Camera analysis and guide

I have the Sony RX100 on order, but a rumor is that Canon will introduce an APS-C sized sensor version in the same form factor. Wow now that is pretty cool. It will might have competion from Fuji as well with a low-end X-mount camera. Wow what a time to be looking at these things! Hopefully the RX100 and my old 350D will tide me over while the onslaught of amazing cameras continues…And of course there is the ricoh GR and the Nikon Coolpix A. Both are APS-C cameras with fixed lenses at F/2.8. So the market is pretty confusing right now that are highly rated and what is coming.
Scott points out that you can’t really look at cameras without looking at lenses. That a 21MP camera with a real world lense might only resolve the equivalent of 12MP. So that it is possible for a 12MP camera with a perfect lense to resolve much better. In the end, you can’t take the megapixel rating and compare them. You have to talk about specific lenses paired with them. In fact, in many ways, when buying a camera, you first have to look at the lense quality as in the current state of the art, lenses pretty much completely define the state of the art.
Second learning is that even cameras with the “same” sensor resolution (24MP) can have radically different effective resolutions. As an example the Nikon D3200 is effectively just a 5MP camera whereas the D600 is effectively a 16MP camera if consumer piece of glass in front of it like the Tamron 70-200mm F/2.8.
For instance, my old Canon 350D is 8MP, but it would never resolve all of that with even the best lense. So you might get 6MP equivalent. Also most zoom lenses are not as good as prime lenses because they, well, have to zoom. That means my Sigman 18-200 on my Canon 350D for instance only gives about 4MP resolution average. These are averages by the way. Most zooms have a favorite aperture and focal length. Most primes are the same way. They usually like to be about F/8.
This is one reason why micro-4/3 (m43) has been so popular. It is very mature and the lenses are excellent quality. So even though they theoretically have fewer megapixels and when you just look at sensors, they actually have about the same resolving power. Most of the ILCs and compacts peak at about 11MP effective (even if you have a 24MP sensor like the NEX-7!).
So I looked again all of this (at Scott’s suggestion), using the DXOmark camera-lense database. They are the only folks who have gone to the incredible trouble of testing most lenses against most cameras. Very eye opening!
So for the cameras that I own (or am about to own), here are the effective megapixel stats (by the way, technically speaking this is the average of the modulation tranfer function or MTF multiplied by the sensor effective megapixels averaged over the various zooms and apertures, confused yet!).
Current camera analysis using DXO where they essentially average these scores:

  • Color depth is a measure of its range and every additional bit double the colors it can see and is measured at the lowest ISO. Obviously the color depth changes as you go to higher ISOs and when you get very high, it shrinks, so this measured won’t tell you what cameras hold their colors at high ISOs). 22 bits is considered excellent and you won’t notice a difference below a bit. I’m not clear on how you get fractional bits here.
  • Dynamic range is the lightest to darkest the sensor can see an each EV is one stop. So an extra EV doubles the amount of light that can be registered. This also falls rapidly as ISO rises which isn’t capture by this metric as it is shot with low ISO minimizing noise and maximizing dynamic range. 12 EV is excellent and less than 0.5EV is not noticible.
  • Low light ISO. This is the highest ISO where the camera has a signal-to-noise ratio of 30dB. Doubling the noise means a 6dB decrease in S/N. A difference of 25% in low light ISO is just 1/3EV and you won’t notice it (so ISO of 1000 isn’t very different from an ISO of 1250).

Canon 350D (Rebel XT). Announced 2005 (so it is 8 years old!) with an 8MP sensor. Overall score is 60 with a color depth of 21.8 bits, dynamic range of 10.8 Evs and a low-light ISO of 637. What this means is that you should shoot about ISO 800 and will get a decent photo. Now to see what the effective megapixels, you can look at the lenses I have and see how far off from 8MP you are:

  • Canon EF-S 10-22mm F/3.5-4.5. 5P-Mpix compared with 8MP theoretical maximum. Best score at F/14 at F/4
  • Canon EF 16-35 F/2.8 II USM. 6P-Mpix. Best at 20mm at F2.8
  • Canon EF-S 17-55mm F/2.8 IS USM. 6P-Mpix. Best at 35mm at F/2.8
  • Sigma 18-200mm F3.5-6.3 DC OS. 4M-pix. I’ve always felt that this lense was lower quality (it is best at 18mm F/3.5 and scored a 6Mb DxOmark)
  • Canon EF 24-105 F/4 IS USM. 6P-Mpix. Best at 35mm at F/4 and 9Mb DxO mark
  • Canon EF 50 F/1.4 USM. 7M-pix best at F/1.4. I’ve always thought primes were best and this shows it 🙂
  • Canon EF 70-200mm F/2.8L IS II. 8-mpix best at 100mm at F/2.8

From the above you can see that although you have 8MP sensor, you are really getting effectively about a 5-6megapixel and you can print about a 20-30cm print from this camera with excellent quality and also if you can stand lugging it around the 70-200 F/2.8L is really the best at 8MP so you get the full sensor quality
Canon 5D Mark II with 21MP sensor. Scores 79 with 23.7 bits of color depth, 11.9 EV of DR and ISO 1815 for 30% SNR. There are big improvements everywhere but especially in low noise. Wow, the 70-200 is an amazing lense, but you can see that there are real lense limitations with the 24-105. Even the 50mm doesn’t allow the full resolution of the sensor to show.

  • Canon EF 16-35 F/2.8 II USM. 12P-Mpix. Best at 20mm at F2.8, 20Mb
  • Canon EF 24-105mm F/4 IS USM. 12P-Mpix. Best at 35mm at F/4
  • Canon EF 50mm F/1.4 USM. 15M-pix best at F/1.4. I’ve always thought primes were best and this shows it 🙂
  • Canon EF 70-200mm F/2.8L IS. 13-mpix best at 100mm at F/2.8 with 24Mb

Sony DSC-RX-100. 66 overall score, 22.6 color depth, 12.4 EV dynamic range, ISO 380 for 30% SNR. The main limitation is that it is slower even than the Canon 350D, but it has amazing color depth and dynamic range. I can see why Ming Thein likes it. One interesting point is that it appears to underexpose. That is ISO 800 by the manufacturer is only ISO 548 and ISO 400 is actually 280. That means that you can shoot to about ISO 800 on the manufacturer scale and still not have too noisy an image. Dynamic range wise, you lose two full EV going from manufacturers ISO 100 to ISO 800
Middle-end camera with APS-C and fixed lense (stil thinking about thsi)

  • Ricoh GR.
  • Nikon Coolpix A

Interchangeable or mirrorless lens cameras
Sony NEX-7. Interesting thing is that even with 24MP sensor and APS-C size the actual measured sharpness is much lower. We will have to see how it does with the new high-end NEX9 and NEX-7n coming shortly

  • Sony E 35mm F/1.8. 11P-Mpix
  • Sony E 10-18 F/4. 9P-mpix

Olympus OM-MD5. 71 overall score, 22.8 Color, 12.3 EV DR and 826 ISO low light. Unfortunately Dxomark hasn’t tested any of these (??!), but to get a sense of it, you can see that even with a small sensor, it is getting great resolving that equals the Sony NEX-7 and is just a touch lower than the full frame 5D2. But the main issue (as Scott points out) appears to be the zoom lenses which do give up something compared with primes.

  • Panasonic Lumix DMC-GH2 with a 16MP sensor
    • Olympus M.Zuiko ED 75 F1.8 scored 11P-mpix so equal to the Canon 5D Mark II!
    • Olympus 25mm F1.4 scored 10P-mpix
    • Panasonic 35-100mm F2.8 scored 9P-mpi
  • Olympus PEN EPL2 with just a 12MP sensor scored
  • 9M-pix with an Olumpus M75mm F1.8 equalling the Sony albeit with a zoom

Indicating that even though the micro-4/3 is smaller, you get about the same resolving power as the Sony. In other words, Sony glass isn’t as good nor is Canon glass. Or put another way, the micro-4/3 fixed lenses are fantastic.
Then there are some other ones not tested yet:

  • Fuji XE-1. APS-C, Trans X sensor and X-mount. Low-end X-mount coming shortly but sadly Dxo mark hasn’t tested it so not clear how it would fair. It did test the older X100 with 12MP sensor giving it a 73 score with 22.9 color, 12.4 EV dyanamic range and 1001 low light ISO so on par with the Sony NEX-7 even with a sensor with half the pixels but it isn’t clear how it resolves things as they didn’t test that.
  • Canon relaunches their not super successful EOS-M

dSLRs (mine broke, may not need every again :-0) so I was curious to see how the state of the art in full frame is going
Nikon D800. 36MP full frame is in a word what a camera! 95 overall score (?!!) with 25.3 bits of color depth, 14.4 EVs and 2853 low light ISO. That is terrific performance. These are some pretty amazing resolution scores, but even with 36MP, you are losing about a third of the sharpness. What it means is that with cameras like the D800, you have to spend lots of dollars for the very best glass to get the resolution you want, although with any glass, the camera is going to rock in low light, capturing sports or for landscapes to get the amazing colors and contrast between light and dark.

  • Sigma 35mm F/1.4. 23P-Mpix. This by the way is amazing lense. On a Canon 5D3, it is a 19P-Mpix!
  • Zeiss Distgon 25mm F/2. 22P-Mpix
  • Nikon AF-S Nikkor 85mm F/1.4. 22P-Mpix
  • Nikon AF-S 70-200MM F/4F ED VR. 21P-Mpix. Wow that’s pretty incredible a real keep of a lense. Does lsight better than the F/2.8 which is 20
  • Nikon 50mm F/1.8D. 17P-Mpix so brings you a little down to earth, you are losing half the resolution even with this piece of glass. 
  • Nikon 70-300 F/4.5-5.6G. 12P-Mpix. So with the piece of consumer glass, you are actually better off buying a NEX-7 really if you care about resolution, although of course the D800 is going to be able to shoot at much lower light and with great dynamic range and color. 

Canon 5D Mark III. Ok, this is a camera that isn’t nearly as good as the Nikon D800, but I was surprised that even though it is “only” a 24MP camera, that somehow, its resolution is about the same (?!!) as the Nikon D800. So the glass quality is actaully pretty good:

  • Canon EF 70-200 F/2.8L IS II. 21P-Mpix (I really have to trade in my older 70-200, this type ii is incredible if I keep the EF mount which was only 12)
  • Sigma 85mm F/1.4. 20P-Mpix
  • Canon EF 24-70 F/2.8L II. 18P-Mpix (again, an upgrade is needed, but amazing piece of glass the 24-70 which is “only 16M-pix”
  • Canon EF 50mm F/1.4. 17P-Mpix. Interesting that with the 5D2, this was a 14P-Mpix piece of glass, so the 5D3 is different.
  • Canon EF 24-105 F/4L. 15P-Mpix. But was only 12 on the 5D2
  • Canon EF 16-35mm F/2.8L. 14P-Mpix.

Then there are some cameras that are only just rumors right now…
Canon will challenge the Sony RX100? EOS-M with lens sells for $399. | Mirrorless Rumors

The Sony RX100 showed us hot to make a “real compact” camera with larger sensor. And it sounds like Canon is going to make a competitor with the same size btu with even larger APS-C sensor! CanonWatch reports that “Canon may be working on a compact camera with a large sensor. Not a successor of the PowerShot G1 X (price & specs), but a camera like the PowerShot S110 (price & specs), just a bit a larger. The camera should have a fast lens and the new APS-C 18MP sensor featured on the Rebel Sl1/EOS 100D (price & specs), and the DIGIC 6 Image Processing Unit. No viewfinder. Pricing should be less than the PowerShot G1 X when it was released. We expect Canon to announce this camera at the end of Summer 2013.

Coming soon: Fuji’s entry level X system camera! | Mirrorless Rumors

Patrick at FujiRumors got feedback from trusted sources and they confirmed a new X entry level system camera is coming soon from Fuji. These are the key info:
– It will be announced in Summer
– It will be smaller than the current Fuji X-E1
– It hess less controls and dials than the X-E1
– There is no built.in viewfinder

Mirrorless Rumors

Canon didn’t capture our hearts withe their Canon EOS-M mirrorless camera. And the rumored Canon EOS-M with built.in viewfinder was juts…a rumor :) But Canonrumors reports new rumors (true???) that “Canon plans to “relaunch” the system in the coming months with 2 new cameras“.
The first camera will replace the actual EOS-M model and the second camera will be a completely new high end model. The autofocus will be dramatically improved and at least three lenses will be announced within the year.

Share
%d bloggers like this: