[robertchow999]

Hmm, good point. If my typical Milky Way shot is a 30-second exposure, then it would take 2 minutes to capture all 4 exposures.

If the astrotrace and pixel shift could work together, you would have to make an 8 second exposure, which the K3II would repeat four times during the pixel shift operation. That may require some processing horsepower.

Also, combining multiple exposures with astrotrace and pixel shift would be phenomenal, if possible. I think adding an espresso machine function would be easier, at the cost of portability.

[Ian Stuart Forsyth]

To compensate earth's rotation by sensor movement the lens axis must be calculated against the rotation axis of the earth. Both are directional vectors.The lens axis is a position vector. How could a directional vector (azimut and inclination) be determined only by GPS? GPS delivers 3-D data only about the Position. To solve the necessary equations you need Information on where the lens is pointing to. To get the pointing angles (azimut and inclination) of the lens additional data is needed. As far as I know O-GPS1 determins azimut by inside magnetic compass and tilt/inclination by acceleration sensors in the camera. This is why a precise calibration of these sensors is necessary before every operation of the astrotracer.


For sure the measurement of magnetic field to get exact azimut (tolerance perhaps a few arcminutes) is a weak method in this case especially if the field is disrupted by adverse influences.

Once the position is solved the azimuth is calculated and you do not need to move, We on a daily basis derive our azimuth on a single point (stationary setup on a tripod) setup using C/A for engineering and land surveying all the time. If we go back to the same setup the following day and setup on the same point we will have to ensure that we use the same coordinate files but azimuth would be the same from day to day. If we tie into the same points on the second day we will find that all the bearings are with 0 degrees 00 minutes 15 seconds if we have something greater than that we have a problem.
Now when you are walking with GPS and you slow down it may appear that bearings to your next way point looks less accurate but that's not the case as GPS always has the correct bearings
GPS BEARING versus DIRECTION OF TRAVEL

---------- Post added 04-30-2015 at 08:52 PM ----------

Right, a GPS can’t calculate azimuth if It’s not moving.

GPS BEARING versus DIRECTION OF TRAVEL

[Gimbal]

Right, the GPS can calculate the bearing to a known point from its current position and give you that number, but it does not know in which direction you point the GPS. You can twist and turn it but all the GPS know is where it is, it does not know it's orientation, for that you need additional sensors.

If you move the GPS it can calculate the direction it is moving, but it still doesn't know how you hold it. You can hold it backwards, it won't know the difference.

[Pete_XL]

Once the position is solved the azimuth is calculated and you do not need to move, We on a daily basis derive our azimuth on a single point (stationary setup on a tripod) setup using C/A for engineering and land surveying all the time. If we go back to the same setup the following day and setup on the same point we will have to ensure that we use the same coordinate files but azimuth would be the same from day to day. If we tie into the same points on the second day we will find that all the bearings are with 0 degrees 00 minutes 15 seconds if we have something greater than that we have a problem.
Now when you are walking with GPS and you slow down it may appear that bearings to your next way point looks less accurate but that's not the case as GPS always has the correct bearings
GPS BEARING versus DIRECTION OF TRAVEL

---------- Post added 04-30-2015 at 08:52 PM ----------



GPS BEARING versus DIRECTION OF TRAVEL

I fear we get a bit off topic with our ´"expert discussion".....


I am an experienced geologist and understand of what you are talking about. But we are NOT having the question of "gps bearing versus direction of travel". With a C /A GPS System you take the gps coordinates determined by the system and can then calculate the bearing to the KNOWN coordinates of another point (or the system does it for you). It is a simple mathematical procedure to calculate bearing from two sets of known coordinates.


