The Cold Pixels

Turns out physics is (more and more) setting performance limits on digital cameras. (Title is a reference to the controversial Tom Godwin story “The Cold Equations”).

I’ve been maintaining two camera systems for a while now—Nikon, and Micro Four Thirds (my bodies have all been Olympus). That developed sort-of accidentally; I got an E-PL2 to replace a Panasonic LX3 as my “point and shoot”. But of course, over time, lenses accumulated, and it started to be a significant camera for more than just snapshots. And when I upgraded to an OM-D EM-5 it was pretty good at video, too (most of my work on the Cats Laughing reunion concert at Minicon 50 was done with it). And after the body upgrade it became my main camera, except for sports action (almost all roller derby) and nasty low-light (often music in bars).

Having both systems leads to having the wrong one, or carrying both, on trips. And to ongoing expenses. And a certain level of duplicate lenses.  And to having only a single body on either side.

And, just recently, the OM-D EM-5 body has packed it in (“beyond service life” according to Olympus service). This kind of brings the question of just what the heck I should do to a head. Without spending too much money, of course.

The Nikon gear (both what I have, and the models I might buy) is an old-school DSLR. Viewing is optically, through the taking lens, via a pentaprism and a moving mirror, which means I have to line my eye up with the lens to see anything (crawling on the ground or whatever as necessary), and that in low-light situations it can be hard to see.  It also means manual focus is hard, since the view can be dim and the focusing screen is not optimized for manual focus.  Auto-focus is by phase-detect sensors in the bottom of the body, fed by semi-transparent spots on the moving mirror plus pivoting sub-mirrors on the back (how does that ever work?).

The Micro Four Thirds gear is of the more modern “mirrorless” design. There’s no big mirror flapping around to make noise and cause shake. Older models including my dead one use contrast-detect auto-focus based on the data from the main image sensor; some more recent ones like the OM-D EM-1 mark II and the Sony A7R II integrate phase-detection AF sensors on the main sensor, making AF much faster (phase-detect sensors tell it which way to adjust, contrast detection does not, among other issues).

Watching the development of cameras over the last decade, I think I see that we’re past the time for silly flappy mirrors. I’m starting to feel about them a little like the way Heinlein described internal-combustion engines in The Rolling Stones:

 The prime mover for such a juggernaut might have rested in one’s lap; the rest of the mad assembly consisted of afterthoughts intended to correct the uncorrectable, to repair the original basic mistake in design—for automobiles and even the early aeroplanes were “powered” (if one may call it that) by “reciprocating engines.”

A reciprocating engine was a collection of miniature heat engines using (in a basically inefficient cycle) a small percentage of an exothermic chemical reaction, a reaction which was started and stopped every split second. Much of the heat was intentionally thrown away into a “water jacket” or “cooling system,” then wasted into the atmosphere through a heat exchanger.

What little was left caused blocks of metal to thump foolishly back-and-forth (hence the name “reciprocating”) and thence through a linkage to cause a shaft and flywheel to spin around. The flywheel (believe it if you can) had no gyroscopic function; it was used to store kinetic energy in a futile attempt to cover up the sins of reciprocation. The shaft at long last caused wheels to turn and thereby propelled this pile of junk over the countryside.


Apart from the risk of simply being completely wrong, I’m wondering if it’s too early to go all-in on mirrorless designs. The main players are Olympus and Panasonic (in the Micro Four Thirds collaboration) and Fuji. Canon and Nikon have minor lines of no particular market or technological significance (though rumor has it that Nikon is coming out with a full-frame mirrorless line next year), Leica makes a few full-frame models at Leica prices, and Sony has some very interesting full-frame models (but the lens lines for them are limited and pretty expensive).

Oh, and there are some medium-format mirrorless models now, from Leica, Hasselblad, and Fuji at least; those aren’t oriented towards sports-level action, and are getting into five figures instead of mid 4 in price, so they’re not anything I should or can think about.

Mirrorless cameras are very easy to adapt to use any old-style SLR lenses (and many others), in strictly manual mode. Because they don’t have to have space for a mirror to flip up, the back of the lens mount flange is very close to the sensor; and when making an adapter that’s one of the inescapable limits (the other is the coverage of the lens being adapted). And, with electronic viewfinders and modern technology like “focus peaking” and just magnifying the image, manual focus is vastly more usable than it was with SLRs. I do still need AF for fast-moving subjects, though, especially sports, and it’s convenient some of the rest of the time. However, my current mirrorless camera wasn’t modern enough to have focus peaking, so I don’t have more than passing direct experience with it (playing with other people’s cameras).

