Forgot Password
Pentax Camera Forums Home
 

Reply
Show Printable Version Search this Thread
08-07-2016, 11:37 AM   #16
Loyal Site Supporter




Join Date: Jun 2009
Location: Tumbleweed, Arizona
Photos: Gallery | Albums
Posts: 5,001
I don't disagree that wide angle lens distortion is involved and is a contributing component. Also, that the wider the lens, the more extreme the distortion becomes, is a factor at play here as well. In terms of distortion within an image from a non tracked camera, I went looking and found this example. It is from a full frame body with a 14mm lens, at 23 sec exposure. For a location - mid to upper latitudes I would guess from the trees. [Note - for the sake of consistency, I'm trying to use a full frame body with a 14mm lens as a basis on this post. The 14mm lens on a full frame body has an FOV of 114 x 81 degrees.]Here is another example - full frame, 14mm @ 30 seconds straight from the camera - it's the 4th image down. There is also another example - at 60 seconds, 14mm on the 7th image down.To continue with a set of examples, here is a tracked (iOptron SkyTracker) full frame camera body using a 14mm lens, with a stacked set of results. Looking close you can start to see the stars trailing along the edges/corners.Extending the exposure time to 3 minutes, on a tracking mount, still using a full frame body along with a 14mm lens we have. [Note - to be fair in the upper corners, the sky/stars do appear to be brighter and probably pulled to an extent. The image is a result of a stack of 4 at 3 minutes each.]So, where I am going with this is to wind up with the K1, using a 14mm lens at 3 minutes with Astrotracing. The K1 has the new 5 axis sensor mount (if it's used to its fullest extent - I don't know). Over in the Forum's Astrophotography area is the first K1 GPS tracked image that I have come across. It is interesting in several respects.The image is a full 3 minutes. The stars are tracking outwards on each of the 4 corners - which is somewhat unique when compared to other tracking solutions. The star trailing appears to be somewhat accelerated when compared to the images above. I admit that there is not a single consistent baseline of 3 min, full frame sensor, 14mm with the same position on the earth and pointing angle across all of the examples.

Having said all of that, this thread has touched on - either directly or indirectly a number of possible reasons for the trailing. We have so far:
  • Sidereal - which is essentially a timing error. Some of the physical trackers take in to account either full or partial sidereal day timing. If Pentax takes this into account is not known.
  • Direction and shapes of the star movements over time- especially if you are taking in a large swath across and between both poles and everything in between. https://cdn.photographylife.com/wp-content/uploads/2015/03/Vortex-star-trail...t-of-Maine.jpg Image #3/6 If this were the sole cause, then we would have the star trailing more aligned with the sets of direction and shapes and not shooting out to the 4 corners from the center.
  • Wide angle lens distortion - This tends to be mitigated by the type of tracking system and the tracking duration to a large degree. Physical trackers are not showing large distortions. That said, they do rotate the camera body in terms of tracking.
  • Full body rotation - This certainly is a factor on the physical trackers. That said, the K1's 5 axis sensor movement, should be able to supply a similar movement to some degree - but certainly constrained due space constraints - which would be a time constraint.
  • Pointing error(s) - a combination of compass and elevation measurement errors
In summary, we have the..
  • Pentax GPS tracking - a black box solution. Very little is known about the internal implementation and constraints - other than the time limitations based on focal lengths. In terms of the K1 we don't know if the sensor tracks rotation or not, and if so to what degree (and with what limitations). It will certainly be limited when compared to a physical tracker since there are hard physical limits on how far the sensor will be able to be moved/rotated. We do know that by limiting the exposure time we can substantially decrease the star trailing.
  • Physical trackers - a white box solution - where quite a bit is known about the system implementation and its limitations. These tend to mitigate a lot of the wide angle lens star trailing that the Pentax GPS trackers see (either using the original SR implementation (k5, etc.) or the newer 5 axis implementation (K1)).
  • Just a camera - no trackers - When you capture just star trails, there is a consistency to their movement.
It appears that the Pentax GPS Astrotracing does a reasonable job, when the exposure time is reasonably minimized (however it is several times longer than just using a camera exposure time without the aid of trackers). Using the rule of 200, you can capture 10-13 second exposures with out tracking and trailing, while capturing 60 second exposures using 18mm lenses with out trailing. Also, you can mitigate the problem by moving away from ultra wide angle lenses. 14mm (and wider) through about 18+mm appear to amplify the problem, while lenses in the high 20's to mid 30's mm tend to mitigate the problem. Also, stitching and stacking tends to mask the problem when wider glass is used. That is the good news.

