Tokyo wide open

Canon’s 50mm f/1.0L is a one of a kind optic – the fastest autofocus lens ever manufactured. You can’t buy one new – it’s been replaced with a slightly slower and considerably cheaper f/1.2 lens – but there are still plenty around second hand and via rental houses. Visiting Tokyo for a brief work trip I thought it would be fun, and maybe instructive, to limit myself to just this one lens and see what sort of dreamy, “Lost in Translation” look I could manage amidst the Tokyo neon.

The advantage of the f/1.0 is, self-evidently, its speed. The chief disadvantage is cost, cost and size. The two disadvantages are cost and size and focus speed. The three disadvantages are cost, size, focus speed and corner sharpness. Among the many disadvantages are… well you get the idea. It’s not a general purpose tool. This is a lens that you buy or rent because, wide open or slightly stopped down, it yields a unique look that isn’t going to be mistaken for an iPhone image. Only a fool would stop it down, and with that in mind my original plan was to shoot everything at f/1.0. That turned out to need more skill and experience than I managed to acquire in my brief time with the lens, and I shot a number of frames where the background – which was intended to give a hint of a recognizable image – was blurred beyond recognition (it doesn’t help that this is a lens that, at f/1.0, opens up when you press the shutter). Of the f/1.0 frames, my favorite is this image of pilgrims praying at the Senso-ji shrine:

Senso-ji shrine

Pilgrims at the Senso-ji shrine, Asakusa, Tokyo (f/1.0)

Sake barrels Meiji shrine

Sake barrels at the Meiji shrine (f/1.0)

Kashiwa at night

In the suburbs… Kashiwa at night (f/1.0)

In daylight, you won’t be shooting the bare lens at f/1.0 as a shutter speed of 1/8000 is too slow even at ISO50! You’ll need a neutral density filter. Lacking one of those, most of my daylight shots were “stopped down” to f/2.0. The lens has quite a distinctive, and to my taste attractive, character even then.

policeman tokyo

Traffic police, Tokyo (f/2.0)

The lens is also usable for available light photography and for portraits, though for these purposes Canon and others make cheaper and / or sharper optics (e.g. the new Sigma 50mm f/1.4, or the Canon 85mm f/1.2) that would probably win out in a head-to-head comparison.

Hilton Tokyo view

Panorama from the Shinjuku Hilton, twilight (f/2.0)


Chris, at the New York Grill (f/1.2)

Overall, my impressions from just a couple of days shooting with the f/1.0 were pretty much a mirror of the numerous reviews you can find on the web. The lens is sharp wide open at the center – albeit with a very shallow depth of field – while away from the center it’s soft with a crazy hotchpotch of optical peculiarities and mishapen bokeh. Although that sounds bad on paper, it’s a look that’s very easy to love. One area where I’d quibble with most reviews is the autofocus speed. No, it’s not fast, but it’s not go away and brew a cup of tea slow either. For most situations you’d use a lens like this for, I found it more than adequate. I’d be hard pressed to recommend buying one – let’s face it this is a $4000+ lens that would quite possibly be impossible to repair if it broke – but I was sorry to part with it and am already thinking of other trips where renting it again would make sense.


Panorama from Park Hyatt Tokyo

Back from a week in Japan. More photos to come but as a teaser… the panorama from the bar on the 52nd floor of the Park Hyatt hotel in Tokyo. Kudos to the location scout for Lost in Translation as you really can’t go wrong here!

The Scottish Highlands

A quick trip through the Scottish Highlands over Spring Break… starting from Stirling then up through Glencoe to Fort William before circling back along the edge of the Cairngorms to St Andrews on the East coast. It’s amusing to note that we made it almost to 57 degrees north, not far south of the latitude of Churchill in Manitoba where polar bears roam! This was not a photographic trip, but the snow on the peaks and Scotland’s notoriously fickle weather made for some interesting scenes, captured here with the Sony RX100m2. I don’t like the lack of an optical or electronic viewfinder, but the image quality of the little Sony continues to impress.

Stitched Panorama

Approaching Glencoe, Scottish Highlands

Stitched Panorama

The rounded peaks of the Cairngorms, at twilight


Isolated houses in the Trossachs

Boulder Flatirons dawn

Stitched Panorama

Busy month at work including a few days with early starts to join video conferences on the East Coast. Not much fun, but I did get to see a couple of sunrises on the Flatirons from my office… always a treat!

New cameras: what we’ll get vs what we want

It’s been a pretty quiet year or two in the Canon world, at least for those of us who aren’t in the market for Cinema EOS. Nikon, Panasonic and Sony have all announced or shipped some distinctively different cameras (the Nikon Df, Sony A7r, Panasonic GH4), while from Canon we’ve had the 6D (protecting their position at the low end of full frame) and a bunch of low-end Rebels. Even going back a bit further, the 5D3 – currently my main camera – was undoubtedly a less dramatic introduction than Nikon’s D800. So I’ve been mulling over what we might expect from Canon this year, along with what I might actually want, given that the 5D3 satisfies pretty much all my current photographic needs.

