Pixel density for bird photography

[exwintech]

Tannin - I'm surprised that anyone took any of those "pretend figures" seriously....

The post was just a tease-dig for the "Bird Forum" folk....

There is a funny side to it - some of us, on fixed low-income - take quite some time to be able to "select the best available tools" at the camera-shop. Which in turn forces "us inferior folk with our inferior tools" - to just plod-on and do the best we can with them.... (Note: - Joking..... )

Which is both entertaining and educational - though, sometimes if one is having a joke at one's self - it can backfire - somebody will take it seriously - which is funny....

> Okay - I won't do it again.... Shall remain verrry-serrrious.....

Regards, Dave.

[Kym]

The new D7000 and K-5 have 4.75 µm pixels. (23.7 x 15.7 16 Mpx (4992 x 3284))
http://www.dxomark.com/index.php/en/...rand3%29/Canon

Using Tannin's formula its 44058 px / mm² x 135mm² (cropped area) = 5.947Mpx

1.9MP - Nikon D700 (1.0 crop, 12.1MP)
1.9MP - Nikon D3 (1.0 crop, 12.1MP)
2.0MP - Canon 5D (1.0 crop, 12.7MP)
2.2MP - Nikon D40 (1.5 crop, 6.0MP)
2.5MP - Canon 10D (1.6 crop, 6.3MP)
2.6MP - Canon 1D III (1.3 crop, 10.1MP)
2.6MP - Canon 1Ds II (1.0 crop, 16.6MP)
3.3MP - Canon 20D, 30D (1.6 crop, 8.2MP)
3.3MP - Canon 5D II (1.0 crop, 21.0MP)
3.3MP - Canon 1Ds III (1.0 crop, 21.0MP)
3.6MP - Nikon D40x (1.5 crop, 10.0MP)
3.6MP - Nikon D60, D3000 (1.5 crop, 10.0MP)
3.6MP - Pentax K10D, K200D (1.5 crop, 10.0MP)
3.8MP - Nikon D3x (1.0 crop, 24.4MP)
4.0MP - Canon 40D, 400D (1.6 crop, 10.1MP)
4.2MP - Canon 1D Mark IV (1.3 crop, 16MP)
4.4MP - Nikon D90, D5000 (1.5 crop, 12.2MP)
4.4MP - Nikon D300, D300s, D90 (1.5 crop, 12.2MP)
4.5MP - Pentax K-x (1.5 crop, 12.4MP)
4.9MP - Canon 450D (1.6 crop, 12.2MP)
5.4MP - Pentax K20D, K-7 (1.5 crop, 14.5MP)
5.9MP - Nikon D7000 (1.5 crop, 16.3MP)
5.9MP - Pentax K-5, (1.5 crop, 16.3MP)
6.0MP - Canon 500D (1.6 crop, 15.1MP)
6.0MP - Canon 50D (1.6 crop, 15.1MP)
7.2MP - Canon 550D (1.6 crop, 18MP)
7.2MP - Canon 7D (1.6 crop, 18MP)

But the ISO performance of the K-5 and D7000 means you can shoot 2 or 3 stops faster !! (to be proven at this stage)

[Tannin]

the ISO performance of the K-5 and D7000 means you can shoot 2 or 3 stops faster !! (to be proven at this stage)

You would expect some improvement insofar as they have newer sensor designs but somewhat lower density. 2 to 3 stops improvement? Where did that number come from? It sounds utterly ridiculous - a fair-dinkum 1 stop increase would be a very big improvement indeed. Remember that, these days. we already have the big improvements - low-noise CMOS, on-chip interpretation, on-chip NR software, full-coverage microlenses - as standard in all new(ish) cameras. So where are they going to pull 2-3 stops from? Out of their bottoms might be the only option. :shock:

But note - depends a bit on what you measure from - the 7D is quite a bit better, noise-wise, than the old 50D. Maybe half a stop, maybe a bit more, I haven't seen figures but certainly noticable. It wouldn't be hard to get one clear stop over the 50D. But three stops? Tell 'em they're dreamin.

