p.1 #1 · The main (theoretical) differences between APS-C and Full-Frame?
I post this in the spirit of a question rather than an assertion because I don't have a strong understanding of the issues involved.
Sony is reported to introduce an A6700 aps-c camera on July 12th that is claimed to better the Fuji X-T5. I find aps-c cameras appealing because of their size and light weight, though the full-frame A7C is also very appealing for the same reasons.
But the appearance of the A6700 again raises the question of the trade-offs between the aps-c format and full frame. The differences that are due to sensor size differences are often stated as:
1. Increased depth of field but limited selective focus for aps-c sensors.
2. That the Sony aps-c sensor gets only half the light as a full-frame sensor (because less light falls on the smaller sensor) making low-light photography more difficult and also having other effects, as listed in #3 below.
3. That noise is greater with an aps-c sensor because of the lesser amount of light and the often-smaller pixel size of an aps-c sensor (leading to a worse signal to noise ratio).
4. That the aps-c sensor will usually have less resolution because it will usually contain fewer pixels than a full-frame sensor.
My question really concerns the idea that the aps-c sensor gets less light. This is often stated as the smaller sensor getting a stop less light, as though f2.0 on an aps-c is really like f2.8 on a full-frame. But I wonder if this is a misleading way of stating the difference in light and one that leads to misunderstandings.
While the aps-c sensor as a whole gets less light, isn't it the case that the amount of light per square centimeter of each sensor is the same in the two formats and that there are simply fewer square centimeters of sensor on aps-c? This means that the smaller sensor must be more enlarged than the full-frame sensor to reach the same image size. This is why the aps-c image shows more noise and less resolution. But it is not the case, as is sometimes implied, that a scene shot at f2.8 on full-frame requires an exposure of f2.0 on aps-c.
If so, then the main differences between an image from an aps-c and one from a full-frame sensor would really be a matter of the amount of noise (and often but not always the degree of resolution).
If so, given the improved efficacy of noise-reducing software and the higher resolution of modern sensors, these differences may be less important than they once were. This would make aps-c sensor bodies more appealing relative to full-frame than they once were.
As I said in my first sentence, I make this post as a question rather than an assertion because of my limited knowledge. Is my understanding of the effects of sensor size correct, as stated above? If not, what have I misunderstood or left out?
Jun 28, 2023 at 05:46 PM
berimbolo Offline • • • [X]
p.1 #2 · The main (theoretical) differences between APS-C and Full-Frame?
p.1 #3 · The main (theoretical) differences between APS-C and Full-Frame?
I took some older aps-c shots and reprocessed them recently with impressive results. So while the improvements apply to all the sensors, I'd not be surprised if some/more people find aps-c quite fine for their uses. If they like the A7C/A6xxx size advantages, great. I noticed quickly and still do that my A7riv and 28-200 was larger and pound or so heavier than my A6400 and 18-135. For some use, a hot new aps-c could be quite well received.
p.1 #4 · The main (theoretical) differences between APS-C and Full-Frame?
These equivalence arguments always get confusing because many different assumptions need to be made, and they're not always stated. So in this case, if I have a 24 megapixel full frame camera and a 24 megapixel APSC camera, then the pixels of the APSC camera will be smaller, so they will gather less light. If there is less light on a pixel its signal to noise ratio will be lower, which shows up as sensor noise.
But if you had a full frame camera whose pixels size was the same as an APSC, like the A7riv and v then the noise per pixel would be the same as the noise per pixel of the APSC camera. In that case, the argument is different. The full frame camera has many more pixels than the APSC. So to compare the full frame to the APSC fairly you would downsample the full frame from 60 to 24 megapixels. When you downsample you get to average some of the noise away. So the full frame image will still have less noise at the same output size. This is a factor that people who pixel peep the riv and rv sometimes forget.
But in the end, this is all theoretical. If a small APSC camera lets you carry it places where you wouldn't take the full frame, then who cares about a stop more noise. Or if you're shooting in good light, and are at ISO 100, will you see the added noise at all?
p.1 #5 · The main (theoretical) differences between APS-C and Full-Frame?
"While the aps-c sensor as a whole gets less light, isn't it the case that the amount of light per square centimeter of each sensor is the same in the two formats?"
The amount of light per unit surface correlates with the ISO, not the f value. One can reasonably expect that a cropped-sensor camera at a high ISO, e.g. 12800, should produce a noisier image compared to an equivalently framed image acquired at the same ISO with a full-frame camera.
"This means that the smaller sensor must be more enlarged than the full-frame sensor to reach the same image size. This is why the aps-c image shows more noise and less resolution. But it is not the case, as is sometimes implied, that a scene shot at f2.8 on full-frame requires an exposure of f2.0 on aps-c."
No, it is the case. This is what actually happens when a TC is added to a lens on a full-frame camera. The TC effectively optically crops the image coming from the lens and projects the crop onto the entire sensor. When I add a 1.4x TC to a f2.8 lens, the camera tells me that my aperture is now effectively f4. With a 2x TC, the camera knows that the aperture is effectively reduced to f5.6 even if the aperture ring on the lens remains at f2.8.
Ignoring the optical artifacts added by the TC, one can expect that the above image obtained with the lens and a 1.4x TC should look like the uncropped f2.8 image obtained without the TC and subsequently cropped by the factor of 1.4 in post. That is, it does not matter how an image is cropped; the corresponding effective f value increases with the crop factor (1.4 = one stop, 2 = two stops, etc.).
All of the above applies to the same sensor or equivalent sensors. Understandably, if the sensor technology is different in two cameras, then it becomes difficult to know the actual differences in the signal-to-noise ratio between the two. What remains unchanged is the difference in DOF. For example, if the crop factor is 2 then equivalently framed images obtained using the cropped-sensor camera and a full-frame camera should have the same DOF when the f value of the former is 1/2 of the f value of the latter.
There are very different reasons for choosing or not choosing a cropped-sensor camera. My own reason for sticking to FF is the the MP count. I like the ability to crop 50 MP images from my A1 when this is needed. Buying a "pre-cropped" camera is not very attractive, from this particular point of view. For others, the obsession with ever smaller, lighter, and more compact camera systems may be a compelling reason to own and use a cropped-sensor camera. I personally don't care much about the size of my lenses or the weight of the entire system in my hands.
p.1 #6 · The main (theoretical) differences between APS-C and Full-Frame?
Sounds like you have a pretty good understanding. I think it's important to consider the various tradeoffs. One of the most important is lens availability and how lens selection compares after applying equivalence. For me the main drawback to aps-c on Sony is lens selection vs FF Sony. There are some issues related to noise, but as you mentioned this may not be a big deal with modern software. Also the issues with noise are not always present in many cases.
For equivalence (to make an "identical" picture) we need to adjust focal length, aperture, and Iso. Shutter speed remains the same. For a real world example:
We might have
FF 35mm f/2 1/60s at ISO 400
vs APS-C 23mm f1/4 1/60s at ISO 200
apart from possible resolution it should be very hard to tell these photos apart. And Sigma makes both of these lenses in the contemporary line.
The Sigma 23mm/1.4 DC DN is $550 with 3.1" long at 330g
The Sigma 35mm/2.0 DG DN is $640 with 2.7" long at 325g
We would expect equivalent lenses to be roughly the same size, and in this case they are close. The first difference we find when shooting these lenses is the noise when there is sufficient light. If we have enough light to shoot the FF at ISO 100, it will have less noise than the APS-C because the base ISO of the aps-c camera is ISO 100, so on the aps-c we need to increase shutter speed and now the photos are no longer equivalent and the FF has 1 stop less noise (Full well capacity difference noted in lens rentals article). Note this difference in noise with equivalent lenses only happens when you have enough light to expose the FF at ISO 100. I personally don't care about noise differences between ISO 100 & 200 they both look great. In bright light you may have a problem with shutter speed limitations on the aps-c as it will need to operate the shutter a stop faster.
When thinking about lens selection there are limitations, we could get a 35GM with f/1.4, but there is no 23mm f/1.0 equivalent lens for aps-c. This will allow more light at the expense of less depth of field compared to the nearest equivalent of the 23mm/1.4. A similar comment can be made for f/2.8 zooms, we typically don't find f/2.0 zooms. The major exception being the Sigma f/1.8 zooms made for canon EF, but those have a shorter zoom range than available in FF. The reverse example can be made as well. You won't find a FF zoom that covers the 100-500mm range the same size as the 70-350mm for aps-c.
Jun 28, 2023 at 10:16 PM
AmbientMike Offline • • • • • • [X]
p.1 #7 · The main (theoretical) differences between APS-C and Full-Frame?
I've looked at it like this: the smaller the pixels the less light falls on them individually. And the more it has to be amplified so you have more noise.
p.1 #8 · The main (theoretical) differences between APS-C and Full-Frame?
Everything has been written on the topic already here but I think it wasn't mentioned explicitly that dynamic range (DR) is also affected with a penalty of about a stop...
p.1 #9 · The main (theoretical) differences between APS-C and Full-Frame?
I'm no expert, but I think you've got it right. I agree that the "gets less light" concept is misleading. The sensor isn't a solar cell that stores more or less energy depending on size. Exposure is the same, regardless of sensor size. But smaller pixels are generally less efficient and more noisy, and the image would be enlarged more to create the same output size (if that's the goal). The "requirement" of using a wider aperture isn't a requirement so much as a way of getting a similar depth of field (if that's the goal). Aps-c is good overall, but does have a little deeper DOF (if using the same aperture), a little more noise, and a little less detail — all for the reasons you stated.
p.1 #10 · The main (theoretical) differences between APS-C and Full-Frame?
AmbientMike wrote:
I've looked at it like this: the smaller the pixels the less light falls on them individually. And the more it has to be amplified so you have more noise.
It is worth noting that the highest pixel pitch Sony camera is the A7RIV/V, which match the 26MP APS-C sensors. You will have to go to Canon (32.5MP @ 1.6x) or Fuji ([email protected]) to get higher pixel density and smaller pixels than FF from an APS-C body today.
However the more pixels you have, the less visible per-pixel noise is in a given composition. It's a tradeoff and generally with modern tech the defining factor in visible noise is total illumination of the sensor (or crop area), not pixel pitch, the smaller pixels are offset by having more of them. This is also why evaluating noise at 100% view is mostly useless on high-MP cameras, it does not well represent the visible noise in the final image presentation.
Jun 29, 2023 at 08:28 AM
Steve Spencer Offline • • • • • • • Upload & Sell: On
p.1 #11 · The main (theoretical) differences between APS-C and Full-Frame?
chiron wrote:
I post this in the spirit of a question rather than an assertion because I don't have a strong understanding of the issues involved.
Sony is reported to introduce an A6700 aps-c camera on July 12th that is claimed to better the Fuji X-T5. I find aps-c cameras appealing because of their size and light weight, though the full-frame A7C is also very appealing for the same reasons.
But the appearance of the A6700 again raises the question of the trade-offs between the aps-c format and full frame. The differences that are due to sensor size differences are often stated as:
1. Increased depth of field but limited selective focus for aps-c sensors.
2. That the Sony aps-c sensor gets only half the light as a full-frame sensor (because less light falls on the smaller sensor) making low-light photography more difficult and also having other effects, as listed in #3 below.
3. That noise is greater with an aps-c sensor because of the lesser amount of light and the often-smaller pixel size of an aps-c sensor (leading to a worse signal to noise ratio).
4. That the aps-c sensor will usually have less resolution because it will usually contain fewer pixels than a full-frame sensor.
My question really concerns the idea that the aps-c sensor gets less light. This is often stated as the smaller sensor getting a stop less light, as though f2.0 on an aps-c is really like f2.8 on a full-frame. But I wonder if this is a misleading way of stating the difference in light and one that leads to misunderstandings.
While the aps-c sensor as a whole gets less light, isn't it the case that the amount of light per square centimeter of each sensor is the same in the two formats and that there are simply fewer square centimeters of sensor on aps-c? This means that the smaller sensor must be more enlarged than the full-frame sensor to reach the same image size. This is why the aps-c image shows more noise and less resolution. But it is not the case, as is sometimes implied, that a scene shot at f2.8 on full-frame requires an exposure of f2.0 on aps-c.
If so, then the main differences between an image from an aps-c and one from a full-frame sensor would really be a matter of the amount of noise (and often but not always the degree of resolution).
If so, given the improved efficacy of noise-reducing software and the higher resolution of modern sensors, these differences may be less important than they once were. This would make aps-c sensor bodies more appealing relative to full-frame than they once were.
As I said in my first sentence, I make this post as a question rather than an assertion because of my limited knowledge. Is my understanding of the effects of sensor size correct, as stated above? If not, what have I misunderstood or left out?...Show more →
This is a good summary, and as others have said you basically got it right. The only trade off isn't noise, however. You will get about a stop less color depth (it is hard to notice this but in general I can see that a shot at ISO 100 has a bit better color than a shot at say ISO 800, but we are talking here a shot at ISO 100 vs a shot at ISO 200 or so), and that might matter to someone, but in my view is a pretty small difference.
