An explanation of Fujifilm’s Super CCD EXR sensor

A look at Fujifilm’s innovative EXR sensor, the latest iteration of its flagship Super CCD sensor, along with some analysis of images from production cameras. Admittedly this would have been more interesting as a speculative piece a year ago, but better late than never

tl;dr: Fujifilm’s EXR sensor is extraordinary, mostly for its dynamic range. If you’re after the best non-DSLR image quality around, your choices start at the Fuji F200EXR, F70EXR, S200EXR, and end there.

Fujifilm has long been a leader in revolutionary sensor technology, particularly at the smaller scale sensor market where the majority of manufacturers have long been content pumping out traditional, vanilla CCD sensors with square grid-based Bayer Filter Arrays.

In September of 2008, announced plans for their latest sensor: the Super CCD EXR, which combines the unique color filter array (CFA) and pixel binning features of various previous sensors into a single “switchable” sensor that can be optimized in one of several areas (which are typically mutually exclusive when designing a sensor): high resolution, high dynamic range, and low noise.

High resolution

High resolution mode is the default mode, which utilizes the full set of photosites on the sensor and produces an image with a corresponding pixel on each photosite - nothing too special here, though Fuji claims the diagonal layout of photosites (as opposed to simple square grid) helps to improve resolution.

High sensitivity

A comparison of a typical Bayer CFA (left) and the CFA on Fujifilm's new EXR sensor (right)

The second mode of operation for the EXR sensor is a high-sensitivity mode which Fuji calls “Pixel Fusion Technology”, which is fancy marketspeak for pixel-binning (combining reading from adjacent pixels together to produce a better signal). With the EXR’s pair-based CFA layout, Fujifilm claims that interpolation (and thus color resolution) will be more accurate because the binned pixels are closer together (e.g. the pair blue pixels are pretty much in the same location, while they’re separated by two pixel lengths in a standard square-grid Bayer array. I don’t know that I buy this argument particularly well - it’s true that same-color pixel values will be more accurate since they’re closer, but you can’t get something for nothing: for example, the average distance from red-to-blue is going to be increased, which lowers accuracy for interpolating blue values at red pixels.

Read More »

Canon 7D and 1D Mark IV: new 1D and 1D junior

TL;DR version: A long diatribe on how the latest Canon releases completely underwhelm in the face of competition, especially from Nikon.  The 7D is a decent upgrade that’s completely overrated simply due to marketing. The 1D Mark IV sounds nice and has the capability the 1D Mark III probably should’ve had - unfortunately its functionality has been completely eclipsed by Nikon’s D3(s) and even D700, which unlike the 1D’s 1.3x crop sensor, are able to pull double-duty as both heavy duty sports bodies and general purpose cameras.

Canon's 7D, which is essentially a 60D with fancy marketing and a higher price tag

Canon's 7D, which is essentially a 60D with fancy marketing and a higher price tag

It’s interesting to see how much an effect marketing has on the general photography consumer. Over the past few months, Canon has released a couple of moderate upgrades, one of which has been hailed as revolutionary and game-changing, and the other which was met with a big collective yawn and cries that Canon has fallen behind the cutting edge and is playing catch-up with Nikon. The biggest difference? One camera was given an incremental version number, and the other was given a new model number as the start of a different series. Read More »

Tamron 17-50mm f/2.8 VC: An image-stabilized, midrange crop lens for the masses

The Tamron 17-50 f/2.8 VC - currently one of only two f/2.8 midrange zoom lenses on the market, and the only one under a grand ($650 to be exact)

When it comes to midrange lenses, there’s a few different approaches. Of course, a lot of people start at the low end with an 18-55mm kit or so, but eventually most people graduate and there are pretty much two ways to go:

  1. A small range, high-quality, high-aperture zoom.
  2. A large range ultrazoom with (usually) lower quality and smaller aperture

One of the advantages where ultrazooms seemed to gain the leg-up on large aperture zooms was in the image stabilization department, where nearly every single ultrazoom lens does, but up until recently only one large aperture crop lens (Canon’s $1000+ 17-55mm f/2.8 IS) did.

That left just one (very expensive) option for Canon users, and Nikon users completely out in the cold (they pay $1300 for a 17-55 f/2.8 without VR). Third party manufacturers, have as always had cheaper alternatives, such as Sigma’s 18-50mm f/2.8 and Tokina’s 16-50mm f/2.8, but all of these lacked any sort of stabilization as well.

