Pixel density for bird photography

[Tannin]

In another thread, which I didn't want to drag off-topic .....

Tony wrote:
the D3 & D700 don't have enough reach for bird work

Please explain!!

For bird photography, you need pixel density. Simple as that. The D3 and D700 are very fine cameras, but they have the lowest pixel density of any SLRs sold today, and are generally avoided by people primarily doing bird work. Nearly all bird photographers use cameras that have high-density sensors.

[Tannin]

The vast majority of bird photographs you will see here (or in most other places) have been cropped, quite often heavily cropped, because it is a very rare circumstance for the photographer to find himself as close to the subject as he would really like to be. The more paitent you are, and the more skilled you are - and indeed the luckier you are on the day - the closer you can get, but you just about always would rather be closer still.

Of course, you have already got the longest lens you can afford to buy and/or afford to carry with you for a given task. (Note that as the lens you carry becomes physically larger, heavier, and more cumbersome, you become correspondingly less able to move (or sometimes even sit) inconspicuously enough to achieve a close approach to your subject.)

In the end, no matter how long a lens you carry, most of the time you are going to wind up cropping the shot - i.e., throwing away megapixels worth of effectively useless information and keeping only that part which shows the subject. Obviously, the more pixels you have of actual subject matter, the better the result (assuming only that the pixels in question are of decent quality - recent model P&S cameras need not apply).

Perhaps the clearest way to illustrate this is to work some numbers. I'll take the Canon bodies as my starting point, but the point is equally valid for all other brands.

(Actually, "other brands" pretty much means Nikon - none of the others are used much if at all by serious bird photographers. Even Nikons are rare in the bird photography world. There are several reasons for this, but probably the main four are that (a) it's only been in the last couple of years that Nikon have started manufacturing a reasonably full range of modern, image-stabilised lenses suitable for birding; (b) until about the time the D300 arrived, Nikon sensors were regarded as having inferior high ISO capability (high ISO ability is usually less important than pixel density, but important just the same); (c) most of the cost in bird photography is lenses, and the big Nikon lenses usually sell for a couple of thousand dollars more than the directly equivalent Canon models - I suspect that this puts a lot of people off; and (d) Nikon still don't have a high-density, full-frame professional standard body (we expect this to change any time now, however).)

Assume you have what is probably the most popular birding camera around at present, a 40D. (Or any other 10MP camera/lens combination with the same field of view - that won't change the numbers.) Most people would be fairly pleased to get a shot out of the camera that looks like the one immediately below. We haven't filled the frame, but we have a fair bit of bird in the shot and at least it isn't a tiny dot somewhere in the distance.



(10.1MP: as shot.)

Obviously, we are going to want to crop it a bit. Let's start by cutting out everything except the bird itself. (Plus some leftover background, because the bird isn't exactly rectangular.)



(2.2MP: just the bird.)

Ouch! We have a 10MP camera and we have less than 2MP worth of actual subject in the frame! Unfortunately, this is the general rule - getting much better than this ratio is the exception, even for skilled photographers with good equipment. Just the same, you wouldn't use so severe a crop in real life, so let's do one that is representative of what most bird photographers would actually wind up using:



(4.0MP: as printed.)

[Tannin]

Now, let's apply the general rules we have discovered to some different cameras. I'm going to take the "as-printed" crop above - where we wound up with 4.0 usable megapixels from our typical 10.1MP 1.6 crop camera - and see what would happen if we used a variety of alternative cameras.

1.9MP - D3, D700 (1.0 crop, 12.1MP)
2.0MP - 5D (1.0 crop, 12.7MP)
2.5MP - 10D (1.6 crop, 6.3MP)
2.6MP - 1D III (1.3 crop, 10.1MP)
2.6MP - 1Ds II (1.0 crop, 16.6MP)
3.3MP - 20D, 30D (1.6 crop, 8.2MP)
3.3MP - 1Ds III, 5D II (1.0 crop, 21.0MP)
4.0MP - 40D, 400D (1.6 crop, 10.1MP)
4.4MP - D300 (1.5 crop, 12.2MP)
4.9MP - 450D (1.6 crop, 12.2MP)
6.0MP - 50D (1.6 crop, 15.1MP)

Remember, this is taking the exact same shot from the exact same place with the exact same lens. We can't use a longer lens - if that was practicable, we would already have done it. And we can't get any closer - we would have done that too if we could have managed it.

So what we get is what we get. A D3 (or a 5D) give you less than half as many pixels worth of actual picture.

[ricktas]

Very well written Tony, and should be of benefit to all those would be birders out there.

I won't mention the benefit of using an APS-C sensored camera over a full frame, due to the automatic crop factor/perceived distance benefit. Oops, I mentioned it.

[Colinz]

Thanks Tony- very helpful. DP review has added mp/cm2 to the camera specs and the claim has been that the lower this figure is the better the image. So the D3 has 1.4 and the 50D has 4.5 which relates inversely to the cropped image figures you have posted.

What I have found interesting with my FZ28 is that at what is called extended optical zoom set a 3mp ie cropping the sensor, the image is noticeably sharper than the same image at 10mp cropped to the same size. So most of my shots are taken using the cropped sensor.

[arthurking83]

Also, for those that don't realise it but that bird was captured at 700mm!(that'll be 500mm plus 1.4xTC)

So the importance of Tony's post about how important pixel density(resolution) is to birders is even more important, considering that the majority of average users wil have easy access to a lens of say 300mm, and may want to try their hand with a cheap-ish(~$1-1.5K) xxx-400 or xxx-500mm lens.

Chances are that you will crop quite some and most likely more than Tony has done here!.. so keep it in mind folks

Great post Tony.

[Tannin]

Now set the Nikon models aside for the moment (we will keep things simple by just considering one brand at a time) and let's look at what cameras bird photographers actually use most often. Overwhelmingly, the answer is one of the higher density 1.6 crop models: a 40D, a 400D, or something similar. There are two reasons for this. Obviously, not everyone can afford more than a 50D or a 450D. But we can quite easily control for that confounding variable by only considering photographers that also own one of the big whites - we can pretty safely assume that most people keen enough to spend $10,000-odd on a lens are also keen enough to spend a good deal more than $2000 on a camera if that's what they want.

And what do we find amongst bird photograhers who own big whites? They fall into three main groups. I wouldn't like to sound too certain of the exact break-up between the three, but all three are well-represented, and at a guess I'd rank them in the order below:

Very high-density 1.6 crop bodies (40D, 50D, 400D, and 450D, though still a few 30D, 20D and 350D users around as well. 40D is the most common, 50D is catching on fast.)
Medium-low-density 1.3 crop bodies. (1D III, still a few 1D IIs around as well.)
High-density 1.0 crop bodies, (1Ds III, 1Ds II).

What happened to the 5D? Where is it listed? It isn't listed - because practically no-one uses a 5D for bird work.

Why would you want to use a 5D, given that a typical half-decent shot opportunity is only going to give you 2MP worth of bird picture? A D3 or a D700 would give you a pro-build body, faster frame rate, and better high ISO performance (compared to a 5D) but even worse finished-product resolution, at just 1.9MP. All three 12MP full frame cameras, none of them cheap, give you fewer pixels on the bird than an ancient 10D would. All three are very fine cameras, but none are suitable for bird work, except under exceptional circumstances - such as if your name is "Andrew" and your target species is the Laughing Kookaburra.

So why, if the crop cameras like the 40D and 50D give you so much more finished-product resolution, do many dedicated bird photographers not use them? Why would you spend all that extra for a 1Ds or a 1D knowing that you are going to have lower finished-product resolution?

Many, perhaps most, of the really, really good bird photographers use a 1Ds because they judge that they are good enough at it to get really, really close and fill that much wider frame up. It's significantly harder work, but if you succeed at it, the results speak for themselves. Also, notice that the latest generation of full frame cameras now have pixel density just as high as the old 20D and 30D had - so that's getting pretty respectable. And, of course, these are cameras with top-class autofocus systems, superb viewfinders, and just about every other possible useful feature - once you use a 1 Series body, everything else seems pretty clunky.