I think the major differences is the lenses that are available and how they perform. At this point there are some really nice quite wide lenses like the Voigtlander 21 f/1.4, the Sony 20 f/1.8 G, the Zeiss Batis 18 f/2.8, the Zeiss Loxia 21 f/2.8 and 25 f/2.5, that really have nothing close to the same performance at the same angle of view available for APS-C. Perhaps that gap will be filled but for me that would be the major reason I would have trouble switching to APS-C.
p.1 #12 · The main (theoretical) differences between APS-C and Full-Frame?
Yes to Chiron. The same logic applies to m43. There you get effectively two stops of extra depth of field and 2 stops worse noise, but counterbalancing this, because most shots require more rather than less depth of field (unless you are obsessed with shooting at f1.4 for everything) you gain back the 2-stops of light because you can open the lens up two stops for the same depth of field. Only when you are at the limits of ISO sensitivity does the FF advantage come to the fore (in very dim environments and you have a requirement to stop action so have to set a fast shutter speed).
p.1 #13 · The main (theoretical) differences between APS-C and Full-Frame?
Steve Spencer wrote:
I think the major differences is the lenses that are available and how they perform. At this point there are some really nice quite wide lenses like the Voigtlander 21 f/1.4, the Sony 20 f/1.8 G, the Zeiss Batis 18 f/2.8, the Zeiss Loxia 21 f/2.8 and 25 f/2.5, that really have nothing close to the same performance at the same angle of view available for APS-C. Perhaps that gap will be filled but for me that would be the major reason I would have trouble switching to APS-C.
Indeed. This becomes even more apparent if you like to shoot very wide. If you want a 12mm FF field of view, especially with a wide aperture like the 12-24mm GM, there are no equivalents on APS-C. Or the excellent 14mm ƒ1.8; no remotely comparable option available.
p.1 #14 · The main (theoretical) differences between APS-C and Full-Frame?
Replying to original post with my take
There is only 1 difference between APS-C and Full Frame: The sensor size.
If you take a picture on your full frame sensor, and crop down to APS-C you have now simulated the difference. That is it - a reduced FOV.
All of the equivalencies are based on recouping that FOV by either moving closer to your subject or changing the focal length - both of which have many theoretical implications (magnification, DoF, light on sensor etc etc)
In addition, Crop Sensors TEND to have smaller pixels historically and tend to be cheaper and older meaning they have worse noise performance. But If Sony made an A1 that only had crop mode (due to smaller sensor) that would be the only different - a reduced FOV.
p.1 #15 · The main (theoretical) differences between APS-C and Full-Frame?
If the A6700 ends up being 26MP crop sensor then that would have all the same properties as taking an A7RV image in APS-C mode or cropping a FF A7RV image to APS-C size after the fact. If everything is held constant, lens, subject distance, aperture etc.
p.1 #16 · The main (theoretical) differences between APS-C and Full-Frame?
chiron wrote:
If so, given the improved efficacy of noise-reducing software and the higher resolution of modern sensors, these differences may be less important than they once were. This would make aps-c sensor bodies more appealing relative to full-frame than they once were.
Yes and no. When some of the newer AI based noise reduction programs work, it's like magic and the output is fantastic. But they are a hassle to use because they sometimes produce weird artifacts so I always have to go over every image very carefully. So my preference with noise is always better initial capture --> fairly safe methods for noise reduction with low likelihood of artifacts (like stacking, exposure blending) --> AI based noise reduction.
p.1 #17 · The main (theoretical) differences between APS-C and Full-Frame?
sbay wrote:
Yes and no. When some of the newer AI based noise reduction programs work, it's like magic and the output is fantastic. But they are a hassle to use because they sometimes produce weird artifacts so I always have to go over every image very carefully. So my preference with noise is always better initial capture --> fairly safe methods for noise reduction with low likelihood of artifacts (like stacking, exposure blending) --> AI based noise reduction.
I treat AI noise reduction like healing brush and content aware fill. It's just easier to do it right to begin with than to fix it digitally in post. I do shoot a lot each time, so AI RAW treatments can be a bit inefficient.
p.1 #18 · The main (theoretical) differences between APS-C and Full-Frame?
FF outperforms APS-C not just in noise but in potential acuity. At equal pixel sizes (24MP 1.5x APS-C / 54MP FF) per unit of area on the sensor, the FF image will yield higher resolution for lenses of equal performance due to the lower lp/mm demands placed on the optics. In other words, a lower magnification factor yields better results.