[Kym]

Stops... (see some of the real world samples over on Pentax Forums)

D7000 ISO 4000 http://s1239.photobucket.com/albums/...t=DSC_0162.jpg

D7000 ISO 25,600 http://i52.tinypic.com/142e8nc.jpg (Topaz Denoise applied)

K-5 ISO 51,200 http://www.bertin.ca/tmp/K-5/ISO/K5__9935_MED_1.jpg (Topaz Denoise applied)

My K-5 is due in today (fingers crossed)

So instead of shooting ISO 400 I'm hoping to get similar results at 1600 or even 3200 - I'll let you know

[Kym]

Another one http://www.bertin.ca/tmp/K-5/K-5_ISO6400_ACR6_MED.jpg 6400

[Kym]

@Tannin: To back up my assertion that 2-3 stops faster shooting will be possible with the K-5 (for the same noise)

http://theonlinephotographer.typepad...umber-one.html

Featured Comment by Dan Bridges: "The DxO DR tested performance of the K-5 is amazing. It has the same DR at ISO800 that my K20D has at ISO100. If you use the 'Print' tab on the DxO charts to get a normalised 8 MP result which is useful for comparing sensors with dissimilar MP, at ISO80 there is an impressive 14.1 stops of DR.

I know he is saying DR, but also look at the DxO noise charts esp. SNR 18%. I'll know fairly soon as mine was due in yesterday - hopefully today.

[Tannin]

Oh. Well, I was\ prepared to take this seriously until you mentioned DxO. DxO reckons the unloved Nikon D200 has lower noise than a 5D! (Or was it some other absurdity? Can't remember. In any case, there are lots of them.)

* Warning. The above post is designed to be provocative. This is something I learned from a bloke named Kym. I am actually quite prepared to look at this seriously when I get a minute, but (obviously) I'll disregard any "evidence" presented by DxO.

[Othrelos]

I really don't pay a huge amount of attention to DXO ratings. It's true the Leica M9 has a below-par score compared to other full frame DSLR's when it comes to High ISO. But when you have the use of a 50mm f/1.0 lens, it isn't such a big issue.

Sigma 100-300mm F/4 APO EX DG On a Pentax k10
Using good glass is essential if you're going to be cropping.

[larrywen]

How about Canon 7D?
Opps, I just found the answer in page 3

[Sarge]

The new 16.5mp APS-C 1.5 crop on the Pentax K-5, Sony A55 and Nikon D7000 should be good

5.9MP - Sony A55 (1.5 crop, 16.7MP)

It's good. I wish Sony had the long glass like Canon/Nikon and the excellent AF system I have heard the 7D has.

[davearnold]

So what do you do on a quiet day at work, you read threads that make your head hurt ..... was interesting seeing my old 450D as a contender for a good bird camera ( a long time ago) .. a good thread, and I think I learnt something.

[Dug]

Just an up date for the hell of it

Using Tony's formula, check my figures though.

1.9MP - Nikon D700 (1.0 crop, 12.1MP)
1.9MP - Nikon D3 (1.0 crop, 12.1MP)
2.0MP - Canon 5D (1.0 crop, 12.7MP)
2.2MP - Nikon D40 (1.5 crop, 6.0MP)
2.5MP - Canon 10D (1.6 crop, 6.3MP)
2.6MP - Canon 1D III (1.3 crop, 10.1MP)
2.6MP - Canon 1Ds II (1.0 crop, 16.6MP)
3.3MP - Canon 20D, 30D (1.6 crop, 8.2MP)
3.3MP - Canon 5D II (1.0 crop, 21.0MP)
3.3MP - Canon 1Ds III (1.0 crop, 21.0MP)
3.6MP - Nikon D40x (1.5 crop, 10.0MP)
3.6MP - Nikon D60, D3000 (1.5 crop, 10.0MP)
3.6MP - Pentax K10D, K200D (1.5 crop, 10.0MP)
3.6MP - Canon 5Dmk3 (1.0 crop, 23.4MP)
3.8MP - Nikon D3x (1.0 crop, 24.4MP)
4.0MP - Canon 40D, 400D (1.6 crop, 10.1MP)
4.2MP - Canon 1D Mark IV (1.3 crop, 16MP)
4.4MP - Nikon D90, D5000 (1.5 crop, 12.2MP)
4.4MP - Nikon D300, D300s, D90 (1.5 crop, 12.2MP)
4.5MP - Pentax K-x (1.5 crop, 12.4MP)
4.9MP - Canon 450D (1.6 crop, 12.2MP)
5.4MP - Pentax K20D, K-7 (1.5 crop, 14.5MP)
5.7MP - Nikon D800 (1.0 crop, 36.5MP)
5.9MP - Nikon D7000 (1.5 crop, 16.3MP)
5.9MP - Pentax K-5, (1.5 crop, 16.3MP)
6.0MP - Canon 500D (1.6 crop, 15.1MP)
6.0MP - Canon 50D (1.6 crop, 15.1MP)
7.2MP - Canon 550D (1.6 crop, 18MP)
7.2MP - Canon 7D (1.6 crop, 18MP) I get 7.7mp
9.2mp - Nikon 3200 (1.5 crop, 24.4mp)

[hee]

Thanks to Tannin for that splendid post - clear and informative.
Just a (newcomer) query though. Extrapolating from the figures given, my 18MP 7D would give a 7.15MP picture compard with the 3.3 MP from the much more expensive, full frame 1Ds 111.
I assume that you can't just deduce from this that the 7D is twice as good as the 1D for bird photography so where do the APS-C cameras lose out compared to full frame cameras?

[arthurking83]

when the light gets lower and the need to push ISO higher.

the larger format will usually look nicer, more detailed or have produced better colour.*


*(unless you crop heavily instead of using the correct lens for the situation, or shoot at the right moment)

[Jimv8]

It would be very interesting to see this thread revived in view of the newer cameras and lenses that have come onto the market since the discussion was started, I am thinking along the lines of the 5D 111, the D800 and the likes of Sony's A99

[I @ M]

It would be very interesting to see this thread revived in view of the newer cameras and lenses that have come onto the market since the discussion was started, I am thinking along the lines of the 5D 111, the D800 and the likes of Sony's A99

The figures have been calculated and posted in post #72 for a couple of newer camera models.

[ricktas]

It would be very interesting to see this thread revived in view of the newer cameras and lenses that have come onto the market since the discussion was started, I am thinking along the lines of the 5D 111, the D800 and the likes of Sony's A99

I would be interested is seeing some of your photos, rather than a post every few months just to revive your inactive account. I was thinking along the lines of some photos for CC and perhaps giving some cc to the likes of other members.

[Tannin]

Jim, I think things have changed quite a bit since my original post, which was quite a few years ago now. I haven't tried to quantify this, nor have I performed any formal tests to prove it, but over the last few years I have come to the conclusion that we have reached the practical limit on smaller and smaller pixels, at least for the time being, and at least so far as the purposes of this thread are concerned. Even way back then it was clear that the results to be obtained from very high pixel densities (at that time only to be found in point and shoot cameras) were quite poor. Today, I think that DSLRs are also hitting those limits.

The 7D, as an example, is a wonderful old camera and I still use mine regularly, but it is generally outperformed by the 1D IV even when we are focal length limited. The higher density of the 7D, in other words, is not enough to overcome the other qualities of the 1D IV. I'm open to discussion of the reasons for this, but I suspect that the primary driver is the delivery of significantly lower noise at birding ISOs coupled with reasonably good pixel density. Compare with the 5D II to illustrate this: it has even lower noise, but the IQ under FLL conditions is inferior to the 1D IV and the 7D as well. Presumably, this has mostly to do with its low pixel density (about 24,000 per mm2) but we should also be aware of the huge difference in AF systems - how often do we discount the quality a 5D II image because it's slightly out of focus rather than because of something to do with sensors and density? Perhaps we should consider the 1D III instead.

In short, my view is that for any given level of technology, there is an optimum pixel density for FLL work, just as there is an optimum density for other tasks. Usually (always?) the FLL optimum will be higher than the general-purpose density.