Now finally, Tamron has gone ahead and introduced their VC stabilization to their flagship crop standard zoom, the 17-50mm f/2.8 (or rather, Tamron SP AF 17-50mm f/2.8 XR Di II VC LD Aspherical [IF]), which finally delivers a large aperture, image-stabilized standard zoom for an affordable price ($650 currently).

It’s only knock is that it doesn’t yet have the fast USM or SWM-based autofocus of the Canon or Nikon models, although it’s a feature I’ve long-regarded as over-rated for standard zooms for most people’s actions.

Panasonic does Micro Four-Thirds right with the GF1

In what many see as the next big evolutionary step for digital cameras, Panasonic and Olympus made a bold move with their introduction of the Micro Four-Thirds system, an electronic viewfinder, interchangeable lens (so-called “EVIL”) system that eschewed the mirror assembly found in traditional SLR cameras and offering image preview via a live view feed only.

Aside from the numerous advantages associated purely with live view (and could technically be realized with a traditional DSLR - it’s just that forcing live view only is likely to spur much more rapid development), the one key advantage to Micro Four Thirds (and upcoming systems like it, such as Samsung’s NX system) is that the removal of the mirror assembly allows lenses to sit much closer to the image plane, making for much smaller camera bodies and lenses.

The first few of these cameras - the Panasonic G and GH1 - failed completely to live up to the small form factor potential - they were shaped much like traditional SLRs, albeit slightly smaller.

Panasonic GH1

The Panasonic GH1 - one of the first Micro Four Thirds cameras which didn't quite realize the potential of the formfactor

Next, Olympus released a Micro Four Thirds of its own: the E-P1 “Pen” which harked back to Olympus’ historical line of compact film cameras. Unlike the G1, the E-P1 actually began to approach what some would call “compact” - it was just 1.4in thick, though that’s not taking into account the attached lens.

Now Panasonic is jumping in on the bandwagon with their E-P1-esque GF1, which sports a slim compact-like body. The specs are nothing to get excited about, though it does have the a built-in flash that was notably missing from the E-P1. In a puzzling decision though, Panasonic decided not to implement any sensor-based image stabilization, relying on lens-based IS to counter camera shake. Unless they were denied a sensor IS license by Olympus (a possibility), I’d say this is a rather bone-headed decision, since any stabilized lenses will add weight unnecessarily (or in the case of pancake lenses that are pretty much made for this kind of camera, impossible to add in), defeating the entire purpose of Micro Four-Thirds.

The two kit lenses offered with the GF1 are a bit more appealing than the E-P1 package: a standard 14-45mm OIS kit lens and a 20mm f/1.7 pancake prime. The prime still isn’t quite there to portrait range and gets even further away from all-around wide angle utility than Oly’s 17mm f/2.8 pancake, but it does offer a much larger f/1.7 aperture.

A comparison of the new landscape in premium compacts:

Panasonic GF1 size comparison

Camera Size Focal range (equiv) Aperture (equiv)
Canon G10 4.3x3.1x1.8in 28-140mm f/13-21
Fujifilm F200EXR 3.8x2.3x0.9in 28-140mm f/14-22
Panasonic LX3 4.3x2.3x1.5in 24-60mm f/9.4-13
Sigma DP1 4.5x2.3x2.3in 28mm f/6.7
Sigma DP2 4.5x2.3x2.3in 42mm f/4.7
Olympus E-P1 w/ 17mm f/2.8 4.7x2.8x2.3in 34mm f/5.6
Olympus E-P1 w/ 14-42mm f/3.5-5.6 4.7x2.8x3.1in 28-84mm f/7-11
Panasonic GF1 w/ 20mm f/1.7 4.7x2.8x2.4in 40mm f/3.4
Panasonic GF1 w/ 14-45mm f/3.5-5.6 OS 4.7x2.8x3.8in 28-90mm f/7-11

As expected, the added IS to the Panasonic kit lens makes it much larger (22.6% longer) than the E-P1 setup. Panasonic’s pancake, however, is about the same size as Oly’s 17mm and with its f/1.7 aperture is by far the best in terms of large aperture performance (35mm equivalent of f/3.4)

If you’re in the market for this kind of camera though, the most sensible thing seems to be taking the E-P1 to get yourself sensor-based IS, and combining that with Panny’s 17mm pancake prime. Though you will be losing out on the built-in flash, which is somewhat of a must-have for a camera like this (since again, needing to carry around a huge external flash defeats the size advantage).