The only scenario where APS-C outperforms FF is for reach-limited, abundant light cases, and only when the pixel size of the APS-C is smaller vs FF (ie, higher pixel density), which used to be the norm until a few years ago when high-resolution FF bodies were offered by all major camera makers. But high-resolution FF bodies carry a premium so APS-C still holds a potential cost advantage for reach-per-dollar, aka dollars per pixel per duck.
Pixel size / count has no influence on noise except for ultra-high ISO shooting, where the aggregate read noise is higher and thus DR is lower. Otherwise (shot) noise is the same at normalized resolutions. The situation is opposite at low ISO - smaller (more) pixels means higher DR on an image basis.
APS-C does get less total light vs FF for equivalent FOV situations, thus yields less noise, at least when not DOF-limited where you'd otherwise have to adjust the FF aperture to get sufficient DOF.
p.1 #19 · The main (theoretical) differences between APS-C and Full-Frame?
snapsy wrote:
FF outperforms APS-C not just in noise but in potential acuity. At equal pixel sizes (24MP 1.5x APS-C / 54MP FF) per unit of area on the sensor, the FF image will yield higher resolution for lenses of equal performance due to the lower lp/mm demands placed on the optics. In other words, a lower magnification factor yields better results.
The only scenario where APS-C outperforms FF is for reach-limited, abundant light cases, and only when the pixel size of the APS-C is smaller vs FF (ie, higher pixel density), which used to be the norm until a few years ago when high-resolution FF bodies were offered by all major camera makers. But high-resolution FF bodies carry a premium so APS-C still holds a potential cost advantage for reach-per-dollar, aka dollars per pixel per duck.
Pixel size / count has no influence on noise except for ultra-high ISO shooting, where the aggregate read noise is higher and thus DR is lower. Otherwise (shot) noise is the same at normalized resolutions. The situation is opposite at low ISO - smaller (more) pixels means higher DR on an image basis.
APS-C does get less total light vs FF for equivalent FOV situations, thus yields less noise, at least when not DOF-limited where you'd otherwise have to adjust the FF aperture to get sufficient DOF....Show more →
Yes, I said less resolution, for the same reason as more noise--a greater degree of magnification. But neither of those are necessarily important for many varieties of photographic practice.
I think the way you state the less light issue is inherently confusing and unclear because of the tucked in field of view qualifier that obscures the magnification issue. Not that you are incorrect, but unclear.
I think it is clearer to say that the issue that matters is magnification, not less light. An aps-c sized frame cropped from an A7RIV (equivalent to a 26mpix aps-c sensor) won't be underexposed relative to the exposure of the full-frame image from which it was taken. It will merely acquire more enlargement of its noise as it is enlarged to a size equal to the full-frame shot. The difference is different degrees of magnification of the noise that is present in both, not differences in the light received by the pixel.
p.1 #20 · The main (theoretical) differences between APS-C and Full-Frame?
chiron wrote:
Yes, I said less resolution, for the same reason as more noise--a greater degree of magnification. But neither of those are necessarily important for many varieties of photographic practice.
I think the way you state the less light issue is inherently confusing and unclear because of the tucked in field of view qualifier that obscures the magnification issue. Not that you are incorrect, but unclear.
I think it is clearer to say that the issue that matters is magnification, not less light. An aps-c sized frame cropped from an A7RIV (equivalent to a 26mpix aps-c sensor) won't be underexposed relative to the exposure of the full-frame image from which it was taken. It will merely acquire more enlargement of its noise as it is enlarged to a size equal to the full-frame shot. The difference is different degrees of magnification of the noise that is present in both, not differences in the light received by the pixel....Show more →
I guess it comes down to personal preference. In my mind, using terms like "underexposed" when comparing sensor sizes is pretty confusing. Same for "enlarging noise". How does one form a mental picture of that?
And the consequences of the magnification differences only apply at equal FOV, so using it as the overarching base to explain all the differences between APS-C and FF doesn't quite work. The APS-C portion of the projection from a 300mm lens will be identical between APS-C and FF sensors of equal pixel design/pitch, ie identical acuity and noise for the duck. This is the reach-limited scenario, which is a significant consideration when deciding on sensor size. If one only shoots wildlife then the reach-limited scenario may be their only consideration.
but I think it wasn't mentioned explicitly that dynamic range (DR) is also affected with a penalty of about a stop...