Two key questions remain:

1: are any of the new breed of ultra-high density DSLR-size sensors worth using for bird work, or do they sacrifice too much noise and DR? Sony seems to like them, Canon thinks very poorly of them. Not sure what Nikon is up to these days.
2: is the large downgrade in resolution introduced by the 1DX justifiable for FLL use because of other improvements? It sounds very unlikely to me - indeed I went out of my way to find a 1D IV to upgrade my 1D III to rather than a low-res IDX, but I haven't tried the newer model for myself.

[Tannin]

As posted some years ago above, I no longer see more pixels as translating directly into more reach. Ten years ago that was certainly true, but the state of the art has changed a lot since then. I'm not sure that I would be rushing to draw too many conclusions from the scores in this list.

Nevertheless, I find value in being able to see at a glance (for example) that my go-to birding camera today (a 1D IV) delivers the same pixel density as the 40D I used back when I started this thread. When pushed for reach, however, I often switch to a higher-density body (7D until recently, now a 7D II). Similarly, I can see that (all else being equal) the 5D IV I've got my eye on will have only a little more reach than a 40D, and noticeably less than a 50D or an ancient Pentax K-5.

Anyway, I've updated the list with some representative current Canon and Nikon bodies, and if members think it's of interest, I'll plug in the numbers for whatever the current leading models are in other brands.

• 0.94 - Nikon D1, D1S (1.53 crop, 2.6 MP)
• 1.14 - Canon1D (1.34 crop, 4.1 MP)
• 1.25 - Canon D30 (1.62 crop, 3.1 MP)
• 1.47 - Nikon D2H (1.55 crop, 4.0 MP)
• 1.70 - Canon 1Ds (1.01 crop, 11.0 MP)
• 1.85 - Nikon D3, D700 (1.00 crop, 12.1 MP)
• 1.94 - Canon 5D Mark 1 (1.00 crop, 12.7 MP)
• 1.97 - Canon 1D II (1.26 crop, 8.2 MP)
• 2.10 - Nikon D1X (1.53 crop, 5.9 MP)
• 2.13 - Nikon D70 (1.52 crop, 6.0 MP)
• 2.15 - Pentax K100D (1.53 crop, 6.0 MP)
• 2.16 - Pentax *ist D, *ist DS (1.53 crop, 6.0 MP)
• 2.16 - Nikon D50, D40 (1.53 crop, 6.0 MP)
• 2.17 - Nikon D100 (1.52 crop, 6.1 MP)
• 2.42 - Canon D60, 300D (1.59 crop, 6.3 MP)
• 2.46 - Canon 10D (1.60 crop, 6.3 MP)
• 2.48 - Nikon D4S (1.00 crop, 16.2 MP)
• 2.48 - Nikon D4 (1.00 crop, 16.2 MP)
• 2.53 - Canon 1D III (1.28 crop, 10.1 MP)
• 2.53 - Canon 1Ds II (1.00 crop, 16.6 MP)
• 2.73 - Canon 1D X (1.00 crop, 17.9 MP)
• 2.84 - Olympus E1 (1.95 crop, 4.9 MP)
• 3.04 - Canon 1D X II (1.00 crop, 20.0 MP)
• 3.17 - Nikon D5 (1.00 crop, 20.7 MP)
• 3.19 - Canon 350D (1.62 crop, 8.0 MP)
• 3.20 - Canon 20D, 30D (1.60 crop, 8.2 MP)
• 3.21 - Canon 1Ds III, 5D II (1.00 crop, 21.0 MP)
• 3.37 - Canon 5D III (1.00 crop, 22.1 MP)
• 3.55 - Nikon D200, D60/80/3000, D40x (1.52 crop, 10.0 MP)
• 3.57 - Sony A100, A200 (1.53 crop, 10.0 MP)
• 3.58 - Pentax K10D, K200D, K-m (1.53 crop, 10.0 MP)
• 3.59 - Pentax 645D (0.77 crop, 39.5 MP)
• 3.70 - Nikon D750 (1.00 crop, 24.2 MP)
• 3.70 - Nikon D610 (1.00 crop, 24.2 MP)
• 3.73 - Nikon D3X (1.00 crop, 24.4 MP)
• 3.73 - Sony A900, A850 (1.00 crop, 24.4 MP)
• 4.04 - Canon 40D, 1000D (1.62 crop, 10.1 MP)
• 4.06 - Canon 1D IV (1.29 crop, 16.0 MP)
• 4.32 - Pentax K-x, K-r (1.52 crop, 12.2 MP)
• 4.32 - Nikon D300, D90, D5000 (1.52 crop, 12.2 MP)
• 4.32 - Panasonic DMC-L1 (1.96 crop, 7.4 MP)
• 4.32 - Olympus E330 (1.96 crop, 7.4 MP)
• 4.33 - Nikon D2X (1.52 crop, 12.2 MP)
• 4.40 - Sony A700 (1.54 crop, 12.2 MP)
• 4.59 - Canon 5D IV (1.00 crop, 30.1 MP)
• 4.68 - Olympus E300, E500 (1.96 crop, 8.0 MP)
• 4.91 - Canon 450D, 1100D (1.63 crop, 12.2 MP)
• 4.96 - Sony A350, A380, A390, A290 (1.52 crop, 14.0 MP)
• 5.07 - Sony A550 (1.54 crop, 14.0 MP)
• 5.24 - Pentax K20D, K-7 (1.54 crop, 14.5 MP)
• 5.24 - Nikon 3100 (1.56 crop, 14.2 MP)
• 5.53 - Nikon D800, D810 (1.00 crop, 36.2 MP)
• 5.53 - Nikon D810 (1.00 crop, 36.2 MP)
• 5.73 - Sony A580 (1.53 crop, 16.0 MP)
• 5.76 - Nikon D7000, D5100 (1.53 crop, 16.1 MP)
• 5.81 - Pentax K-5 (1.54 crop, 16.1 MP)
• 5.85 - Olympus E3, E410/20, E510/20, Panasonic DMC-L10 (1.96 crop, 10.0 MP)
• 5.97 - Canon 50D, 500D (1.61 crop, 15.1 MP)
• 7.11 - Canon 7D, 60D, 550D, 600D (1.61 crop, 17.9 MP)
• 7.14 - Olympus E3, E30 (1.96 crop, 12.2 MP)
• 7.38 - Nikon D7500 (1.53 crop, 20.7 MP)
• 7.68 - Canon 5Ds (1.00 crop, 50.3 MP)
• 7.83 - Canon 7D II, 70D (1.60 crop, 20.0 MP)
• 8.63 - Nikon D5500 (1.54 crop, 24.0 MP)
• 8.63 - Nikon D7100 (1.54 crop, 24.0 MP)
• 8.63 - Nikon D5600, D3400, D7200, D5500 (1.54 crop, 24.0 MP)
• 8.63 - Nikon D7100 (1.54 crop, 24.0 MP)
  