But we are here discussiong the situation of a tripod-mounted camera that only knows its OWN coordinates by O-GPS1 while its lens is pointing to an arbitrary point in the sky. The coordinates of that point are NOT known and can NOT be derrived from gps data. Also, from the sensors point of view, this point is rotating around the earths rotation axis. To do the mathematical calculation for the sensor movement you have to determine 3 Parameters of the POSITION VECTOR from camera to the aiming point of the lens and set it into mathematical relation to the rotation vector the earth to get the relative "moving vector" of the aiming point. The 3 Parameters are:1) 3-D coordinates of camera on earths sphere (by gps). 2) Azimut angle of lenses axis . 3) Inclination angle of lenses axis. The Parameters 2) and 3) can NOT be derrived from gps data. They must must be seperately measured by camera and/or O-GPS1 to be then computed in the Astrotracer software (measurement by magnetic sensor and by acceleration/tilt sensors).

[Stone G.]

As much as the accuracy of the O-GPS1/Astrotracer can be argued about, it appears to me that one factor has been overlooked in this dicussion about Astrotracer-Pixel Shift compatibility:

In the case of a camera mounted on a "perfect" equatorial mount, consecutive images of a given star field will also fall on "exactly" the same spot on the sensor. This is very different from the situation with Astrotracer, where the camera remains stationary (it does not rotate with the stars) and for each individual exposure, the sensor will always begin it's movement from the same plane-parallel position relative to the camera.

Thus, if your were to take a, say, 20 second exposure with a 16mm lens using Astrotracer and Pixel Shift simultaneously, in each of the four subexposures, a given star (at the celestial equator) will have changed its position on the sensor by 16*20*0,000073 milimeter = 0,023 milimeters, which is 6 K-3 pixel-widths from frame to frame or 24 pixels over the intire set of sub-exposures!

Will Pixel Shift work on a good equatorial mount then? In theory yes, but it will have to be a very stable and precise mount and you will have to work under stable atmospheric conditions (where the stars do not twinkle).

[Pete_XL]

Thus, if your were to take a, say, 20 second exposure with a 16mm lens using Astrotracer and Pixel Shift simultaneously, in each of the four subexposures, a given star (at the celestial equator) will have changed its position on the sensor by 16*20*0,000073 milimeter = 0,023 milimeters, which is 6 K-3 pixel-widths from frame to frame or 24 pixels over the intire set of sub-exposures!

You are completely right! That is the killer-argument against the combination astrotracer/pixel-shift......

[Gimbal]

You are completely right! That is the killer-argument against the combination astrotracer/pixel-shift......

Not really, the sensor position doesn't have to be reset between the 4 exposures. Instead the maximum (individual) exposure would have to be divided by 4. So while doing a 20sec exposure, they would divide the 20s into four 5s exposures. Using an electronic shutter and pixelshift every 5 sec without resetting the position.

So it is possible, sort of, the result though would probably not be that great with the added problem of the distortion from (some) lenses causing the stars to make trails even if everything is perfectly set up. But then sometimes it could work just great.

But I don't think they have messed with the astro tracer software at all. Instead they have just added the PS part as a function separated from all the rest. Just a guess though.

---

Last edited by Gimbal; 05-01-2015 at 05:50 AM.

[Pete_XL]

Not really, the sensor position doesn't have to be reset between the 4 exposures. Instead the maximum (individual) exposure would have to be divided by 4. So while doing a 20sec exposure, they would divide the 20s into four 5s exposures. Using an electronic shutter and pixelshift every 5 sec without resetting the position.

So it is possible, sort of, the result though would probably not be that great with the added problem of the distortion from (some) lenses causing the stars to make trails even if everything is perfectly set up. But then sometimes it could work just great.

But I don't think they have messed with the astro tracer software at all. Instead they have just added the PS part as a function separated from all the rest. Just a guess though.

If I am not using a wide lense I (like probably most of us) do everything to maximize the signal to noise ratio. The signal of deep sky objects like galaxies, nebulas and star clusters is very low with only few photons making the lightyear journey onto my sensor. I struggle a lot to maximize exposure time to catch all of them I can get in the given time pulling also ISO quite high. I shoot darks and use flats. Then everything is stacked to get the noise down again and optimize the signal.