On the downside, when the Astrotracing is allowed to expose for longer durations - it exhibits the star trailing in corners and along the extreme edges. It certainly appears that this is due to a host of items that are cumulative in nature. There does not appear to be a single main item that can be isolated.

When comparing the physical trackers to the Pentax GPS tracker implementations - it appears that the physical trackers tend to calibrate out their tracking errors with their polar alignment which appears to be stable (does not drift). The Pentax GPS tracker appears to depend on GPS location and an electronic compass and elevation angle measurement, for its equivalent to a polar alignment. This pointing measurement may 1) not be completely accurate and 2) may drift over time. Don't know - since very little is known about the Pentax GPS implementation - in particular the compass. We just have the resulting images to work with.



08-07-2016, 12:55 PM   #17
Pentaxian




Join Date: Apr 2007
Location: Sweden
Posts: 1,887
QuoteOriginally posted by interested_observer Quote
That said, the K1's 5 axis sensor movement, should be able to supply a similar movement to some degree - but certainly constrained due space constraints - which would be a time constraint.
The added 2 axis in the 5 axis SR doesn't add any new capabilities to the actual image sensor movement.
The image sensor can still only move up/down and left/right, and by combining the movements up on the left side and down on the right side (for example) it can rotate slightly. That's is exactly the same capabilities as in for example the K-5. The added 2 axis doesn't require any new ways for the sensor to move, it's a couple of new accelerometers added in to the mix to detect up/down and left/right movements. Up until now Pentax SR has only reacted to angular movements using gyros.

Thus the 5axis SR will have no impact on the astro tracking function. On the other hand the whole SR mechanism is new and might have precision improvements that will help, but the basic functions are the same as in earlier models.


QuoteOriginally posted by interested_observer Quote
When comparing the physical trackers to the Pentax GPS tracker implementations - it appears that the physical trackers tend to calibrate out their tracking errors with their polar alignment which appears to be stable (does not drift). The Pentax GPS tracker appears to depend on GPS location and an electronic compass and elevation angle measurement, for its equivalent to a polar alignment. This pointing measurement may 1) not be completely accurate and 2) may drift over time. Don't know - since very little is known about the Pentax GPS implementation - in particular the compass. We just have the resulting images to work with.

Physical trackers don't calibrate anything, they just rotate at a fix speed. Or rather they try to rotate at a fix speed, but due to mechanical problems they usually oscillate back and forth in speed. If you are shooting wide you will not notice this though. (I have built one myself a few years ago. A microcontroller controlled the motor which had a feedback sensor and thus the motor kept perfect speed. Still the output after reduction gears oscillated slowly due to imperfections in the final worm gear.)

The actual GPS is good enough, a few meters of error doesn't mean squat.
The compass (of the OGPS-1) is what you actually try to calibrate, and it's pretty hard to get it prefect. At best I can get it right to +-1-2 degrees in most directions, but there is almost always some direction where error the will be bigger. Recalibrate and there error moves to some other direction. Then when you start to tilt the camera body the compass sort of loses it's bearings some what. a 3D compass is kind of complicated.

Also the angular measurments are calibrated (or so I believe) as they are normal accelerometers sensing the earth gravity.

But when everything works and the stars are perfectly tracked (in the middle of the frame) then the distortion of the wide rectilinear lenses screw up the edges no matter what.
08-08-2016, 07:09 AM   #18
Loyal Site Supporter




Join Date: Jun 2009
Location: Tumbleweed, Arizona
Photos: Gallery | Albums
Posts: 5,001
QuoteOriginally posted by Gimbal Quote
The added 2 axis in the 5 axis SR doesn't add any new capabilities to the actual image sensor movement.
The image sensor can still only move up/down and left/right, and by combining the movements up on the left side and down on the right side (for example) it can rotate slightly. That's is exactly the same capabilities as in for example the K-5. The added 2 axis doesn't require any new ways for the sensor to move, it's a couple of new accelerometers added in to the mix to detect up/down and left/right movements. Up until now Pentax SR has only reacted to angular movements using gyros.

Thus the 5axis SR will have no impact on the astro tracking function. On the other hand the whole SR mechanism is new and might have precision improvements that will help, but the basic functions are the same as in earlier models.
Apparently I missed the design paper that Pentax put out on their use of the 5 axis design. Yaw and Pitch are certainly available (within physical limitations) in their design - as to if it's used or not - I don't believe that Pentax has been forthcoming with K1 implementation details.
QuoteOriginally posted by Gimbal Quote
Physical trackers don't calibrate anything, they just rotate at a fix speed. Or rather they try to rotate at a fix speed, but due to mechanical problems they usually oscillate back and forth in speed. If you are shooting wide you will not notice this though. (I have built one myself a few years ago. A microcontroller controlled the motor which had a feedback sensor and thus the motor kept perfect speed. Still the output after reduction gears oscillated slowly due to imperfections in the final worm gear.)

