canon plays catchup
Ever since the D60, Canon has been the one to beat when it comes to DSLR technology. Well, that statement had to be qualified with the introduction of the Nikon D200, and possibly revoked with the announcement of the Sony Alpha 100. For the first time since DSLRs hit the mainstream (courtesy of the Canon Digital Rebel), Canon has to play catchup.
Megapixels
Yeah, yeah, it's not all about megapixels, most people already have plenty for their purposes, blah, blah, blah. Bullshit. We all want as many megapixels as we can get, as long as there are no significant adverse consequences. Canon has led the megapixel battle since DSLRs began to hit the mainstream, but this year it has fallen behind at the entry and prosumer levels. At the entry level, it needs 10 mpix just to match the Sony Alpha 100 and some digicams, and don't think for a second Sony is going to rest on its laurels.
On a more philosophical level, Canon may be a bit over-obsessed with low noise at high ISO levels, probably because this is its trump card, and much of the reason it holds a near monopoly on sports photography. If the noise at ISO 400 is truly negligible, it implies that, in applications with sufficient light and proper exposure, Canon could go to half its current photosite pitch (each photosite gathering one-quarter the light), thus quadrupling its megapixel counts. In that case, a 33-megapixel (the resolution of some medium-format backs) APS-C camera is achieveable with today's technology. By also providing pixel-binning, it could theoretically retrieve close to the noise/dynamic range performance of its existing APS-C sensors.
Dynamic Range in Highlights
I believe Chuck Westfall of Canon North America has mentioned that Canon is working on increasing the dynamic range of its sensors. According to DPReview's review of Fujifilm's S3 Pro, Fujifilm's SuperCCD SR has two stops of extra highlight range over Canon. This is just the start - by my estimate, I'd need roughly four extra stops (7 stops brighter than the subject) to cover the dynamic range in a mixed direct sunlight/shadow shot. Although the final output medium likely cannot reproduce that dynamic range, it would make metering much more straightforward (just spot meter for the subject, and let the highlights fall where they may), possibly eliminating the need for manual metering.
Anti-dust Filter
Sensor dust is truly annoying and productivity-sapping, and it's about time Canon and Nikon stopped ignoring it. Other manufacturers have included a partial solution in their DSLRs for years.
Sensor-based IS
Neither Canon nor Nikon are likely to jump on this bandwagon, unless they can make it compatible with their lens-based IS. Even then, they would lose some of their competitive advantage over third-party lensmakers, so I don't see this happening anytime soon.
Interestingly, Samsung just announced a new fixed-lens camera with CCD-based shake reduction. I believe all previous fixed-lens cameras with IS used lens-based IS. Is this just a case of going with whatever technology is most familiar for each company? But Sony has both sensor-based and lens-based Super Steady Shot. There are obvious advantages to using sensor-based IS on interchangeable lens cameras, but why use it on a fixed-lens camera if Canon's claims for the superiority of lens-based IS are true?
Cramped AF Point Layout
Both the pro and prosumer AF systems in the Canon line seem to be designed for crop-frame cameras, and are just dropped without rescaling into full-frame cameras. Is an AF system really so hard to develop that after six years, Canon has yet to introduce a digital-era system? By comparison, the Nikon D2X has a near-ideal 11-point layout with all but the outermost points being high-precision, cross-type sensors. Why can't Canon copy their layout (and scale it up for full-frame)? I suspect the reason for this is that Canon wide-angle lenses, supposedly the primary reason to go full-frame, have poor corner performance - so poor, Canon's contrast-based AF cannot function properly except in the very center of the lens.
3.5% Spot Metering
You're joking, right? Canon finally gets spot metering (standard on every entry-level DSLR out there) into its prosumer line, and the "spot" turns out to be enormous. On a 5D it's even worse, the "spot" covering most of the area defined by seven central AF points, and entirely encompassing the six hidden AF points. Even Nikon's 2% is nothing close to a true spot, but at least it's a lot closer. 1% might begin earning the name "spot".
Controls
The list of encumbrances is long: no mirror lockup button, two-touch button activation on the 1D series, Auto ISO that maxes out at 400.
Look and Feel
Call me picky, but all the prosumer and entry-level Canons feel like either plastic toys or cheap magnesium bricks with shallow ridges passing for hand grips. By comparison, the new Sony Alpha 100 (which I have not tried out yet) looks organic, shaped for a real human hand.
Sub-optimal Write Speeds
Both Olympus and Nikon write faster, and neither of those seem to be able to stretch a SanDisk Extreme III to its limits (supposedly 20MB/s write). More megapixels require more bandwidth.
Bells and Whistles
Intervalometer
Automatic-detection of steadiest shot
Live histogram
Eye-activated AF
Dynamic Range Optimization
Masks for various aspect ratios
Predictions
Here's my guess of what Canon will (as opposed to should) do:
September 2006
4000D: 350D with 3 AF points, less buttons, smaller. $500 for kit.
February 2007
Merging of the 1D lines. Introduction of Digic III.
1DsIII: 1DsII except with 8fps, larger DR, improved NR, 3" LCD. $6-7k.
5Ds: 5D except 5fps, Digic III, anti-dust, 3" LCD. $2.2k.
September 2007
3" OLEDs from now on.
40D: 30D w/5D AF, 12 mpix, larger DR, Digic III, anti-dust, OLED. $1.4k
400D: 350D with 40D sensor, Digic III, OLED. $800.
February 2008
1Dx: 22mpix, 8fps, still larger DR, OLED, fuel cells, 802.11n. $5-6k.
5000D: 4000D with 40D sensor. $400-500 as kit.
September 2008
1Dx new tech propagates to rest of line.
5Dx: 5Ds w/16mpix, 1Dx DR. $2k.
50D: 40D minor update. $1.2k.
500D: 400D minor update. $700.
2009-11
Digic IV, Liquid lenses, wavefront encoding, WXGA EVF - the end of the SLR
Labels: Photography

Whoa, I'm reading your ancient prediction you made 2 years ago and they're very much within good tolerances! For example, 5Dx (5D2) did in fact come out around Sept with more MP and slightly higher MSRP than your prediction. The 50D did in fact come out with that price range. I guess perhaps if you had been taking in account of USD devaluating and recession, the prediction would have pretty much been right on spot.
Not so close ones: Digic4 came out sooner than your prediction, wayyy sooner. 500D didn't come out, but instead, 400D/450D MSRP are below your 400D/500D predictions. So I guess 550D would be out later.
Overall this is pretty good prediction. I'm impressed. It's been 2 years since you wrote this, got more predictions for the next 2 years? I'm betting on "waaay higher ISO for super low light" capability in the future. Chang's law (my law) says 3 stops of ISO improvement (2^4 * current) every 3 years :)
2:16 AM
Regarding the previous post, I mean 3 stops, and (current acceptable ISO)*2^3. Not 2^4 (that's a typo). So if 3200 ISO is becoming more common for 5D2 users posting to flickr/internet-sized images in 2008 as I've seen lately, they'll have acceptable/equivalent quality at 25600 ISO at around 2011. That's just my guess, or at least hope to happen because I've always struggled with low light picture taking throughout my life and would loooove to have high ISO over IS/VR, MP, smarter flash, etc.
2:31 AM
Post a Comment
<< Home