RAW failure

I like Aperture so far, mostly.  I’ve even managed to memorise a significant fraction of its obscure keyboard shortcuts and combos.  Enough to get around it and make it appear I have some idea what I’m doing.

Unfortunately, Aperture relies on Mac OS X’s built-in RAW support (as does every other imaging app, which is handy in some respects – everything from Preview to TextEdit can work with RAW files), and that support is really quite buggy.

For example:

There’s nothing wrong with the file itself.  Nikon ViewNX 2 and other RAW readers can display it just fine.  Only Mac OS X has problems.

Something like 1 in 20 of my images come out this way.  It’s very frustrating.  Generally I’ve been taking multiple shots of things, so thus far I haven’t had it ruin an image without substitute, but it’s only a matter of time.

A radar has been filed, of course, though I get quite depressed when I think about the odds and time frame involved in seeing it actually get fixed.

Edit: Well shit, now that I look at it again I’ve realised that Nikon’s software shows the exact same corruption; it’s just ViewNX 2 shows the preview by default, which I knew, but I thought I had successfully told it to show the RAW instead.  Apparently not – when I went to change the exposure, then it really showed the RAW.  Damnit.  I hope the RAWs aren’t truly corrupt.

Edit: It looks like they are.  #%@!  On the upside, though, embedded in every NEF file is a couple of JPEGs, including one that’s full resolution (albeit not as sharp as the real raw data).  Thanks to exiftool, I’m able to extra the JPEGs, so the images aren’t lost entirely.

Update: It ultimately turned out that this is the result of a single toggled bit in the entire file.  It appears to be a communication problem between the SD card and my iMac (and happens with two SD cards, albeit both SanDisks), as ejecting the card, physically, and re-inserting it usually fixes the problem.  I’ve now collected thirty or so examples of this problem, both the corrupt and good versions of the RAWs.  I had aspirations of writing a tool to detect the corruption (by comparing the various embedded previews and the RAW render itself) and ideally fix it (by doing some kind of educated binary search), but… I never got around to it.  Sad panda.

Stars are moving really fast

While we were camping I wanted to try some star photography.  I had borrowed Chris’s tripod specially for the occasion, so I was all excited and set to go.  I hadn’t initially thought of doing any astronomical work with the camera, but I chanced upon a discussion of it, with some example images from old Digital Rebels.  Very impressive images, for the cameras involved.  Just stars – nowhere near the resolving power to show things like planets or whatnot in any detail – but cool nonetheless.

So I set up my tripod and picked a random, likely piece of sky, set zoom to max (300mm), and went for max exposure time (30s), so that I could use ISO 100 and keep the noise down.

And I get this [full scale excerpt of a larger frame]:

Well shit.  Something vibrated.  I had thought ahead enough to use a delayed shutter, so I could take my hands completely off the camera and give it enough time for the vibrations to die down.  Was that not enough?

After a couple of tries with identical results, I was starting to get a bit annoyed.  Was it the shutter itself that was moving the camera?  Sounds crazy, but it wasn’t the most stable tripod ever.  I tried lowering the centre column, tried putting weight on it, shifted the tripod legs, etc.  No good.  Same streaks.

I tried shooting a different altitude.  The streaks changed length and direction.  Ah-ha!  The vibration depends on the camera orientation.  Well, that’s a clue, innit?

Long story short, yes, but my suspicions were totally off-base.  It was Marisa that pointed out the embarrassingly obvious – the stars move.  A quick back of the hand calculation put their motion at up to a ninth of a degree, over a 30 second exposure.  With 300mms of zoom, that’d surely be more than enough to explain the motion.  And why changing the target portion of the sky changed the direction and magnitude.

Then Marisa added further to my intellectual smack-down by showing me how to find the North star, for which I otherwise had no means to find, and thus have a pretty good idea where to look for a stable reference point.  And the end result was this:

Not perfect, granted – looks like a stubby little shuttlecock – but clearly we had identified the problem.  Hah.  I also verified the hypothesis by cranking the ISO way up and doing half-second exposures of previously streaky stars, yielding now much shorter streaks, if any.

FYI assuming 86,400 seconds in a day, the distance travelled by a star on the celestial equator would be exactly 0.125 degrees.  So my 1/9th estimate was pretty bloody good, if I may say so myself, for being done in my head by the side of our symbolic campfire.  Symbolic because there was a park-wide ban on wood/charcoal fires.  So we gathered around our propane lantern.  Just as romantic, only not really.

Also, my camera cannot auto-focus on stars, so I had to resort to manual focus.  I don’t know how close I got to optimal, but the lens reported a focal distance of ~40 metres, so that’s not too bad.  Unfortunately as far as I’m aware there’s no way for me to see the focal distance except long after the fact, in Aperture via the embedded metadata.


For a few months now I’ve been eyeing off DSLRs.  I’m not sure, now, what triggered it… I think just a couple of events, like Jen & Vante’s wedding, where I wanted for a better camera than my little Sony DSC-TX10.  While it’s a great little camera, and was totally worth having for Hawaii alone, it is of course a happy-snapper, and it just doesn’t deal with things like low-light very well.

So I researched and talked with various people and even played with a bunch of Nikons, Canons and Panasonics, along with Steve & Marisa.  I really liked the full-size bodies of the Nikon D7000 and its ilk – they’re the only ones which are big enough to fit my hands comfortably.  The smaller bodies of the entry level Canon & Nikons felt okay, but pretty cramped.  The Panasonic – a DMC-G3, if memory serves – was of course tiny by comparison and even more awkward to hold, though there was some appeal to its diminutive form factor, just for the portability aspect.  I was a bit leery of it, though – I’d never heard of the micro four thirds system, and though the store clerk was singing its virtues, in the end I steered clear of it, if only because my primary goal was good low-light performance, to which end an even smaller sensor made little sense.

After weeks of this, I’d decided that my favourite was the D800.  Of course, $3,300 just for the body seemed just a little extravagant, by my standards anyway.  I tossed and turned on it, and even convinced myself briefly that I would get it – at which point I appealed to Marisa to talk me out of it, but she was quite fond of the idea and encouraged me instead.  Gah!  But in the end I decided it would be wiser to take a smaller first step.

So I went with the D3200.  It doesn’t have quite the features I originally wanted – I kind of wanted a hypothetical D5200, i.e. an updated D5100 – but I didn’t want to wait some indeterminate length of time for that to come out.

So D3200 it is.  I ordered it from B&H Foto, along with Nikon’s common 55-300 (bundled with it at $150 off).  I successfully resisted the urge to go too crazy with lenses, though I’ve since been looking at wide-aperture portrait lenses quite a bit…

Anyway, it’s been a lot of fun to play with.  I put the manual on my iPad and read it cover to cover before the box even arrived – stupid slow-arse 3-day shipping – so now I actually understand what all those modes do beyond “Auto”. 😉

I also bought Aperture, because it seemed apropos.  I was surprised to discover that it shares the photo library with iPhoto – that’s a nice safety net.  Unfortunately the RAW support for the D3200 on Mac OS X, even with yesterday’s update, is buggy – it sometimes renders the lower half of photos as garbage.  It also interprets some photos very badly – totally mangles the colours.  It seems to like cranking the temperature way up (among other things)… I haven’t yet determined if that’s based on the metadata in the photo or its attempts to “auto-optimise” the photo.  After a few days I became pretty good at correcting its mistakes, but it is tedious to have to fiddle with a good half of my photos.  And it probably bugs Marisa to no end that I keep dragging her over to get her opinion on every other photo. 🙂

FYI Nikon’s bundled software, ViewNX 2, displays the photos perfectly (and more sharply to boot).  Just to make it clear that this is a Mac OS X bug.  I filed a radar.

But I digress.

Thus far I’ve taken it with me almost everywhere Marisa & I have gone, from walks around our neighbourhood to weekend gatherings with friends.  Photos slowly percolate their way out of Aperture and into Flickr, as per usual.  I’ve been burning the midnight oil to get them all up this week, but I am finally caught up.  ‘course, this weekend I’ll probably take a thousand more… le sigh… anyone wanna be my post-processing minion? 😉