The first time I saw the O-GPS1, I thought:
"They really got all of that into something the size of a matchbox."
Small.
Light.
And it runs on a single AAA cell.
That alone made it attractive as something to carry into the mountains.
But what really surprised me was what was inside.
Find your position with GPS.
Work out which way the camera is pointing.
Then calculate how the stars will move and, using the camera's shake-reduction mechanism, move the image sensor itself.
In other words:
track the stars from inside the camera.
That is what the O-GPS1 and Astrotracer, released by PENTAX in 2011, were.
Even looking at it as an engineer, I thought:
"How did anyone even think of this?"
Thinking of it may be possible.
But turning it into an actual product is another matter.
GPS, heading, tilt, focal length.
Use all of that to derive the motion of the stars and drive the SR unit.
And do it inside a small GPS unit and a camera body.
I have no idea who led the effort to make it happen, but
they cannot have been an entirely reasonable person.
That is a compliment, of course.
My first job with it was a pre-release O-GPS1
The first time I used an O-GPS1 was before it went on sale, I believe.
It was a job: borrow a prototype and actually photograph the stars with it.
At the time I was still using Nikon as well, and I would go on using it afterwards.
I never much felt that a different manufacturer meant you had to use one or the other exclusively.
A new tool appeared.
It might let me photograph something I could not photograph before.
In that case,
just go and use it.
I took the borrowed O-GPS1 to Yabitsu Pass.
Turning
To use Astrotracer with the O-GPS1, you need to run a precise calibration.
You hold the camera and rotate it, correcting the heading sensor and so on.
The manual explains how.
So I did exactly that.
Turn.
Fail.
Do it again.
Fail.
Change orientation.
Turn.
Fail again.
Still turning
I was at it for about ten minutes, I think.
It would not pass.
About thirty minutes went by.
Still would not pass.
Before long you start to feel:
"What exactly am I doing out here in the dark?"
In the darkness of Yabitsu Pass, a man alone with a camera, turning it round and round.
To anyone watching, it would look distinctly suspicious.
But I had to shoot, so I kept turning.
Check the rear screen.
No good.
Turn again.
I no longer remember exactly how long, but my recollection is that I wrestled with it for close to an hour.
And then, finally, it passed.
The stars really did stop
Once the precise calibration is done, you can use Astrotracer.
I tried it.
The stars were stationary.
Obvious as it sounds:
it really is tracking.
Combined with the K-5 of the time, it could track for up to five minutes depending on conditions.
I think I tried a five-minute exposure too.
The stars came out properly.
"I see. This is usable,"
is what I thought.
But at the same time:
this is going to need practice.
I knew the product worked.
But you cannot spend tens of minutes on a precise calibration every single time you are in the mountains.
There had to be a way.
I searched online
It was after I got home, I think.
I looked online for information about calibrating the O-GPS1.
And I found someone's blog.
Instead of simply rotating the camera about three axes,
move it as though drawing a figure eight.
That was the method described.
The moment I saw it, I thought:
"Ah — that's it."
Electronic compasses in devices like the iPhone use a similar calibration.
When I actually tried it, it passed far more easily.
A minute or two, I think.
What on earth had I been doing, endlessly turning that camera at Yabitsu Pass?
The Juliana Tokyo calibration
You wave the camera in a figure eight.
Watching that motion, I gave it a name of my own:
the "Juliana Tokyo calibration."
(Juliana Tokyo was a famous Tokyo nightclub in the early 1990s, remembered in Japan for dancers waving large folding fans.)
Obviously Juliana Tokyo was not a going concern in 2011 when the O-GPS1 was released.
That was long before.
But the motion looked like waving one of those fans.
That is all there is to it.
With this method, the precise calibration becomes far easier.
But as I kept using it, another thought came:
couldn't this be done with a much smaller motion?
The wrist is enough
I gradually made the movement smaller.
There is no need to draw a big figure eight with your whole arm.
Use your wrist and move it in the directions required.
Then smaller still.
Before long,
it passed with nothing but a flick of the wrist.
So this time I started calling it
the "wrist-flick calibration."
Ten seconds, twenty seconds — around that.
Thirty seconds at the outside, I think.
What had taken me nearly an hour at Yabitsu Pass now took a flick of the wrist.
This is not, of course, the official method the manufacturer printed in the manual.
It is a method I found, through use, that
"this gets it to pass."
Later, as I had more chances to speak at places like the PENTAX booth at CP+, I described this method many times.
I remember being asked by people at the manufacturer to tell that story.
A user calibrating the manufacturer's product by his own improvised method, and then explaining it at the manufacturer's own booth.
Come to think of it, that is a slightly odd situation.
It can track for five minutes
Astrotracer on the O-GPS1 had one big attraction.
You can make long exposures.
High-ISO performance in digital cameras was improving rapidly at the time, but raising ISO has consequences for image quality.
If you can track and expose for longer, you can lower the ISO.
Up to five minutes, depending on conditions.
That is significant.
But when you actually shoot nightscapes with it, another problem appears.
The stars stop. The ground does not.
The Earth is moving
Astrotracer follows the stars.
More precisely, it moves the sensor by exactly the amount the stars appear to move due to the Earth's rotation.
So the stars stop.
But the landscape on the ground is not actually moving relative to the camera.
And then you move the sensor.
Naturally,
the ground moves.
You are moving the ground in order to stop the stars.
So what do you do?
Simple.
Don't shoot for five minutes.
"Can shoot for five minutes" and "shoot for five minutes" are different
In actual nightscape work I ended up using around one to two minutes most of the time.
It can track for five minutes, and I stop at one.
That may sound wasteful.
It is not.
As a product,
"can track for up to five minutes,"
and as a photograph,
"five minutes is the best exposure,"
are entirely different statements.
At one minute the ground shifts very little.
The stars are stopped well enough.
You get the light you need.
So one minute is fine.
I later wrote in an article for Digital Camera Watch as well that when using Astrotracer for nightscapes, keeping to about one to two minutes is a major point.
Using a specification to its maximum is not the same as mastering a tool.
Stop where your photograph needs you to stop.
Isn't a portable star tracker better?
If all you want is to track the stars, a portable star tracker (equatorial mount) is an option.
I bought one later, too.
It is a dedicated device, so for tracking alone it is extremely capable.
So is the O-GPS1 unnecessary?
In my case, it did not work out that way.
If anything, I have barely used the portable star tracker.
I use it for things like solar eclipses.
But for mountain nightscapes there are situations where the O-GPS1 is overwhelmingly easier.
The reason is simple.
Because I move.
Iruka-iwa, again
On Mt. Tsubakuro there is a rock called Iruka-iwa — the dolphin rock.
It is also the place that, in the first story, led me by accident to the lighten composite.
Later I took a different photograph there.
One where the Milky Way
looks as though it is being blown out of the dolphin's mouth.
To take that, you do not simply set up the camera once and be done.
The Milky Way moves with time.
Its position relative to the rock changes.
When I think "a bit further right,"
I move.
Walk a few metres.
Set the tripod down.
Shoot.
The Milky Way moves again.
I move again.
Shoot.
And repeat.
You can do this with a portable star tracker too, of course.
But with a portable star tracker you have to re-align the polar axis every time you move the tripod.
With the O-GPS1,
you just pick up the camera and tripod and walk.
In the mountains, that difference is enormous.
Not maximum performance — usable performance
Compare tracking accuracy alone and the dedicated star tracker is surely better.
But what I want is not
the machine that tracks the stars most accurately.
What I want is
a tool that lets me take the photograph I want to take.
Light.
Small.
Something I can carry into the mountains.
Move a few metres and shoot straight away.
Stop the stars for as long as I need.
If it does that, it is enough for me.
This is the same as the release I wrote about before.
There are times when a three-terminal part stuffed into a battery box is a better tool than an elaborate CMOS circuit.
Performance as a machine and
performance as a tool for me are different things.
What you can do and what you should do are different
When a new technology appears, things that were impossible become possible.
With the O-GPS1, you could track the stars with nothing but a tripod.
That is a big change.
But when one constraint disappears, another becomes visible.
The stars trail.
So you track them.
And then
the ground trails.
So you shorten the tracking time.
And now you have to balance that against light.
With wide-angle lenses, stars towards the edges of the frame can trail as well.
So you change focal length or tracking time.
Technology does not remove all of the problems.
It moves where the problem is.
And the photographer thinks again, within those new conditions.
Just stop the Earth from turning
I want the stars stationary.
I want the ground stationary.
In that case:
just stop the Earth from turning.
That would be the simplest solution.
Which is, of course, impossible.
So the photographer decides where to compromise.
Prioritize the stars?
Prioritize the ground?
Shorten the tracking time?
Depending on the situation, there is also the idea of reducing the tracking amount and taking something in between.
Which is correct?
It depends on the photograph.
A tool doesn't decide the photograph for you
When I first saw the O-GPS1, I was amazed by the technology.
I still think it is remarkable that it was ever productized.
But what I learned by actually taking it into the mountains was the obvious thing:
excellent technology does not, by itself, produce an excellent photograph.
If the calibration is a nuisance, think about how to shorten it.
If it can track for five minutes but one minute is better, shoot for one.
If a dedicated star tracker is more accurate but you are moving around, choose the O-GPS1.
And in the end, decide where to stand yourself.
Standing in front of Iruka-iwa, when I think
"a bit further right,"
I am the one who moves right.
The tool does not decide that for you.
And that is fine.
Better, in fact.
As I wrote before:
a tool is not there to take photographs in place of a person.
It only needs to take on what a person cannot do, or what is tedious, so that the person can take the photograph they want.
For me, the O-GPS1 was that kind of tool.
And looking back, there is something curious about it.
The place that led me to the lighten composite was Iruka-iwa.
The place where I used Astrotracer to chase the composition I wanted was also Iruka-iwa.
I did not plan that.
But:
it was Iruka-iwa again.
That photograph still hangs in my bedroom.
Perhaps the reason for taking photographs is about that simple.
I want a photograph I would like to hang in my own room.
So I use tools.
If they fall short, I improvise.
If that still falls short, I build something.
And in the end, I take the photograph myself.
You cannot stop the Earth from turning.
But
what kind of photograph it becomes is yours to decide.
— To be continued