Light Loss at Wide Apertures

[MarkBarbieri]

Here is an interesting article on light loss at wide apertures. As I understand it, DSLRs auto-adjust the ISO when you use very wide apertures to correct for the fact that the sensor doesn't really get all the extra light. Each sensel is at the end of a little tube. When you use a really wide aperture, the light coming from the edges doesn't make it to the bottom of the tube. The consequence is that you aren't gaining as much light as you would expect from having such a wide aperture. The camera makers have the camera adjust the ISO to compensate, although the claimed ISO value doesn't change.

As an example, the Canon 7D misses about 1 stop of light when shooting at f/1.2. I am inferring from another graph that Canon boosts the ISO by roughly 1 stop over what it would use if you stopped the lens down one stop to f/2. That implies that you won't really see a significant change in noise or DOF between the two aperture settings assuming that you adjusted the ISO instead of the aperture. The noise will be the same because if you shot at f/2 and ISO 1600 or f/1.2 and ISO 800, the actual gain applied by the camera would be the same. The DOF would be the same because the additional light rays coming in from a more oblique angle with the wider aperture will not reach the sensor, so they won't change the DOF.

Maybe I'll do some experiments to see if I get similar results.

Looking at the chart, there appears to be a fairly strong relationship between pixel pitch and the light loss effect. That seems logical because the "tubes" over the sensels would be narrower and block more light. It is also interesting to note that it looks like for a given pixel pitch, the effect is reduced as the sensor technology gets newer. Presumably this is because they improve high ISO performance partly by making the "tubes" shallower. It looks like the T2i actually outperforms the 7D by just over 1/10th of a stop.

The author concludes that people might make different lens choices if they had this information. For example, there is very little value in moving from the 85mm f/1.8 to the 85mm f/1.2 if you are a 7D shooter.

Or maybe I misunderstood the whole thing.

 

[photo_chick]

Interesting.... and now you've given me something new to learn about.

So if I understand this... they adjust the signal gain so that the ISO measurement at the wider apertures is correct, which means that wider aperture shots could be subject to more noise.

One more reason to love my entry level glass! LOL

 

[boBQuincy]

This news is interesting but finding that some "special" processing is going on inside our cameras is not really surprising or new. Some testing has shown that Canon's intermediate ISO settings really use the next higher base ISO, so 125 has the same noise level as 200. Extra (and usually unwanted) processing of RAW images also goes on and we do not know to what extent.

I guess the manufacturers decided the ISO bump was a better choice than having customers wonder why going from f/2 to f/1.2 showed no increase in shutter speed. Perhaps this effect is why we have not seen any EF-S lenses with large apertures, they just don't work on a crop sensor (at least not yet).

Or is this just the next ploy to get us to buy full frame cameras?

 

[MarkBarbieri]

I agree that the insinuation that camera makers up to secret tricks is rather overblown in the article. Assuming that the effect is true, it would be a nice spec, though, for the above mentioned reason that each sensor effectively has a minimum maximum aperture and going beyond that is essentially a waste of time.

I don't see it as a drive to full frame as much as it implies that there is a cost to higher pixel densities in the form of an aperture limit. Improvements in pixel "tube" depths need to keep pace with resolution increases.

 

[photo_chick]

Many of us have known for a while that higher megapixels aren't all they're cracked up to be. Unfortunately the mass market consumers still don't get that.

 

[boBQuincy]

DXOMark has not appeared to test the Foveon sensor used by Sigma but the results could prove interesting. Since the sensor has R, G, and B pixels and they are at different depths there may be significant differences in ISO *by color* with a large aperture lens!

I agree that it is probably not so much that the camera manufacturers are trying to put one over on us as it is they are hoping to prevent a lot of confusion over effective apertures with lenses that the average consumer would not buy anyway. Still, the few who do buy such lenses are entitled to know if they are not getting all that they hoped for (and paid a lot for).

I was considering a 50mm f/1.2 for the shallow depth of field, quick focusing, and nice bokeh. Now I am not sure it would work well on my crop sensor camera, or at least not provide much more maximum aperture than my $100 f/1.8!

 

[Code]

I'm trying to wrap my brain around the argument, but there's something I don't understand. The last graph shows the sensor gain for 3 different cameras, one being the Sony A350 which shows the most gain, but the Sony A350 has a CCD sensor, not CMOS.

 

[Geoff_M]

Part of me is wondering if the author is confusing the theorethical with the practical. There's a point where pixel-peeping doesn't translate well into real world results... and I'm thinking he's crossed it.

Bottom line: Due to the complexity of design and manufacture (let alone the high cost and weight) of large aperture lenses, one may actually end up with better results at virtually the same ISO and depth of field using lenses with more modest maximum apertures.

Somehow, I'm not quite ready to buy that notion... but it does sound like an idea that Ken Rockwell would give two "thumbs up" on!

 

[photo_chick]

... but it does sound like an idea that Ken Rockwell would give two "thumbs up" on!

 

[MarkBarbieri]

I did a small experiment using an 85mm f/1.2, a Canon 7D, and a lens align. I haven't focus adjusted the lens yet and it appears to be a little front focused. From what I can see in the pictures (these are 100% crops), it looks like f/1.2 and f/1.4 look virtually identical in terms of DOF. With f/1.6, I can start to see a difference and DOF increases noticeably from there.

You can see the pictures here. Obviously, you need to view them at the original size.

So what is the moral of the story? I don't know. It definitely shows that there is a difference (however small) between the DOF at f/1.8 and f/1.4, so anyone trying to decide between 85mm lenses with those apertures can be confident that even with an APS-C sensor, you are getting something for your money. If you are trying to decide between an f/1.4 and an f/1.2 on an APS-C, well, I can't see a difference in my very simple test.

It also shows that shooting a black and white ruler angled closely to the camera at a relatively short focusing distance is a recipe for chromatic aberration. That's the red and green fringing you see. The rule is plain black and white. It isn't a special Christmas ruler. I've never seen any CA with this lens in real world shooting, so don't let that get you worked up.

Maybe I'll run the same test on the 5D later. I did take one quick frame with it at f/1.2 and it was noticeably shallower than the 7D, but I would expect that even at slightly narrower apertures anyway.

 

[Experiment_626]

How much difference would we expect to see in a situation like that? Both f/1.2 and 1.4 have very limited DOF in any situation, and focussed very close like that, I'd imagine the difference would be extremely small.

 

[AlbertZeroK]

I will have to look at this next time i shoot. I often shoot at 1600 iso a stop added to that would easily be shown. But i can not believe.canon would be able to get away with something like this for this long.

 

[photo_chick]

I wouldn't expect to see a huge difference in depth of field between f/1.4 and f/1.2. Or even for the lens to be much faster. The math supports an almost negligible difference in both and while you do get something, it's not a big jump or anything. But then again it's not just about a wider aperture. You also get a better build quality, sharper optics and a better performing lens all around in most cases.