The 1D III is another popular choice amongst this same group of photographers (ones who own big white glass and can thus be presumed able to use whatever camera they want). (There are still some 1D IIs around as well.) On the face of things, this one seems harder to explain - after all, you take a big hit in resolution compared with something in the 40D class. You don't get as much resolution so you arguably need to be even closer than you do with a 1Ds, but you still get all the pro body features like viewfinder and top-class AF system, plus two other things: a very fast shutter repeat rate, and very good high-ISO performance. It's a trade-off. Some people think it's worth it, some don't. (In theory you could take this approach even further and have essentially the same sort of camera but with even better high ISO at the cost of even worse pixel density - that's what the D3 amounts to. Most bird photographers apparently regard this as one bridge too far, and I agree: of all the currently available cameras from Canon and Nikon, I'd put the D3 second last on my wish list, superior for bird work only to the 5D, which is almost equally low resolution but much slower and lacking things like the pro AF system too.)

And, of course, a good many top-class bird photographers with an effectively unlimited choice of gear nevertheless use 1.6 crop cameras, mostly 40Ds and now 50Ds. They would rather work within the limitations of a non-pro body, with non-pro auto-focus and less than stellar feature set, but not have to crop so hard. Yes, the larger cameras deliver better quality pixels, but there is still much more detail available from a real-world crop of a focal-length-limited (e.g.) 50D image than there is from one taken by a lower-density body.

Personally, I think there is merit in all three approaches. I really like the wonderful picture quality delivered by the 1D III (there is a lot to be said for fat pixels), the clarity and rich, subtle colours, the focus speed and accuracy, the ability to push the ISO any time I need to, the big, clear viewfinder, and the all-round handling qualities. In exchange for that, I sacrifice a lot of reach - with the cumbersome 3.9kg 500mm f/4 lens, the 1D III delivers less detail than a 50D does with a little 1.4kg 100-400/5.6. Or, to put it another way, to get an equally detailed picture I have to get just as close to the bird with the 1D III and 500/4 as I do with the 40D and the 100-400, and that can be quite a lot harder. But, most of the time, I think it's worth it. Apart from the focus system, the frame rate, and the richer resulting image, I've got an extra stop of aperture up my sleeve if needed. On balance, this is my preffered strategy.

But there is a lot to be said for using something like a 50D instead, and quite often I do. (Especially since I replaced the rather lack-lustre 40D, which to my mind never really improved much over the old 20D, and certainly didn't deliver any better image quality, with a 50D.) In general, I think it's usually better to use 50D and 1.4 converter than 1D III and 2X converter, and quite often better to use 50D bare lens than 1D III and 1.4 converter. The reach is about the same either way, but the focus is faster (perhaps not quite so accurate) and you have an extra stop of aperture available.

As for the full-frame, high-resolution method, I can't say from practical experience. One day I'll buy a 1Ds III and see for myself ....... except that by the time I can afford one they will be up to the 1Ds LXVII.

[JM Tran]

So where does the new A900 fit in? haha

I have used the A900 at the launch with some models provided to test the extreme cropping, and boy the crop results rival that of digital MF backs

wasnt very impressed with the high ISO capability though, for obvious reasons

[dbax]

Thanks Tony, and others, facinating read, great info and plenty of food for thought

[arthurking83]

..... DP review has added mp/cm2 to the camera specs and the claim has been that the lower this figure is the better the image. So the D3 has 1.4 and the 50D has 4.5 which relates inversely to the cropped image figures you have posted.

What I have found interesting with my FZ28 is that at what is called extended optical zoom set a 3mp ie cropping the sensor, the image is noticeably sharper than the same image at 10mp cropped to the same size. So most of my shots are taken using the cropped sensor.

Exactly!!(but not exactly too!)

Define 'better image' ?

Some want resolution, other want zero noise, and as alluded too in the other thread there are different feathers for different birds(or something like that)

Better images is usually taken to mean better quality of pixel data, and therefore light.

A P&S at 12Mp is no match for a D3, in terms of sheer quality of image when printed large.

If you never print(as I have never seem to have done, but will one day!) this is moot.
You'd never really worry about it.

But if you want the image printed(reasonably large, like an A4 or bigger) you wil see a small improvement in image quality of the D3 over the P&S. As the print gets bigger you see it more vividly.

You can also do a whole lot more(editing) with those larger pixels!!
Here's the point of having those bigger pixels, even though there's less of them!

Each pixel can be stretched to a higher point of processing brightness/darkness/sharpening/color saturation before the quality of that piece of data starts to deteriorate.
Remember each pixel is just a teeny amount of data. But that data had to be captured and converted from a light source(analogue). The better that source of analogue info, the better the ability to process it to the nth degree.

of course technology also increases vastly over time, so you 2Mp camera from years gone by, is still not going to produce as good a quality as a current higher resolution camera can!

And that brings us to another point often brought up(and I think we're going off topic now). As technology improves there's a tendency to want to upgrade to a better camera. Lenses stay static, and take many more years to improve upon, compared to a camera body. Get a great lens that allows you to capture great images from the outset, and it seems to improve the technological durability of your camera body too!
Remember if the quality of the incoming data is good(with good lenses) then the quality of the data you output(regardless of the amount of processing your do) is going to be better too.

Sorry to go OT with that, but it is an important aspect to consider when the topic revolves around resolution.

Hope that makes sense Colin.

It really looks as though the APS-C sensor seems to have reached a saturation point.
DPR's summary of the 50D was something along the lines that it can have worse quality compared to the 40D. Does that imply that 12 or 13 or 14Mp is the realistic limit for an APS-C sensor.
FF still has a long way to go in terms of pixel density, and the issue that's going to play an important role in that arena is one of storage and movement of the impending vast amounts of data from the coming of the 50Mp DSLR!!

[Tannin]

Arthur, you are too quick for me! As I was about to post .....

Actually my example image was taken with a 1D III at 700mm - but that really doesn't matter from the point of view of using it as an example. It's a 10MP image that is a pretty fair representative sample of the sort of 10MP image you could sensibly hope to get with something like a 40D and a 100-400. (If I'd realised how much time I was going to put into this "short" post before I started, I'd have picked an actual 40D & 100-400 shot, instead of just grabbing the first one I saw in the folder I happened to already have open.)

But actually, if you are used to thinking in terms of apparent focal length - what you see when you look through the viewfinder of a 40D or a D300 - we shouldn't think of it as a "700mm" shot. Being taken with the 1D III, it fills the frame in the same way that, from the same spot, you'd fill a 560mm frame on a crop camera. Or, putting it another way, you get almost exactly the same field of view from my 500mm lens on the 1D III as you get with a 400mm lens on a 40D.
happend

By the way, below is how I eventually chose to crop that image for my website. Mostly I prefer a more generous crop than this, but I wanted to get rid of those branches and I didn't have a better view of a White-winged Triller - I always find them a bit challenging.



Looking at it again, I think it needs better PP: it's a bit over-bright and could be sharpened a bit more. There is another job to put on my list.

[Tannin]

So where does the new A900 fit in?

Until such time as Sony produce some serious long glass for it, nowhere, I'm afraid.

I doubt they will: it's a seriously expensive undertaking developing that kind of glass and you'd need to sell a good number of them to break even on the investment. Canon sell lots of long glass, Nikon haven't in the past (200-400 VR apart) but have a big customer base that doesn't seem to mind spending buckets of money, so they should do OK. The same goes for Pentax: no current range of modern long glass, and arguably Olympus too. Oly apparently have a very good if rather expensive 300mm unit, but even on a 2.0 crop body that seems a bit too short for birding.

A second factor keeping Pentax, Sony and Olympus out of the wildlife and sport specialties is their in-camera image stabilisation system. For shorter lenses, apparently, in-camera systems work pretty well, and stabilise all your lenses, but the longer the lens (the conventional wisdom goes) the more of an advantage it is to have your IS system in the lens rather than in the camera. I can't really see a Sony or a Pentax developing an in-lens stabilising system just for 400mm and 600mm lenses, and if they introduced (say) a modern 600mm f/4 lens with the latest digital coatings and so on - this would cost around $12,000 retail - who would buy it when Nikon and Canon have direct equivalents with IS/VR?

[Colinz]

So to summarise- get close enough to fill the frame and the lower mp/cm2 produce the best image from any standpoint. Can't get close and the higher mp/cm2 allow for a better crop. There being a limit past which noise becomes the main problem.

I keep harking back to the days (of film) when I learned photography- restricted to iso 25 then 64 so you had to get close and have plenty of light. People of the digital era don't know they're alive! It was so difficult to get a good bird shot that didn't look like it was taken at midnight. I might see if I can pull a couple from my slide archives and post them for interest sake.

PS thanks for the input Tony

[JM Tran]

Until such time as Sony produce some serious long glass for it, nowhere, I'm afraid.

I doubt they will: it's a seriously expensive undertaking developing that kind of glass and you'd need to sell a good number of them to break even on the investment. Canon sell lots of long glass, Nikon haven't in the past (200-400 VR apart) but have a big customer base that doesn't seem to mind spending buckets of money, so they should do OK. The same goes for Pentax: no current range of modern long glass, and arguably Olympus too. Oly apparently have a very good if rather expensive 300mm unit, but even on a 2.0 crop body that seems a bit too short for birding.

A second factor keeping Pentax, Sony and Olympus out of the wildlife and sport specialties is their in-camera image stabilisation system. For shorter lenses, apparently, in-camera systems work pretty well, and stabilise all your lenses, but the longer the lens (the conventional wisdom goes) the more of an advantage it is to have your IS system in the lens rather than in the camera. I can't really see a Sony or a Pentax developing an in-lens stabilising system just for 400mm and 600mm lenses, and if they introduced (say) a modern 600mm f/4 lens with the latest digital coatings and so on - this would cost around $12,000 retail - who would buy it when Nikon and Canon have direct equivalents with IS/VR?

well as a full time pro I turn towards Sigma for my long end needs when i was still using Pentax, Sigma probably has the most diverse range of telephotos available at a more affordable pricing than C & N offerings.

So i can imagine in the near future many A900 owners turning towards the Sigma line up, instead of waiting for some rediculously overpriced Sony or Zeiss telephotos to be produced.

true that at the telephoto end SR in my pentaxes never fared as well as the IS in my canon lenses. BUT, in-body still gives me up to 2stops advantage with telephoto - so it works fine, if slightly less effective than IS in-lens for telephoto

[Tannin]

It really looks as though the APS-C sensor seems to have reached a saturation point.
DPR's summary of the 50D was something along the lines that it can have worse quality compared to the 40D. Does that imply that 12 or 13 or 14Mp is the realistic limit for an APS-C sensor.

DPR's "review" of the 50D has been (quite correctly) howled down from right, left and centre. DPR is just plain wrong. See for yourself here: http://www.ausphotography.net.au/for...ad.php?t=18890

The funny thing is, I agreed with your take on this until just after the 50D came out. I was certainly not impressed by the "improvement" that the 400D and then the 40D brought to the table in the wake of the (admittedly excellent) 20D/30D twins. On the whole, I generally prefer the images my 20D makes to those of the 40D, though there isn't a lot in it. (Why then did I buy two 40Ds? Because the image quality was about the same as the 20D and I wanted some of the other features it had, notably sensor cleaning. In any case, I'd promised to give my 400D to a family member and I had to replace it with something.)

But the 50D has changed my mind. It delivers significantly more detail than the 40D (or the 20D), so I was wrong. How much further crop cameras can go than the current high water mark of 15MP remains to be seen, but I thought they had hit the wall at 8MP, so who knows? I'll bet you a bottle of good red medicine that the D400 is 16.something MP, or maybe a little more than that, and produces beautiful images.

[I @ M]

Now set the Nikon models aside for the moment ------------ such as if your name is "Andrew" and your target species is the Laughing Kookaburra.

Pfffft,

Real men use Nikons ----
Real men don't crop ----
Real men only need 10mm lenses for birds ----



Sorry Tony, but you did bring it up.