What we’ll likely get from Canon is probably the easier question to answer. Nikon and Canon remain, for now, a virtual duopoly in the market for high-end cameras, and duopolies stay that way by paying close attention to each other’s moves. By all appearances the Nikon D800 has been very successful, so the chances of the next Canon full frame body having a substantially higher pixel count must be high. Whatever Canon might espouse about 18 or 22MP being the “right” resolution for full-frame, they’ll surely seek to stop Nikon having a monopoly on a high resolution body as soon as it’s technically feasible (all of Canon’s current full-frame sensors are apparently manufactured on a less than state-of-the-art 0.5 micron process). Likewise dropping the anti-aliasing filter hasn’t led to howls of complaint from Nikon D800E or Sony A7r users, and I’d expect Canon to follow suit. The logic of Canon’s own recent products (the 6D and 70D) says we’ll get phase detection on the sensor, built-in WiFi (and maybe GPS), and 4k video. The last might get delayed a generation to continue to differentiate the 1DC, but with the GH4 on the way it’s virtually certain the flagship Cinema EOS DSLR will go upmarket, perhaps with RAW, 10 bit color, Apple ProRes codec, or similar features appealing to the video crowd. On the lens front there’s been quite a bit of innovation lately, but there are still plenty of “L” lenses awaiting their makeover to make them sharper wide open for current digital bodies. More telephotos with built-in extenders is also a pretty safe bet.

Just this me-too and evolutionary stuff would actually be pretty useful, even if it’s not very exciting. As I’ve mentioned before, I was struck over the summer by how obvious an edge the D800E has over the 5D3 in resolution, even when shooting handheld with high consumer-grade glass. I’m also plenty happy to let video cognoscenti pull focus manually, and switch to autofocus enabled by phase detection in live view and video modes. As for 4k, I’d rather play with 120fps slow motion, but we’ll probably get both as a package.

What we won’t get is also predictable. If Canon and Nikon perceive any threat from Sony, and possibly others, launching serious mirrorless cameras, that threat is most likely to come from the inherent cost advantages of simpler systems rather than better image quality or even lower weight. So we won’t see any additional mechanical complexity being added to DSLRs, even if built-in ND filters say would be a nifty feature for video. I also don’t see a “Canon Df” as being very likely – nostalgic Canon types already had their hearts broken way back when Canon dumped the FD lens mount.

What I want is trickier. My 5D2 had an achilles heel – it broke when I tried to use it hiking the Berg Lake trail in heavy rain – but the 5D3 handles the photographic situations I’ve personally encountered satisfactorily. What I most want is some Steve Jobs-like innovation I don’t know I want! In the category of disruptive changes, I’ve read enough favorable reports about the Sigma Merrill cameras, from people whose opinions I respect, to be interested in a body with a non-Bayer multi-layer sensor. I’d certainly sacrifice a stop of high ISO performance for what is said to be a subtler color rendition. Three stops? Well at that point I’m not so sure. I’d also like to see a hybrid viewfinder. The recent addition of grid lines and electronic levels into the viewfinder is already a very useful feature, but more broadly the whole raison d’être of a DSLR is that looking into the viewfinder gives you your best view of the world. Camera companies know this, and take some pains to design the controls so that you don’t have to take your eye away to change them. But of course in the real world we mostly take our eyes away from the viewfinder after almost every shot, to check how it looks on the rear LCD! Logic says we should be able to review images in the viewfinder, and while we’re about it switch from an optical to live view in the finder when it gets gloomy (anyone with a modern body knows all too well that one can take perfectly good pictures long after one can no longer see clearly through the lens).

Then there are the minor things that would surely be useful, but which require software changes that Canon seems reluctant to provide. I do a fair bit of time lapse and night photography, and it’s faintly ridiculous that I need to buy various triggers and gizmos to accomplish what a Canon engineer could code up as an in-built feature in literally a day of work. More substantially, the “Dual ISO” Magic Lantern hack strongly suggests that Canon’s imaging pipelines have qualitatively different modes, that in some circumstances would be of real benefit to the photographer. Certainly there are downsides too. But we already have options – like ISO 50 – that are useful despite lowering image quality, so it would be great if Canon let us officially use more of what their hardware is capable of on a caveat emptor basis.

