Accurately comparing FF vs APS-C sensor performance? An open discussion.

Started Apr 8, 2013 | Discussions thread
rf-design Regular Member • Posts: 496
Re: Accurately comparing FF vs APS-C sensor performance? An open discussion.

I did asking a similar question 3 month ago:

My motivation is to reduce the time to capture a night scene with a required DOF and signal to noise ratio. The result of my considerations and other posters was that there is no difference between FF and APS-C. If you step down the apture also the diffraction limited limited resolution for objects in focus will be equivalent between APS-C and FF.

So only two specific advantages for FF remain:

1. Geometric resolution improvement until diffraction limits

Any lens construction have a geometric resolution which improves if you close the apture. If you scale down a FF lens construction to APS-C it will have the same geometric resolution but the diffraction limit starts 1 1/3 apture step earlier. That is why the best large format lenses start at 5.6, have the peak resolution between 11 and 16 and have still better total resolution than any FF lens.

2. 1 1/3 stop advantages where DOF does not matter and the center of the lens is sharp enough before diffraction starts.

VirtualMirage wrote:

Out of boredom and curiosity, I decided to explore the differences between some of the current top sensors (both full frame and APS-C) as well as the implications one may face when going from APS-C to full frame.  I created a beautiful spreadsheet with tons of information.  I did this more so to better my understanding of the different technologies and how it may or may not suit my needs.  These are my observations, and as such there might be something that might need correction. To simplify this, I will only be talking about the A77 and A99.  Upon open discussion, we can talk about other cameras too.
I won't go into all the advantages and disadvantages of Full Frame versus APS-C.  By now, I think most of understand this and can see a world where both have their place.  But the one thing I do want to talk about is the comparison of two when it comes to noise, dynamic range, etc. and how it may not always seem like those advantages may be visible in certain situations.
Here is something I don't many consider when contemplating the jump to Full Frame:
The relation of depth of field to focal length equivalents and their effects on required ISO when matching or exceeding shutter speed.
Wow! Sounds like a title to a thesis.  Anyways...
Some of you might be scratching your head, so hear me out on this.
On a 1.5x crop APS-C sensor, a 50mm lens has a focal length equivalent (or field of view) closer to a 75mm lens.  I think most of us are familiar with this.  We also know that Full Frame sensors offer a shallower depth of field.  So if an APS-C sensor is using a 50mm lens (75mm equivalent) and we compared it to a Full Frame using a 75mm lens (keeping focal length equivalents identical), the Full Frame sensor will have a shallower depth of field at the same aperture settings of its smaller APS-C cousin.  How much of a difference, you might be wondering?  Well, if the APS-C is set to F/1.4, the Full Frame will need to be set to F/2.2 for a similar depth of field.  That is a difference of around 1 1/4 to 1 1/3 stops.
Ok, so now we know if we want a similar depth of field the lens will need to be stopped down around 1 1/4 to 1 1/3 stops (1 1/3 stops might be closer option since lenses usually will do 1/3 stop increments).  But if we are at a fixed ISO, what does this do?  It slows the shutter speed down.  So what if we need that same, or faster shutter speed, that we are getting on the APS-C for the similar depth of field?  We will then need to up the ISO.  How much?  Again, around 1 1/3 stops to meet or exceed the shutter speed of the APS-C equivalent.  So if we were shooting at ISO 800 on an APS-C, we will need to shoot at ISO 2000 on the Full Frame.
So, for situations like this we see that a Full Frame will need to shoot at 1 1/3 stops smaller aperture and 1 1/3 stops faster ISO speed than its APS-C cousin to match its depth of field and shutter speed.  It is usually generally agreed that Full Frame sensors offer around 1 1/3 to 2 stops advantage (+/-) when comparing ISO to ISO against the APS-C.  So a picture on a Full Frame at ISO 3200 may look equivalent or better (noise wise) than the same picture on a APS-C at ISO 800 - 1250.  But if you need to match the depth of field and shutter speed at the same equivalent focal length of the APS-C, you immediately forfeit 1 1/3 stops of that advantage.  So what does that leave you? Anywhere from no advantage (but no disadvantage either) or maybe 2/3 stop advantage over the APS-C.
For example, take a look at the DXOMark results for the A77 and A99.  I know these are lab results and will vary from the real world, but the lab results are the only ones I can look at with consistent, repeatable results from one camera to the next.  So bear with me here.  Also, to simplify things I am looking directly at the Manufacturer's ISO and not the measured ISO.
Comparing the results side by side, the A99 shows an advantage average of around 1 1/3 stops across the board in all sections.  But when you apply what I just divulged above, the advantage becomes close to nil (a few .1-.3 differences in the A99's favor here and there).  Below are just a few extractions from those results (APS-C ISO versus Full Frame ISO + 1 1/3 stops):
SNR 18% (dB, Screen/Print):

  • ISO 400: 29.2 / 34.0
  • ISO 1600: 23.1 / 27.9


  • ISO 1000: 29.5 / 34.3
  • ISO 4000: 23.5 / 28.3

Difference: .3-.4 / .3-.4

Dynamic Range (EV, Screen/Print):

  • ISO 400: 10.4 / 11.2
  • ISO 1600: 8.5 / 9.3


  • ISO 1000: 10.5 / 11.3
  • ISO 4000: 8.5 / 9.3

Difference: 0-.1 / 0-.1

Tonal Range (bits, Screen/Print):

  • ISO 400: 7.0 / 7.8
  • ISO 1600: 6.0 / 6.8


  • ISO 1000: 7.1 / 7.9
  • ISO 4000: 6.1 / 6.9

Difference: .1 / .1

Color Sensitivity (bits, Screen/Print):

  • ISO 400: 18.1 / 20.5
  • ISO 1600: 14.6 / 17.0


  • ISO 1000: 18.5 / 20.9
  • ISO 4000: 14.9 / 17.3

Difference: .3-.4 / .3-.4

When looking at it this way, it is amazing how much the advantage gap has shrunk.  I was honestly shocked by it.
I tried comparing them based on their measured ISO as well but was just getting confused.  Maybe someone can chime in here.  From my observation, I am seeing that the A99 would lose an extra 1/3 to 1/2 stop in its advantage.  But don't hold me to that.
What I am seeing is if an A77 at ISO 400 is measuring at ISO 318, the closest the A99 can get to matching that is between ISO 500 and ISO 640, which is around 1/2 stop higher ISO.  If the shutter speed is affected more by the measured ISO than the manufacturer's ISO, that means that at manufacturer's ISO 400, the A77 would have a faster shutter speed to begin with.  This would put the A99 at a slightly greater disadvantage.  Am I interpreting that correctly?  Due to that uncertainty, I didn't want to base my observations above on measured ISO quite yet.
So what am I getting at here?
For one, you can't compare the performance capabilities of an APS-C versus Full Frame purely on its noise characteristics at said ISO.  Lenses, focal length equivalent, depth of field, aperture, and shutter speed all play a part in the final outputted image and at what quality.
Saying all that, my comparison still shows that there wouldn't be a disadvantage to going with a Full Frame.  Just in the worst case scenarios when trying to capture the pictures to identically match the look of an APS-C, it should look no worse than APS-C.  But it only goes up from there.  If you can afford a shallower depth of field or can afford a longer shutter speed, then the Full Frame will shine.  And at the extremely low end of the ISO range (ISO 50-100), the Full Frame is even better.  But also, maybe we aren't giving APS-C sensors enough credit?

-- hide signature --


 rf-design's gear list:rf-design's gear list
Panasonic LX100 Canon EOS 40D Sigma SD10 Sigma SD14 Sigma SD15 +57 more
Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow