[jbinpg]

Just so we are all on the same page, the EXIF value for compass heading is given in degrees True, not magnetic. I assume there is a lookup table onboard that does the necessary conversion to True from magnetic before writing the EXIF based on the acquired lat/long.

[phoebus]

Mag Var in Southern England is currently charted at about 1 deg west, so no great difference between mag and true... ...but yes - it would not work in general without some table or algorithm to resolve the magnetic flux into a true heading.

[peterh337]

That's interesting too. Down here the mag variation is small but there are parts of the world where a compass is barely usable. The correction can be done by a polynomial (which works only over most of the globe, not anywhere near the poles) or a lookup table.

I will re-test with say 70mm or longer.

Incidentally, how do telescopes get around this? Presumably they have two degrees of motion: the azimuth and the pitch (declination or whatever it is called in astronomy).

[photoptimist]

That's interesting too. Down here the mag variation is small but there are parts of the world where a compass is barely usable. The correction can be done by a polynomial (which works only over most of the globe, not anywhere near the poles) or a lookup table.

I will re-test with say 70mm or longer.

Incidentally, how do telescopes get around this? Presumably they have two degrees of motion: the azimuth and the pitch (declination or whatever it is called in astronomy).

Telescopes solve this by moving both the lens and sensor together to track the motion of the stars. Equatorial telescope mounts feature one rotation axis that is carefully aligned to the axis of the Earth and then they use a small clock motor to cancel out the rotation of the Earth.

It's also worth noting that this differential motion issue drops with increasing focal length. A few quickie-calculations suggest that at focal lengths of greater than maybe 135mm to 200 mm, the Astrotracer system will hit its motion limits before differential trailing becomes noticeable (the exact threshold focal length depends on some details of astrotracer's algorithm and the user's definition of "noticeable").

It's also worth noting that this differential motion issue is worst when the camera pointing angle is perpendicular to the north star (e.g., looking up and due south). The maths for exactly how much differential star trailing will occur is quite complicated. And when one adds in the effects of errors in your particular calibration in your particular location plus barrel/pincushion distortion of your particular lens, the best advice is to experiment!

Good luck and we look forward to your pictures.

[peterh337]

" Equatorial telescope mounts feature one rotation axis that is carefully aligned to the axis of the Earth and then they use a small clock motor to cancel out the rotation of the Earth."

Would those mounts not have the same issue with wide angle lenses as the K1's system?

[clackers]

" Equatorial telescope mounts feature one rotation axis that is carefully aligned to the axis of the Earth and then they use a small clock motor to cancel out the rotation of the Earth."

Would those mounts not have the same issue with wide angle lenses as the K1's system?

But those astronomical telescopes have long focal lengths, surely? They would rely on stitching multiple shots for any wide pics.

[photoptimist]

" Equatorial telescope mounts feature one rotation axis that is carefully aligned to the axis of the Earth and then they use a small clock motor to cancel out the rotation of the Earth."

Would those mounts not have the same issue with wide angle lenses as the K1's system?

No. The mount's motion is designed to perfectly cancel the rotation of the Earth within the limits of the quality of the motion system's drive and the constancy of the drive motor's speed (good mounts & drives cost money!). You can put any camera and any lens on a good equatorial mount and the stars will remain perfectly fixed on the sensor the entire night unless the star is very near the horizon and atmospheric refraction distorts its apparent location.

In fact, if you left the mount and camera running 24 hours a day, 365 days a year, the stars would never shift or even rotate although they would disappear when the moving mount was pointing the camera below the horizon or when the sun was also in the sky with those stars. You'd also notice a tiny wobble in the locations of some stars that are nearby and are affected by stellar parallax. The mount removes the rotation of the Earth about it's axis and around the Sun but does not remove the looping circular displacement of the camera around the axis of Earth and Sun.

[peterh337]

I have just done some 160mm shots at 2 mins iso100 f8 and they are not good



I will try tomorrow to calibrate the compass yet again...

[photoptimist]

Is there any chance that your tripod has metal/magnetic materials that are disturbing the compass reading? If you put the K-1 in compass display mode and bring it near the tripod, does the reading change?