Pentax K-x

Pentax K-x (space white)

Following up on their K-7, Pentax has now come up with an entry-level K-x. While it doesn’t bring anything groundbreaking that wasn’t already seen on the K-7, it packs in many of the features seen on many competitors’ midrange model, and perhaps pending reviews on image quality and disregarding the overall Pentax system upgrade options, is probably the best choice out there currently for the beginning photographer/student.

The big headline features:

  • 12.4MP CMOS (different from the K-7 14.6MP sensor, but interestingly also uses CMOS unlike all previous Pentaxes which used CCDs)
  • ISO up to 12.8k
  • Live-view with face-detect AF
  • 720p, 24fps video
  • 4.7fps continuous shooting
  • $650 MSRP with 18-55 kit lens (and likely to drop further once it gets off pre-order)
With the specs listed, this is a camera you’d expect in the high-hundreds, competing with the likes of Canon’s Rebel T1i or Nikon’s D5000/D90, yet it’s got a price closer to that of the entry-level Rebel XS or D3000.

A comparison:

Pentax K-x comparison

Camera Canon Rebel XS Nikon D3000 Pentax K-x Nikon D5000 Canon Rebel T1i
Sensor, crop 10MP, 1.6x 10MP, 1.5x 12MP, 1.5x 12MP, 1.5x 15MP, 1.6x
ISO range 100-1600 100-3200 100-12800 200-6400 100-12800
Live-view? Yes No Yes Yes Yes
Live view AF Yes None Yes, face-detect Yes, face-detect Yes, face-detect
Video None None 1280x720, 24fps 1280x720, 24fps 1920x1080, 20fps
AF 7pt, 1 cross-type 11pt, no cross 11pt, 9 cross-type 11pt, 1 cross-type 9pt, 1 cross-type
Continuous FPS 3fps jpg, 1.5fps raw 3fps 4.7fps 4fps 3.4fps
Image stabilization lens-based lens-based sensor-based lens-based lens-based
Size 127 x 97 x 61mm 127 x 97 x 64mm 122 x 91 x 69mm 127 x 104 x 79mm 130 x 97 x 61mm
Weight 450g 485g 516g 560g 480g
Price (with kit lens, Amazon) $499.95 $529.95 $649.95 $719.63 $781.89

In a comparison with the $500 entry-level cameras, the K-x blows them away in nearly aspect, and goes toe-to-toe or even exceeds the D5000 and Rebel T1i in every single category, despite being significantly cheaper (especially once the street price drops lower from MSRP)

Interestingly enough, the Pentax K-x will come in a variety of colors, including an ultra-spiffy red (below), the space white shown above, and your ordinary black.

Interestingly enough, Pentax Japan features a site where you can come up with your own custom color scheme, and apparently order it as well, which personally is an insanely appealing prospect.

Pentax K-x (red)

Pentax K-x custom design - design your own!

Pentax K-x custom design - design your own!

I think this is a ranking of custom designs that users have created

I think this is a ranking of custom designs that users have created

Pentax K-x press release

Past three months in camera news: Nikon

Apologies to all for dropping the ball for the past three months - it’s been a whirlwind start to the semester here. Big, recent developments:

Nikon SLR Refresh

Nikon introduced a couple of new SLR updates, the updated D300s and D3s. The D300 is a pretty incremental upgrade to the mid-level D300, offering a modest +1fps improvement in continuous shooting (up to 7fps), and bringing the video capability that’s now standard on every new DSLR.

The bigger story came a few months later, in the form of the D3s. While still not a revolutionary introduction, it is much more than a software refresh. Among the features of note were a video mode (at 24fps!, albeit only at 1280×720 (720p) resolution), 11fps available in a higher-res crop mode (it now crops only 1.2x instead of 1.5x), and an increase in ISO range, up to ISO12.8k natively with a boost to ISO100k. The D3s presumably packs a different sensor, though it still maintains the same 12.1MP resolution.

