Originally Posted by Daniel Browning
That's just insane.
You don't have to understand the differences between ff and smaller sensors to take great pictures. But you *should* know the difference before you write an article about it. Sigh.
thanks Daniel, what a great subject to talk and think about it!
I'm resurrecting this thread because it seems like a more appropriate place to reply to HiFiGuy1's post, and I didn't want to start a brand new thread called "My reply to HiFiGuy1's post"
I believe (and not everyone agrees []) is that a big advantage of FF
is that you get better IQ at wide apertures. The reasons (explained below) hold for pretty much any fast lens.
This isn't exactly what I said: I that 1DsIII + 200 @ f/2.8 would outperform 7D + 135 @
f/2, and there are two reasons I think this.
First, looking at Bryan's ISO12233 comparison it seems to me that the
200 has a clear (though not what I would call huge) advantage over the
135 (again, both used wide open). The 135 is great, but few lenses perform as well at f/2 as the 200 does at f/2.8.
When you crop, this difference will
only get larger...
When you crop, you don't just change the angle of view, you throw
away resolution. Of course, you might object that the 7D has more pixel
density, so even with the smaller sensor it has about the same
resolution as the 5DII. True. But I when I say "throw away
resolution", I don't just mean sensor resolution, but lens resolving
power.
Put another way, when you crop, you don't just crop, you crop then
enlarge to the same size you started with. So all lens flaws are
enlarged. I think this may have been more obvious in the film days because when you start with a small film, you have to enlarge more to get to a given size (say 8x10). Even if the 7D has higher resolution "film" to mostly compensate for this, all other flaws in the picture- including those caused by the lens- will be magnified more.
(Heck, there has to be *some* reason people pay huge $$ for medium frame cameras and lenses that have less versatility than their 35mm counterparts)
When lens quality is high enough (and this probably means stopping down), maybe this lens resolving power
isn't so important because lens flaws are smaller than pixels (ie, we are pixel limited) or maybe even smaller than diffraction discs (ie, diffraction limited). In this case, the
camera with more pixels has better sharpness, and 7D and 5DII are about
the same in this department. But I believe the 135 f/2 is significantly
lens limited when wide open, especially at the high pixel density of
the 7D.
Thus I'm less confident that 7D + 135 @ f/2.8 is sharper than 5DII +
200 @ f/4.5, because this is closer to being a pixel-limited situation.
Not only that, but both lenses are in a zone where stopping down has
less effect on IQ, so the advantage the 200 gains by operating at a smaller f number is
reduced.
Still, I don't think the 135 @ f/2.8 is totally pixel limited. I
think the 5DII + 200 @ f/4.5 would indeed have better IQ.
What matters isn't DLA but how many lines per picture you get with a given DOF. You have to stop down more with full frame by a factor of about 1.6 for far away subjects. This factor changes as you get closer to your subject, and furthermore the size of diffraction discs grows faster than linearly as you get close to 1-1 and beyond. On the other hand, even if you have bigger diffraction discs with FF, the sensor is larger, so it might not be clear which gives better resolution...
At least it wasn't to me until I spent about 20 minutes once writing a bunch of equations
want a particular DOF at a particular framing, your diffraction discs
will be exactly proportional to sensor size, and thus you get the same
resolution no matter what your sensor size is. (Sensor resolution is a different issue... but what this says is that diffraction is exactly equally important for two 18mp cameras taking pictures framed the same way and with the same DOF, no matter what the sensor size, assuming all else equal- which it never is.)
I don't think there is any question that the 17-40 would be happier
on FF. When you buy a 17-40 you sacrifice IQ, cost, and weight to get a
lens that has a wide angle. When you then put it on a 40D, you then
throw most of that wide image away (together with lens resolving power). If I wanted wide angle on a 40D, I'd get an EF-S lens.
Of course
the 17-40 gets wider on APS-H than APS-C, but then you give up the
ability to use EF-S lenses. Thus, IMO APS-H is the worst size for wide
angle. (Though it is great for a lot of other stuff)
Makes me sad to say it, but I probably ought to sell my beloved 17-40. []I am not doing it justice on my camera. I will miss it, though. It was the first lens I ever bought, and I didn't even have a body yet! Oh well, anybody want an EF 17-40 f/4 L? I guess I need to forget FF and just get a 17-55 f/2.8. [
Wow. Back then, you *really* weren't using its full potential
No, but I was so smitten with the idea of DSLR photography that I wanted to jump in ASAP. That was my first opportunity so I took it. I bought the 17-40 based on Bryan's comments on it being an excellent value in a wide angle L zoom. I also thought I'd be getting a 5D, which I actually did, in a kit with the 24-105, but it didn't have the speed and responsiveness I wanted, so I returned it after a couple of weeks.
I came across a great deal on a NIB 40D kit with the 28-135 that I couldn't pass up, so I got that next. Then came the understanding that my wide angle and my new body weren't necessarily made for each other, and ever since I guess I have been trying to will them to be happy with together. Alas, it was a union that was not meant to be.
This evening on the road on the way to work, I have been pondering alternatives, and I think that if I sell my 17-40 and my 28-135, I will have enough money to get an EF-S 15-85, which seems to be highly regarded. I know it isn't constant f/number across the zoom range, but it certainly seems like an excellent choice from almost every other perspective, and it will work ideally with my 40D, and of course any future APS-C body I might end up with.
To get back to our other discussion, though, I think we are bogging down with the wide open comparison on the two actual lenses that I chose originally, when the point of the discussion is at least somewhat theoretical. That is why I shifted to the 5DII + 200 at 4.5 and 7D + 135 @ 2.8. That allows the lenses to be nearly interchangeable optically and DOF, and then we are only comparing the performance of the combos from a resulting image standpoint, with the consideration for an 8% crop of the FF to match the framing of the other combo. This even makes the resulting resolution similar.
The rationale, though my train of thought may have been derailed by the long drive, is that for a lot less money, I can get FF results with an APS-C and a carefully chosen companion lens. I know there are certain things that will only be possible with FF, because it is mathematically impossible to achieve them with a crop body, but for the vast majority of my shooting, with properly chosen lenses, I won't miss the FF experience at all.
I had also been questioning the efficacy of APS-H, as it relates to the compromise with both APS-C and FF, which I still am interested in hypothetically, but I decided on the way here that I am having a crisis of conscience financially, and can't make myself spend the money right now on a 1D3 or 1D4. I could talk myself into it in a heartbeat, but I shouldn't do that for the moment, and I am trying to be responsible.
My aspirations for now are to be able to shoot residential (indoor) architecture, portraits, macro and birds/air shows. I feel like I will be able to do this at least adequately with two lenses for now, which are the 15-85 and either a 300/4 or 400/5.6, though I am leaning towards the 300 because it will also double as a near-macro like the 400mm Sigma TeleMacro I just sold (Nikon mount). The 15-85 should be wide and long enough to get the architecture and portraits. I am also looking at the 17-55, but honestly it doesn't seem like it is better in practice than the 15-85, and it lacks width and length, and has a less sophisticated IS, a full stop short. It is almost too good to be true that the 15-85 is $400+ cheaper. It lacks the 2.8 constant, but especially on the wide end, I would be stopped down anyway, probably. If I really needed the narrow DOF on the longer end, I could just get an 85/1.8 and stop it down to 2.2 or 2.8. The combo would be about what the 17-55 costs alone. What do you think about that?
No, you're not. Oh boy. Let me start by saying that I don't want to re-launch my discussion with Jon, but at the same time, we touched on several of these issues recently. My main point from that discussion was that you really cannot compare bodies and lenses at the same time like that. If you read Bryan's comments on how to use the ISO 12233 charts, he makes a clear point that lenses only should be compared with other lenses on the same body, and bodies should be compared with other bodies only when using the same lens (for that, he recommends the 200mm f/2L and the 200mm f/2.8L II, since he's shot the charts with several bodies using those lenses, presumably intentionally to allow comparison across bodies). The only way to actually compare those two combinations is to shoot the same test shots with those specific combinations, i.e a 5DII body, a 7D body, and the pair of different primes (or a zoom lens set to different focal lengths and apertures) to compensate for the sensor crop factor.
But, HiFiGuy1, let's take a few of your statements and look at them:
No, this isn't really true. With an APS-C body, you can choose a lens to match the aperture and focal length that you would achieve on a FF body with a different lens (e.g. 85mm f/1.2 on crop body, 135mm f/2 on FF; or, 50mm f/1.2 on crop body, 85mm f/1.8 on FF). The focal lengths and max apertures will be equivalent. But the results will not be the same. First and foremost, ISO noise will be higher on the crop body - that alone will make a big difference in the types of things you want to shoot. One of Jon's valid points from our recent discussion is that cost is also a consideration - in the examples above, getting the 135mm f/2 and85mm f/1.8 to use on FF would cost about ~$2K less than getting the85mm f/1.2 and50mm f/1.2 to use on a crop body (that difference would almost buy the 5DII!!).
You're right thatcertain things that will only be possible with FF - the razor-thin DoF of an f/1.2 lens on FF is one example.
I don't think it's a good compromise, either. Personally, the combination of a good crop body (e.g. 7D) with a 5DII would make a much better option than a 1DIII or 1DIV body, for about the same cost.
Wide enough for architecture, probably. But the other issue there is distortion. Buildings have straight lines. At the wide end, the 15-85mm suffers badly from barrel distortion (3.15%). By comparison, the 17-55mm at 17mm has 2% barrel distortion, and the 10-22mm at 10mm is quite good, with only 1.25% barrel distortion. Also, some interior spaces really need wider - that's where the EF-S 10-22mm f/3.5-4.5 is useful; the minimal amount of distortion with that lens make it one of the better choices for architecture (short of a tilt-shift lens).
On the long end, I think you'd find that the 15-85mm for portraits will, in a word, suck. I should qualify that by saying that it would work in a studio setting, where you have total control over lighting and especially background (pull-down backdrops). But if you can't simply lower a contrived featureless background into place, the OOF blur and subject isolation you want in a portrait aren't going to happen at 60-85mm f/5.6.
So, on the whole I think the 15-85mm is probably not a great lens for your uses - too much distortion for architecture at the wide end, too narrow an aperture for portraits at the long end. It's a compromise lens, but with too many compromises. I think for a point-and-shoot user coming to dSLRs, it would be a very nice lens.
That makes sense. The 85mm f/1.8 is a great close-up and head/shoulders portrait lens on a 1.6x body.
You might actually consider 10-22mm, 85mm f/1.8, and the 100-400mm instead (I find the 100-400mm to be much more versatile than the 300mm prime; you mention air shows, Bryan calls the 100-400mm the best air show lens).
Just my 2¢. Good luck with your decisions!
--John
neuronatomist,
"...but I decided on the way here that I am having a crisis of conscience financially, ... and I am trying to be responsible."
When you suggest the 10-22 + 85/1.8 + 100-400, I think you forgot about the above. [] I don't want to go into worse (or inappropriate, in the case of the 17-40) glass from my current 17-40 and 28-135, but I can't really spend a bunch more money. Also, the trio you suggested leaves a hole in the general purpose area. I do love the idea of a 17-55, but I am not sure it has enough range, and it is almost twice the current price of a 15-85. Is that lens THAT bad? Unless I am misreading the review, Bryan seems to quite like it.
If I had the cash, I would probably like to have the 10-22, 17-55, 85/1.8, 100/2.8 IS macro and a telephoto plus a 1.4x. I won't say I wouldn't own one, but I have used the 100-400 and really don't dig the push-pull zoom. If they made that range in a normal zoom lens, I would be all over it. I have no real problem with the IQ, just the functionality. I might get used to it, but I haven't had that opportunity yet.
It's a very nice lens, when used as a general purpose zoom (especially outdoors). But although architecture and portrait shots may fall into the focal length range covered by a general purpose zoom lens, those types of shooting are specific, not general. Although the 15-85mm is a fine lens, personally, I don't think it's well suited to architecture and portrait shooting. Vacation pictures, pictures of a family outing or gathering (even indoors with an external Speedlite), etc. - for that sort of thing, it's a good option.
I think the reasons I gave apply to any lens... it's just that the difference matters less for lenses that are more pixel limited.
I don't think there are
mathematical or physical limitations on small sensors, at least, not
ones that are relevant to our discussion.
If you set things up so the shutter speed is the same, the noise will *not* in general be higher on the crop body. Photon noise will be the same, assuming the sensors have the same quantum efficiency (I believe the 7D has a slight edge over the 5DII here). Read noise could be higher or lower, who knows. The 7D has lower read noise overall, but there may be particular situations in which the 5DII has lower read noise.
Results will indeed, modulo specific attributes of the particular lens and body chosen (like IQ [
Posting Permissions
Reply With Quote