Thanks very much for a very clear and concise post on the subject Tony.

and as an afterthought for anyone thinking about the sort of lens that Tony is using to capture the exquisite images he does ------

Tony in background and Princess Col looking "down the barrel" ( image courtesy of Debra Faulkner)

[Tannin]

^^^ Well, I know Col is looking for the little birdie that makes the "clack" sound when I press the big black button - but what am I looking so worried about?

[arthurking83]

So where does the new A900 fit in? haha

I have used the A900 at the launch with some models provided to test the extreme cropping, and boy the crop results rival that of digital MF backs

wasnt very impressed with the high ISO capability though, for obvious reasons

A900 would make a brilliant studio or landscape camera for sure!... but for high speed action, or for birding, it's probably lacking in a lot of areas(feature wise)

Usability at low ISO is definitely it's forte.

Apparently Nikon is supposedly using this same sensor for the D3x(maybe called D4, but highly unlikely!)
There seems to be a huge delay in it's release, and I stil reckon it's due to it's (inherently)higher noise quality.
Word is that Nikon have given a few D3x's to certain photogs in the NPS system to test.
(lots of rumours abound on the net) and considering the way the D3 was introduced, it's highly probable.. but this thread is about hi-res cameras for birding.
Tony points out the features required to maintain that kind of photography, and more to the point, the availability of lenses for that purpose.

I'm glad it was introduced into the equation so as to highlight that it's probably not a camera fit for the purpose of birding due to it's features list(frame rate/focus system/speed/etc) and the lack of birding lenses available for it.
Zeiss do make a fabulous sounding lens in the 135mm f/1.8 though, and I think that points to it's intended market!!!(Portaiture!)



Tony, my point about the lens used(700mm) was simply to point out to the folks that have aspiration for a telezoom, and find themselves looking at (up to) 300mm types. Even on a crop body, it's not realistically enough for great opportunities, as as you said 400mm seems to be the minimum length.
Your use of the 1.4TC highlights something I kind of suspected too, in that you can 'easily' substitute a TC of capable performance(on a equally capable lens of course!) to make up for a lack of pixels.
One thing I've noticed(and Colinz alluded too with his findings on his Panasonic) is that you can sharpen a big pixel more so than you can a smaller pixel(working with available data, if that data is good enough to work with).
One thing that the D300 doesn't tolerate too much is sharpening.. no where near as much as the D70s can, nor the D3(even better!)
Just after the D3 was announced(but not introduced) a while back there were D3 sample images available for download, and the quality of the images were a pleasure to work with (in NX).

So!!

Would you say that having the ability to compensate for the loss of one stop(in using a TC), can be made up for by a better quality pixel(ISO)?

One thing I have noticed is many images of the Nikon 200/2 against the Nikon 300/2.8
(remember I'm Nikon, and research Nikon stuff!)
Both lenses are equally fine lenses, and there is either nothing between them both in terms of IQ, and equally so if you add a 1.4xTC to the 200/2!!(makes for a 280/2.8, which is damned close)

My reply is to ascertain if it's feasible to have your cake and eat it too.
ie. A one camera fits all purposes situation.
A D700 is my ideal camera for what I prefer to do.
A higher Mp sensor will(may??) introduce unwanted noise, and lower DR.
Could I plausibly add a 200-400 f/4 and a few TC's and compensate with using higher ISO(and I'm referring to 1600 to 6400 here!).
I've seen images taken with Nikons TC's on their long lenses that are definitely fit for the purpose.

Sar Nop(what happened to him? ) has posted fabulous images taken with this combo in the past!
If you can't get closer, and you can't crop too much on a D700/D3, then I'd want to use it's larger, more processable pixels, by compensating with TC's.

Colin

[SpaceJunk]

wonderful thread guys, very useful explanations and nicely garnished with the humour, the UWA kookaburra shot always brings a smile to my face

[Kym]

4.9MP - 450D (1.6 crop, 12.2MP)
6.0MP - 50D (1.6 crop, 15.1MP)

So a Pentax K20D or K-7 at 14.6MP and 1.5 crop with a Bigma (500mm) is quite reasonable for birding

Andrew's 10mm Kookaburra notwithstanding