People have been going goo-gah over the last spec in particular, especially given such a high linear number for ISO (and from here, it’s just four more stops til we get to ISO1.6 MILLION), though it’s really just +1 stop natively and +2 stop boost over the previous D3. And it’s important to note that the simple availability of an ISO capability says nothing about image quality at that level - that would be the same mistake as having the maximum shutter speed expanded from 30 seconds to a minute, and somehow thinking this magically makes photos at 1/500s less blurry. Given that the resolution (and thus pixel pitch) remained the exact same, I certainly wouldn’t expect quality to be any worse than the D3, and quite probably will be a tad better (although I have extreme doubts about the ISO100k mode, which is digitally boosted 3 stops; things have always looked terrible at just +2 stops digitally, even boosting ISO100+2 to 400.)

All in all, about as much as you could expect from Nikon, who seems to do very incremental updates and waits a long time to deliver big, revolutionary refreshes. Here’s hoping we see that D3s sensor in a D700s soon, though 1080p at 24fps would be nice (and completely feasible: 1920×1080x24fps = 49.8MP/s throughput, while we definitely know that the D700 supports 12.1MPx9fps = 108.9MP/s throughput in its continuous shooting mode).

Nikon D3s press release (I don’t know why they keep referring to it as “D3S”, past history and even the logo in the image clearly denote “D3s”)

The Nikon DS3, now with 720p video and ISO up to 12.8K/100K(boost)

The Nikon DS3, now with 720p video and ISO up to 12.8K/100K(boost)

Nikon also announced a couple of lens refreshes, with Version II’s of their popular 18-200mm VR ultrazoom and a long-awaited update to the 70-200 f/2.8 VR to optimize it for full-frame (FX) sensors. As Nikon had long trumpeted 1.5x crop DX sensors before their introduction of the full-frame D3 in 2007, they cut corners with their introduction of the 70-200 f/2.8 VR in 2003, building a lens that was technically full-frame but had an abysmal drop-off in performance once you actually got to the corners outside of a 1.5x imaging circle. This wasn’t found out until a bit after the D3 was released, finally giving digital photographers a platform to test the lens’ full frame performance, which resulted in tests like these:

http://www.dpreview.com/lensreviews/nikon_70-200_2p8_vr_n15/page6.asp

http://www.dpreview.com/lensreviews/widget/Fullscreen.ashx?reviews=17&fullscreen=true&av=3&fl=105&vis=VisualiserSharpnessMTF&stack=horizontal&lock=&config=/lensreviews/widget/LensReviewConfiguration.xml%3F4

The new 70-200 II promises to fix all of these problems with a new optical design and coatings, and promises to throw in a more effective “4-stop” VR system as well. There haven’t been too many authoritative tests yet to show how it performs (if you’ve found any, send me a link!), but presumably they should have no problem building such a lens - Canon has had two 70-200 2.8’s that’ve performed flawlessly on full-frame, and Nikon itself had a great 80-200 2.8 lens prior to the 70-200 VR I.

The one stickler? As if Nikon’s $2019 price on the original 70-200 I wasn’t enough, the 70-200 VR II will now set you back a cool $2400.

Nikon 18-200 II and 70-200 2.8 VR II press release

A real TZ-killer: Fujifilm’s F70EXR

Possibly the biggest announcement in the compact sector since the Panasonic TZ1 - Fujifilm finally puts together not just a compact ultrazoom but an ultracompact ultrazoom (0.9 in thin), and manages to fit in a half-inch SuperCCD sensor to boot.  If ever a camera came along with the potential to dethrone Panasonic’s vaunted TZ/ZS-series, this is it.

Have got a big night tonight - will update later, but for now you can munch on the details of the press release and Imaging Resource’s short overview/analysis.

Olympus E-P1 - a size comparison

In a release that they’ve hyped for weeks now, Olympus finally pulled the covers off their first Micro Four-Thirds format camera, the E-P1.

Good product photography - sure looks dainty doesn't it?

Good product photography - sure looks dainty doesn't it?

For those of you not already in the know, Micro Four-Thirds is a new interchangeable lens system developed by Olympus and Panasonic which is the first mirror-less digital camera system to feature interchangeable lenses.  The removal of the mirror (and associated prism and optical viewfinder) and the exclusive usage of live view for image preview enables a drastic size reduction for both cameras and lenses, and as you can see here, the new Olympus E-P1 is tiny tiny tiny.