Finally there are lenses. I’m happy with Canon’s recent and legacy offerings, though along with every man, woman and child on the planet I wonder when the 100-400mm f/4.5-5.6L will be updated. There’s also an obvious gap in somewhat more affordable super-telephotos. At 300mm, the f/4L is old. At 400mm, the f/4 DO occupies a lightweight / expensive / optically not-so-great niche of questionable interest, while the f/5.6 is so old it doesn’t even have IS. Beyond 400mm, nothing. It seems obvious that some updating and rationalization of the 300mm and 400mm offerings is due, and that a 500mm f/5.6 would be an interesting optic. Losing a stop of light is hardly a big deal with current bodies that are at least two stops more sensitive than early pro models, and that sacrifice would be more than outweighed by the lower weight and likely significantly lower cost.

A star trail panorama


Moonrise at Bear Lake, Rocky Mountain National Park


Took advantage of co-operative weather this weekend to have a first go at a star trail panorama, at Bear Lake in Rocky Mountain National Park. I’ve visited this place many times: it’s beautiful, dark, and close to a huge parking lot that so I could retreat from the cold and wind while the cameras were firing away. The setup went as planned, though with two DSLRs mounted to the rail tripod stability was not great. I packed snow around the legs to try and cut down on vibration. For star trails dark frame subtraction is a no-no, so I set the exposure at what was intended to be a balance between depth and noise – 30s, f/3.5, ISO 1600. Once it got dark enough (for future reference that’s about an hour after sunset) I locked down remote releases and retired to the car for dinner! (Not without some irrational unease at abandoning all this gear in the backcountry.)

I ran for two and a half hours, with the Moon in the sky to illuminate the peaks for the last half hour. The result was a set of 280 images from each camera, which were processed in Lightroom and stacked with StarStaX as usual. The nearly full Moon proved to be pretty bright, so I used only about two hours of exposure. Here are the individual stacks.


Pointing ultra-wide lenses up toward the sky gives a lot more distortion that I’d realized, and despite the generous overlap Autopano Pro balked at stitching this automatically. The problem was mostly noise and a lack of clear features in the foreground, and the stitch improved dramatically after I added back the EXIF data (removed by StarStaX), lightened the foreground, and applied noise reduction before stacking. Even then the stars were a mess, but that’s easily fixed by adding manual control points to the sky.

I reckon more sky and less foreground would have been better, and even thought as much while setting up. But I wimped out seeing the unexpected distortion and fearing that my tripod might topple in the wind if pointed any further skyward. Happy with the result for a first effort though!

Stitching images shot at different focal lengths

Stitched Panorama

This is a test. This is a test… The boring panorama of my apartment building is a stitch of two frames, on the left a full-frame shot from a Canon 5D3 with an 18mm lens, on the right a crop sensor frame from a 7D at 10mm (16mm equivalent). The only point is that it works! My stitching software (Autopano Pro) combined two frames, taken on different cameras with different pixel scales using lenses of different focal lengths, with no problems. (It seems to interpolate to the smaller pixel scale, so I ended up with an enormous 66 MP image.) The panorama here has a horizontal field of view of about 120 degrees, but that was with a lot of overlap, and in theory with these two lenses you can get close to 180 degrees horizontally.

Normally of course you’d shoot a panorama like this with one camera. I’m testing using two cameras because I want to try shooting star trail panoramas, for which the left and right frames have to be shot simultaneously. Originally I feared this would need two identical cameras and lenses (which I don’t have), but at least for a scene like this the software can cope with different cameras and lenses without difficulty. A plate to mount two cameras to my tripod is on order, and once it arrives I’ll see how it works with the night sky…


Saw in 2014 with a long weekend in Rome. As you may have heard there’s quite a lot to see and do in Rome, so on this, my first visit, I was mostly occupied with the famous historical sites. Well known or not, the Colosseum, the Pantheon and the rest are incredible.


Glimpsing the Pantheon


Basilica of St Paul’s outside the walls


At the Colosseum

Stitched Panorama

The Colosseum, twilight

I shot everything with the Canon 5D3 and 24-70 f/2.8L lens, hand held, stitching where necessary. It’s a good combination for church interiors and night scenes. In practice, ISO 12,800 seems pretty clean and potentially useable for a good-sized print, while 25,600 is OK but a stop too far without serious work on noise reduction.

Crazy stacking experiments

What would it look like if you shot 250 frames over about 8 hours, starting at night and continuing until just after dawn, and then stacked them using star stacking software (which retains the brightness pixel across the stack)?


If you answered “it would look weird!” you’d be right. That’s the Moon you can see in the extreme upper right. However with the right scene – maybe ideally with no clouds – there could be potential for some interesting images with this technique.

Update: once one gets into the spirit, interesting images are not so hard to find. Here’s a stack made from the time lapse I shot in Rocky Mountain Park the other night. No night to dawn transition here, but the clouds rolling through result in some hallucinogenic sky effects…