For that I really cannot imagine how quartering the exposure time to get more resolution by pixel shift can bring an advantage to the final picture quality of astrophotos in real life application. I would always trade the additional resolution for additional exposure time and would in parallel work on best focus on long lenses by e.g. using a bahtinov mask.

I use a K5. The first of the following pictures of M51 (crops of Whirlpool Galaxy pictures) was the best I could get out of O-GPS1 on that object. More than 15 seconds of exposure with 200 mm lead to small startrails no matter what I tried. Therefore I did 111 pictures of 15s at F3.5 and ISO 3200. A few days later I bought the small and relatively cheap Skywatcher Star Adventurer mount. The second picture is a crop of my first try with it. It was made of only 20 x 118 s at ISO F4.5 and ISO 800. I could have extended to 180 seconds without trails but then clouds came up and I went to bed. 120 or even 180 seconds with a 200 mm lens and O-GPS in my case are far from what is possible. The comparison speaks for istself I think.

[Gimbal]

The question is how much you lose in quality by dividing, for example, a 20sec exposure into four 5s pixelshifted exposures that are then stacked together for the final result. The total exposure time is still 20sec.

Pixelshifting and your skywatcher would make an interesting combo though.

[CNunez]

We have to hope that the sensor shift is done rapidly. Otherwise it doesn't matter what speed we shoot at.

[Pete_XL]

The question is how much you lose in quality by dividing, for example, a 20sec exposure into four 5s pixelshifted exposures that are then stacked together for the final result. The total exposure time is still 20sec.

I agree if the luminance of the 4 x 5 s LRGB is added 4x to result in 20s of "light". If another algorythm is applied it could result in less exposure. This would be negative for astrophotography especially with long lenses aiming at deep space objects.

Pixelshifting and your skywatcher would make an interesting combo though.

Alltogether it would be a revolution in amateur (and perhaps expert) astrophotography if pixel shift could be combined with tracking systems guiding stand alone camera or teleskope/camera combinations.

We here say "Versuch macht klug" what should be the same as "learning by doing" in your language. A friend of mine owns the new Olympus O-MD E M5 Mark II which does pixel shift as the first camera in the world as far as I know. I will try to persuade him to borrow it for a test with pixel shift and Skywatcher when the moon is less disturbing than it is at the moment. I will report.

[ChrisA]

The more I read about the K-3II the less impressed I am with the new features. None of them appear to be designed to work together. Its like Ricoh just took a bunch of different technologies and put them together without integration.
.

Excellent joke, very funny.

Oh, wait...

[philbaum]

The New Superhero That Could: The Pentax K-3 II, How Nikon and Canon Are Lagging Behind, and Why It Doesn't Matter | Fstoppers

I apologize if someone already posted this link.

But this review takes Nikon and Canon to task for not doing innovative stuff like Pentax K3ii. And they mentioned specifically the GPS and pixel shifting exposures.

AT least this creates some buzz for Pentax instead of a popup flash, which many cameras already have. Go Pentax!!!

[Spodeworld]

I read it when it first came out. He applauds pentax, chides Nikon and canon, but says ultimately it won't shift the dial. At least, that is his view.

The New Superhero That Could: The Pentax K-3 II, How Nikon and Canon Are Lagging Behind, and Why It Doesn't Matter | Fstoppers

I apologize if someone already posted this link.

But this review takes Nikon and Canon to task for not doing innovative stuff like Pentax K3ii. And they mentioned specifically the GPS and pixel shifting exposures.

AT least this creates some buzz for Pentax instead of a popup flash, which many cameras already have. Go Pentax!!!

[philbaum]

I read it when it first came out. He applauds pentax, chides Nikon and canon, but says ultimately it won't shift the dial. At least, that is his view.

Yep, he's a gloomy prognosticator :-( But one can never be that sure about the future. Lots of new products die a quick death, and then occasionally a "GoPro" comes along and turns the marketplace into something different.

Hope this time it makes a difference.