[Gimbal]

I have just done some 160mm shots at 2 mins iso100 f8 and they are not good



I will try tomorrow to calibrate the compass yet again...

I think you have to settle for less time at 160mm, 30sec is hard enough. It’s better to get a shorter but reliably time and then take several shots and stack them.
So if one shot gets ruined for some reason you haven’t lost that much time.

[peterh337]

The tripod (a £500 Gitzo one) doesn't have any effect but I am getting weird results. Basically the camera seems to calibrate and then it indicates the heading correctly, and shortly afterwards, after some movement, it is 10-20 degrees off. I can't get any consistency however.

I did find a way to cause a 10 degree change: move away from the house. The heading changes over about 2 metres, even when one is 2 metres away from it already.

I will re-test tonight...

For many, this feature will not be usable due to the proximity of houses and any steel in them. For example I fly a light aircraft which has a fluxgate (very very carefully calibrated in an open field, with with all enroute electrical loads selected) magnetometer driving a gyro stabilised slaved compass system and just taxiing along a taxiway, I get the magnetic compass swing by 10 degrees just because there is a hangar 100 metres away!!

There should be a manually inputtable azimuth value.

I bet you anything that the azimuth of a proper telescope derives from a surveyed value and not from some little electronic compass!

In the pic above which shows the huge trails, the heading was about 20 degrees off. But if it was just 2 degrees off, you would still get unacceptable trails. But there is no way you will ever get this type of compass more accurate than that. An aviation fluxgate sensor can be calibrated to something better than 1 degree but it weighs 0.5kg and costs $1000.

[Gimbal]

As already mentioned (or maybe it wasn’t?) in some directions just moving the LCD screen at the back of the camera will affect the compass reading quite a bit. Fold it out and get one heading, tuck it in and get another.

To address this compass problem I made a “device” that attaches to the flash shoe, it’s a rotatable knob with two small magnets inside. By rotating this knob I can set whatever compass heading I like.

Due to weather problems I haven’t tested it more than briefly, but it turns out that things move quite fast and the heading has to be updated often. And of course you have to know the correct heading to set.

But it was interesting to see that adjusting the heading had a large impact on the result.

Maybe it would be better to have a couple of larger magnets attached to the tripod? Hm…

[peterh337]

"To address this compass problem I made a “device” that attaches to the flash shoe, it’s a rotatable knob with two small magnets inside. By rotating this knob I can set whatever compass heading I like."

What a brilliant solution!

That is exactly how you calibrate aviation compasses. The vertical ones have two screws, one for N-S and one for E-W - e.g. here: http://www.peter2000.co.uk/aviation/lowz/panel-2014-900px.jpg

I ordered some tiny fridge magnets from Ebay and will use them to "fix" the camera to read right. That will save a lot of time.

[Sagitta]

Few observations:

The camera is pointing in X degrees *magnetic* north, not polar. The difference is likely the difference between the two.

Regarding star rotation/trailing, the wider the lens, the less you'll have every star perfect due to the fact that the stars will spin at a different rate the further you are away from Polaris. I can't recall the scientific term for this off hand, but if you think of looking down the guts of an umbrella and twirling it, the center won't have to move nearly as much as the outer edges.

This will show in your photos, and the only option is to reduce shutter speed to minimize the effect if you don't like it.

I shot my house last summer on a nice dark night, and you can see it in the stars on the edges of the frame. I imagine on a full frame the effect would be even greater.

EDIT: This was almost due east, FWIW, north would be to the left off frame.

DOUBLE EDIT: You can see the effect with the scenery as well - house seems sharp, trees on the sides definitely not so.

K-3, Sigma 10-20mm f/3.5 @ f/3.5, 60 seconds ISO1600

[peterh337]

I am getting much better results now, 20m away from the house, though far from great

70mm:



1:1



This one came out well in terms of sensor movement (but is out of focus)

133mm:



1:1



Why the difference I don't know. The last one shows a heading of 198 which looks about right



The data on the ones which showed the movement did not show any significant difference.

100 seconds F8 (70-300 Tamron) in all cases, on a tripod and with a radio trigger.