While I lived with film for decades, I now casually expect modern levels of low-light sensitivity out of my equipment (and I am, after all, mostly competing with people who either have it, or don’t want it).  Micro Four Thirds uses a sensor half the area of a “full-frame” sensor. The Fujis and the Canon are APS-C.  The bigger sensors will always capture more photons per pixel at any given resolution (each pixel is simply physically bigger).  And while they’ll all get better at capturing, and at processing the data they capture, they’ll all be reasonably in step on those improvements. Bigger will always win.

Here are the DXOMARK stats on some of the cameras I have or am considering:

DXOMARK numbers

My old Nikon D700 has a “sports” score (which is basically high ISO quality) of 2303. The fancy new OM-D E-M1 Mark II has a score of…1312; not that much better than half as good. And the Sony A7R II, one of the very top low-light cameras,  scores 3434. In the nearly 10 years since my D700 was released, sensors and processing haven’t  improved enough for any smaller sensor to catch up with it, but sensors the same size have moved well past it. So the latest fancy Micro Four Thirds body would be a considerable step backwards in something I care about (possibly mitigated by a lens a stop faster from 200mm to 300mm, see below).

Just to confirm the DXO methodology, here are lab test shots of some of the choices from

DPReview high-ISO examples

(Note that the Nikon D750 is considerably more recent than my D700; but they didn’t have anything that old in the database, not even the D3).

That’s kind of interesting, in that the Sony doesn’t look as good as its rating would suggest.  The E-M1 Mark II does look somewhat better than the E-M5. And of course the D750 beats them all, but that’s a modern full-frame DSLR.

Going solely to any mirrorless system requires buying some high-end lenses, too; at least their 70-200mm f/2.8 equivalent. And there, Micro Four Thirds wins big; the Olympus equivalent (40-150/2.8 Pro) has equivalent angle of view of an 80-300, which is enough longer to be very nice (a stop faster anywhere beyond 200mm equivalent than anything I have now), and costs “only” $1400. The only choice for the Sony (no Sigma or Tokina models available) costs $2600.

Oh, and there are some fairly significant Olympus rebates and trade-in deals for the next couple of weeks.

Decisions, decisions!

Annoying Aspects of Modernity

My primary camera came back as “beyond service life” and hence unrepairable.  Even the guy at the camera store was surprised, he had to look up when it was released (it was announced in February of 2012).

Olympus OM-D EM-5 “beyond service life”

Meanwhile, I could easily get a Leica M3 (made around the same time I as) or a Nikon F (about 5 years newer than the Leica) repaired. Of course those two are special cases, they’re both regarded as important classics. And, being old-school completely mechanical cameras, the parts they need can be manufactured pretty easily today, without any help from the original manufacturers.

This was my only video camera and my primary still camera (though the D700 still does a much better job on roller derby and in dark bars and music circles).

I’ve been playing around in my head with where to take the camera collection from here, given that both sides (this Micro Four Thirds body and the lenses for it, and my Nikon D700 and those lenses) are getting old by modern standards (the D700 is even older, having been released in July of 2008; I’ve already had one autofocus system repair). This is rather financially constrained, among other things.  (Some of the Nikon lenses I’m using I bought in 1981, and they still work fine, and still could be repaired though perhaps not by Nikon themselves.)

I think it’s time to abandon flappy mirrors; they’re a silly idea in digital cameras. However, full-frame sensors do seriously better in low light than smaller ones (no smaller sensor has yet matched the specs of my 2008 D700 full-frame sensor), and DSLRs have better auto-focus for tracking and fast action than any mirrorless (except possibly maybe the hugely expensive top-of-the-line Sony A9, which doesn’t take any of my lenses). And there aren’t many full-frame mirrorless lines; there’s the Sony, and a Leica (which makes the Sony look cheap).  But the state of the art in sensors and electronics is advancing constantly; while no smaller sensor has caught up to my D700 yet, they’re close, and no doubt will catch up soon. Of course today’s full-frame sensors are five generations (or some such) better, and still well ahead, but at some point something becomes “good enough” and it’s not worth paying hugely for small improvements for most kinds of photography. Nikon is allegedly about to release their own mirrorless full-frame system, but how well it will work with old Nikon lenses is anybodies guess. For that matter how well it will work at all is still up in the air.

With financial constraints, concentrating back into one system is nearly certainly the way to go, and for cost and flexibility the Micro Four Thirds seems to be the best choice starting from where I am.

Have to think about it; I wonder what I’ll actually do—and when?

Dave Romm has died

Kind of a shock; nobody seems to have known of any reason to expect any such thing any time soon. A phone call had reached Sharon and Richard (who were hosting the Minn-StF meeting yesterday) just 5 minutes before Lydy and I walked in.