How tiny?  The exact specs on the E-P1 are 121 x 70 x 35mm (4.7 x 2.8 x 1.4 in) and 335 g (11.8 oz) - body only, with no batteries - which firmly plants it in compact camera category.  Of course, you’ll need to attach a lens at some point before shooting, which will add some bulk, but as of now the E-P1 indisputably offers the most compact interchangeable lens solution.

The following is a run-down of things you might have already picked up from other news sources or blogs. The real interesting stuff is the size and equivalent aperture/focal length comparison, at Size Comparison.

Read More »

Camera Guide, June 2009

I get a lot of questions all the time about which camera to buy, and most of the time I don’t quite have the time to respond to all of them, so I’ve decided I’d start a regular camera guide feature.  These aren’t in-depth rationales or anything, just quick summaries that tell you, unequivocally, which is the best of the best; I also write digital camera recommendations for a site called bestinclass.com - I might recommend visiting there if you’re looking for more in-depth analyses and comparisons.

Digital cameras don’t come out all that often, so this may be a quarterly guide.  This also may not reflect the very cutting edge - it takes a little while for reviews for the newest cameras to come in, and it’s impossible to really assess the cameras until they do.

All prices based on the lowest of amazon.com or bhphotovideo.com

Have you got a use case/need that isn’t covered here?  Feel free to post it in the comments, and I’ll keep it in mind for future guides.  And if you think differently about any of the cameras, feel free to share that too!

General advice:

To give you all an idea of the perspective these recommendations are written from, here’s a few guidelines I mostly go by:

Features trump image quality: With modern cameras, image quality differences are mostly a consideration of the past.  Almost every camera released today has megapixel resolution far in excess of what’s needed (or even usable) for most applications, and in most daylight scenarios there is practically zero difference between cameras, especially among the top tier of manufacturers.  The main differentiator in your photographic experience and capability, then, is what features you’ll have to work with - being able to take a wide shot with a 28mm wide-angle lens, or having a fast 4fps continuous shooting mode for action shots, for instance, is going to go a long way towards getting you the photographs you want compared to minute differences in image quality or resolution.

Price/performance: The recommendations for different categories will mostly recommend the camera with the best value proposition - a lot of these are often written in the format of: Best budget camera under $200, best midrange camera under $300, best premium camera under $500, etc. While oftentimes, yes it’s true that Camera Xa has a slightly bigger LCD screen than Camera Xb and is therefore better, and the $50 premium still puts it under the $300 budget, as a knowledgeable consumer you wouldn’t want to spend that much more on a mostly cosmetic difference, and as an informed friend you would do best by recommending Camera Xb to your friend.

Simple Ultracompact

For many people, cameras are just cameras, and all they need is something that, for lack of a less-hackneyed phrase, they can “point and shoot”.  They’re not interested in photography and don’t need nor want full manual controls, and rarely would use and can make do without a huge zoom range.  They’ll take snaps while they’re out at social events or just randomly at home or in their room, but that’s about it.  For this group there’s the simple ultracompact - a basic camera that has a few useful features (possibly wide-angle lenses for indoor group photos and image stabilization for low-light situations) but otherwise just provides good overall quality and a small formfactor that can be slipped just about anywhere.

Simple Ultracompact, midrange: Canon SD960

A plethora of Canon, Panasonic, and Fujifilm cameras rule the roost in the ultracompact category, all roaming around the $300 range.  For a little less than $300, Canon’s SD960 puts together the best combination of image quality and features, with a 28mm wide-angle, image stabilization, and even 1280×720 (720p) video capability.

  • 12MP resolution
  • 28-112mm (4x) zoom range
  • f/2.8-5.8 aperture
  • 1280×720, 30fps video (720p)
  • Image stabilization
  • 145g (5.1oz)
  • 99.1 x 53.3 x 22.9 mm (3.9 x 2.1 x 0.9 in)
  • 200 shots battery life (CIPA)
  • $281 on Amazon

Read More »

The Demosaic Project

So over the past couple of weeks I’ve been working on a little project called Demosaic.  It’s a little online demo that interpolates image data from (simulated) raw sensor output, similar to what almost every digital camera used today has to do.

http://www.thedailynathan.com/demosaic/

Read More »