frame body. This is where it makes a big difference to the final result.

[ameerat42]

...Noise is primarily a function of the total number of photons captured, which (for any given lens and lighting) is a function of the size of the sensor.
...

I (sort of) know what you mean, and so would MM, Swifty, and a few others, but how is this for the general public?
Is it a "direct" function, or an "inverse" one? As it stands it appears as if "the more photons captured, the more noise".

[Tannin]

Quite so, AM. It is obvious, nevertheless I should have said "inverse" to ensure no misunderstanding.

[MissionMan]

Noise is primarily a function of the total number of photons captured, which (for any given lens and lighting) is a function of the size of the sensor.

Is one stop a show stopper? Of course it is! Blimey, it's the difference between 800 ISO and 400 ISO, or 1600 and 800. A one stop difference means that one camera is twice as good as the other one.

Where noise is a factor (and in bird work that is practically all the time because you can't use slow shutter speeds - as a general rule, you want to be over 1/1000th - and you have limited aperture (generally f/4 for pro-grade lenses, f/5.6 or even f/6.3 for amatuer ones), a larger sensor can make a huge difference.

Now in a lot of bird work, the light is half-decent and you are further away from the subject than you would like. You can use a crop camera and crop a bit, or you can use a full frame camera and crop a lot. Result: it's a bit of a wash. I generally prefer to use the crop camera, but there is no compelling reason either way.

But in poor light (rainforest is the obvious example) you are less limited by focal length because birds are highly visual and will let you get much closer in bad light than they do in bright light. This means that you can often fill the frame even with a full frame body. This is where it makes a big difference to the final result.

Thanks Tannin

Good to know. I don't shoot birds so I guess it's never been a problem for me.

On a separate note, I do notice that a lot of people on forums complaining about "needing" full frame (present company excluded and not this forum, more the technical forums like dpreview) for low light are often the ones who take bad photos of their cats

[swifty]

On a separate note, I do notice that a lot of people on forums complaining about "needing" full frame (present company excluded and not this forum, more the technical forums like dpreview) for low light are often the ones who take bad photos of their cats

I see it all the time. It's a scapegoat and it's easy to blame equipment. I even get ppl looking down at me for using an older generation FF even outside of technical discussions (read: man, I can't believe you're still using a D700, that's so ancient. At least upgrade to a D750).
I get the feeling they might be the same people who say, I can't believe you're using APS-C, m43, Nikon 1 etc etc.
You can't beat them, but in this case I'm not going to join them.
I just smile and move along.

[MissionMan]

I see it all the time. It's a scapegoat and it's easy to blame equipment. I even get ppl looking down at me for using an older generation FF even outside of technical discussions (read: man, I can't believe you're still using a D700, that's so ancient. At least upgrade to a D750).
I get the feeling they might be the same people who say, I can't believe you're using APS-C, m43, Nikon 1 etc etc.
You can't beat them, but in this case I'm not going to join them.
I just smile and move along.

I had the same issue on Sunday. Went for a short motorcycle ride with a mate and stopped at princess pier to take a photo of his bike. Midday, so wasn't ideal timing but it was more about getting him a photo on his bike while we were on route than it being a shoot.

Guy starts watching me take photo with an X100F (it's a $1500 P&S with a leaf shutter and a 23mm prime, so image quality is pretty damn good), and lectures me that I should just take a photo with my iPhone because no one will know the difference and if I want to take good photos, I need something like a D850. Nodded my head, smiled, pretended to care about the fact that the new tamron 70-200 he was buying has a metal body (whilst my 90mmf2 @#$% all over it from a sharpness perspective.)

[John King]

Thanks Tannin

Good to know. I don't shoot birds so I guess it's never been a problem for me.

On a separate note, I do notice that a lot of people on forums complaining about "needing" full frame (present company excluded and not this forum, more the technical forums like dpreview) for low light are often the ones who take bad photos of their cats

Is this one bad enough?

[Tannin]

Very hard to see the bird clearly when it's inside a cat, John. This is actually a very good example of the superiority of using good quality, up-to-date photographic equipment. You need one of these: http://www.medicalequipment168.com/p...y_Machine.html

[John King]

Very hard to see the bird clearly when it's inside a cat, John. This is actually a very good example of the superiority of using good quality, up-to-date photographic equipment. You need one of these: http://www.medicalequipment168.com/p...y_Machine.html

No, Tony, one needs an Olympus operating endoscope ... .

[swifty]

No, Tony, one needs an Olympus operating endoscope ... .

No need. Just look for the feathers in the hairball it'll cough up

[MissionMan]

Very hard to see the bird clearly when it's inside a cat, John. This is actually a very good example of the superiority of using good quality, up-to-date photographic equipment. You need one of these: http://www.medicalequipment168.com/p...y_Machine.html

Lets not turn this into a mirrorless vs dslr discussion or it will really go downhill

[John King]

No need. Just look for the feathers in the hairball it'll cough up

No chance of that with our girls. De-sexed, inside cats ... . Only allowed out with harness, lead and supervised.

- - - Updated - - -

This isn't an attempt to sell you on the virtues of APSC, but it’s clear that many of the people who feel they need full frame because of high ISO noise, dynamic range and any other reason they might have, are probably kidding themselves unless they are really pushing the limits. There is nothing wrong with APSC. The image quality of these is better than full frame was 5 years ago, and most photographers were producing spectacular images then.

Back on topic, Athol.

The differences between formats are grossly exaggerated IMHO. They do exist, but some of the overblown rhetoric one reads boggles the mind.