The information that’s reached me so far is that he probably died quickly of a heart attack in his home about September 4th, but wasn’t found until the 14th (hence the lack of certainty on date and cause of death).

So it was a bit of a subdued meeting with a lot of reminiscence.

Dave, who most of the years I knew him insisted on the capital “E” in DavE (but mostly dropped that after becoming a Baron of Ladonia), moved to Minneapolis quite shortly after I became a permanent resident here, I think in 1978. He was already a photographer and interested in the history of fandom, so we had that in common, and he was an accomplished bridge player and this was during Minn-StF’s bridge period. He moved into the apartment I vacated in the basement of the Bozo Bus Building when I bought Finagle’s Freehold.

I will say that, Facebook profile to the contrary, he did not live in Memphis Tennessee.

Oh, about that Barony—DavE described himself as being “a real baron of a fake country.” The people running Ladonia appear to have properly granted him that honor in line with their historic traditions (which go back to 1996, it looks like).

Here are a few of my photos of DavE over the decades:

Near-Total Eclipse of the Eclipse

We drove down into Iowa Sunday, to put us in striking range of a variety of viewing sites in the totality zone on Monday.

Even Sunday, we encountered traffic and crowding on Interstate 35; we had to wait for a table at a Culver’s down in Iowa.  We did seem to be a bit ahead of the crowd, since our hotel’s parking lot was nearly empty when we got there (or maybe the people on I35 had planned ahead further than us and had hotels down in the totality zone).

Weather forecasts looked fairly dismal. We poked at them a lot, and on Monday morning we chose Falls City Nebraska as our viewing site.

The route there was all back roads, and we encountered no particular traffic. However, the town was quite active, with clear eclipse events at various places, and many optimists hoping to sell $20 parking. We ended up in the parking lot of an empty fast food facility, with a good view of the clouds overhead. A number of other group showed up there shortly.

We can see the sun!

There were thinner spots in the clouds, and occasionally you could tell where the sun was; or even see the disk (filtered by the clouds; at this stage, with the eclipse protective glasses you could see nothing at all, you couldn’t even tell where the sun was behind the clouds).

This raised a technical point I hadn’t really planned for. With the eclipse glasses, or the solar filter over the lens on the camera, you could see nothing at all. Sometimes the clouds were heavy enough we couldn’t tell where the sun was, but when they thinned and you got hints of the sun, sometimes a view of the full disk, you still couldn’t see anything through the filters. This made it difficult to get the camera trained and ready to capture anything that might briefly show through the clouds.  It also left me bouncing back and forth between filters and no filters, trying not to expose my eyes or even the camera sensors to the bare sun, but needing to use bare eyes and sensors to find it much of the time. (There don’t appear to be any weird blind spots in my eyes today, I seem to have been careful enough or lucky enough.)

A bit after first contact, shot through a solar filter (and the clouds).

The clouds thinned out a few times, and we got views near first contact, and occasional views thereafter.

We got some very heavy-duty insectile assistance while waiting for the eclipse to start. Looks like a very big wasp, and we saw it entering and leaving a crack in the pavement, so I guess ground-dwelling. Carrying something even bigger than it was; food for itself or for the next generation I guess.


The light seemed to dim in definite steps as we neared totality, not continuously. It was a weird effect, and repeated a number of times. People watching seemed to agree on when a step had happened.

It did get very dark. We had something like a sunset (except that the sun wasn’t there) all around the horizon; it was dark overhead, but light on all sides, with the sun out there clearly lighting up the clouds. Not sure one would see this effect without the clouds; though just the diffusion of light passing through the air might be enough to give a similar effect.

Half a hair short of totality, through clouds but without solar filter (nothing at all was visible through a filter).

And we got a very brief view in early totality through the clouds, where you could see a bit of corona. I got one photo then, too.

We stop to check out the reappearance of the sun when the clouds suddenly clear after totality.

We hung around to watch the sun come back, and got to see a bit for a while, until the clouds got serious.  Then we headed out—and the clouds cleared and we got bright sun a few minutes later (clouds had been solid to the horizon before) and we got some good clear views of the late stages (I didn’t take photos of the sun then).

Again, nothing much in the way of traffic delays on the back roads, but Interstate 35 was jammed. We had lines for the bathrooms at every stop, and visibly heavy traffic (and much heavier than the southbound traffic) all the way back up to the Twin Cities.  The trip back was over 3 hours longer than it should have been, due to these traffic delays.

Very close to sunset, we did run into a very nice double rainbow.

And got home by 1:30.