Previously, I have quoted the manufacturers' nominal crop. This time I've let the software calculate its own more accurate crop, based on the manufacturer's specifications for sensor size. It is interesting how much variation we see from the nominal 1.5 and 1.6 crops usually cited.

[Cage]

Tony, I've just had a read through this thread again, and while it didn't make my head hurt like it did nine years ago, it still leaves me with some unanswered questions.

It seems to be the general consensus that at the same time, same place, same good light, same lens, same subject, same camera settings, the crop sensor is going to capture more detail due to it's higher pixel density.

I'm assuming we are talking 'good light' here, and that is not only subjective, but variable, depending on the subject.

For example, a Blue Wren in open sunlight is going to be a challenge for any camera and settings, trying to cope with that highly reflective plumage, while striving to get some detail in the dark and light feathers. BTDT ! An exercise in frustration.

And when that same wren hops into some heavy shade, or a cloud covers the sun, or both, does the pendulum then swing toward the extra light gathering ability of the full frame's bigger pixels, or will today's sensors, with their better control of noise at higher ISO settings still enable the crop sensor to win out.

When I look back at my bird shots over the last nine years, I'm hard pressed to find many stand-outs taken in direct sunlight, and most keepers seem to have been taken under the canopy, or when cloudy, or both.

My jury is still out on the best type of birding camera for me, but my gut feeling is that for my usage a 35+MP FF is the way to go. Shame I can't afford a $10K lens to get the best out of it.