The actual GPS is good enough, a few meters of error doesn't mean squat.
The compass (of the OGPS-1) is what you actually try to calibrate, and it's pretty hard to get it prefect. At best I can get it right to +-1-2 degrees in most directions, but there is almost always some direction where error the will be bigger. Recalibrate and there error moves to some other direction. Then when you start to tilt the camera body the compass sort of loses it's bearings some what. a 3D compass is kind of complicated.

Also the angular measurments are calibrated (or so I believe) as they are normal accelerometers sensing the earth gravity.

But when everything works and the stars are perfectly tracked (in the middle of the frame) then the distortion of the wide rectilinear lenses screw up the edges no matter what.
Poor or bad polar alignment with physical trackers is analogous to poor or bad calibration in Pentax's GPS / compass based design. The results from physical trackers are extremely good. Pentax's design and implementation as good as it is, the results are still trailing.

In terms of electronic compasses, I believe that is what I was pointing out initially. Initial position error with GPS is essentially negligible and the position is static during use, so that the main source of error in Pentax's design is essentially the 3D compass (direction and elevation). Designing such systems for the consumer electronics market is extremely difficult. You have very tight space, weight, power constraints along with very difficult cost restrictions, which directly impacts your product's end results.

I'v been involved in the design and use of GPS since the mid 70's, along with some laser ring gyro work. Designed the first GPS based navigation system for suborbital vehicles, and orbit insertion systems. I also did a somewhat unique star tracking system over 20 years ago, for a little telescope - that some folks still use out in Texas.


Last edited by interested_observer; 08-08-2016 at 07:23 AM.
08-08-2016, 08:11 AM   #19
Pentaxian




Join Date: Apr 2007
Location: Sweden
Posts: 1,887
QuoteOriginally posted by interested_observer Quote
Apparently I missed the design paper that Pentax put out on their use of the 5 axis design. Yaw and Pitch are certainly available (within physical limitations) in their design - as to if it's used or not - I don't believe that Pentax has been forthcoming with K1 implementation details.

That Image of the body with the five arrows visualizes the five axis of movement that the SR compensates for, not how the actual image sensor moves to do it. The sensor itself does not move in any new ways compared to the K-5. It is still left/right, up/down and rotate.

The SR mechanism is well known from earlier models, and images of the new mechanism are available on the net. There are no new axels of movement. Angular rotation left/right has been compensated for since the K10d days by shifting the image sensor left/right. The two new axels they added are for when shifting the entire body left/right, and to compensate for that they also shift the image sensor left/right.

So the image sensor movement mechanism is basically the same as in earlier models, although reworked to accommodate a FF sensor.

Reply

Bookmarks
  • Submit Thread to Facebook Facebook
  • Submit Thread to Twitter Twitter
  • Submit Thread to Digg Digg
Tags - Make this thread easier to find by adding keywords to it!
astrotracer, camera, center, distortion, dslr, edge, frame, full frame, full-frame, image, k-1, k1, lens, light, paul, pentax, pentax k-1, shots, viewfinder
Thread Tools Search this Thread
Search this Thread:

Advanced Search


Similar Threads
Thread Thread Starter Forum Replies Last Post
Trailing Stars wkraus Monthly Photo Contests 12 10-11-2015 05:36 PM
Budget flash with trailing sync ability for K-500? PeaceFroggy Flashes, Lighting, and Studio 14 12-03-2014 07:06 AM
Macro I'll be seeing you...seeing you...seeing you.. eaglem Post Your Photos! 10 07-01-2013 03:31 PM
Seeing what you are aiming at on the display? slaktratten Photographic Technique 4 04-11-2008 12:57 PM
Are You Familiar With This One? solar1 Pentax DSLR Discussion 3 05-28-2007 02:47 AM



All times are GMT -7. The time now is 03:28 PM. | See also: NikonForums.com, part of our network of photo forums!
  • Red (Default)
  • Green
  • Gray
  • Dark
  • Dark Yellow
  • Dark Blue
  • Old Red
  • Old Green
  • Old Gray
  • Dial-Up Style
Hello! It's great to see you back on the forum! Have you considered joining the community?
register
Creating a FREE ACCOUNT takes under a minute, removes ads, and lets you post! [Dismiss]
Top