Embedded Files
Animation:
Animation:
Animated GIF of the movement of the Lunar terminator over a six hour period. Images were taken every 15 seconds (1100 total) from 00:30 to 06:30 on 5 March 2010 at the Broemmelsiek Park observatory in St Charles County Missouri (http://www.asemonline.org/broemmelsiek-park-astronomy-site/broemmelsiek-observatory).
26 of those images were used for this animation test.
Click on the image at right to display in a full page (Depending on your net speed, It may start out slow and not run smoothly). This is a 4meg file and is best viewed when downloaded and displayed locally. This image is available here: File Cabinet.
Sunset
A smaller, 2meg version is to the right. It has been cropped to the upper right quandrant.
"Sunrise"
Considering that the original image set was taken during a waning moon (setting sun), reversing the image sequence (running the movie in reverse) gives you an idea what it's like to see terrain "appear" during a rising sun.
The sequence was taken using a Canon A590 connected afocally to a TMB 9mm eyepiece on the observatories 10" LX200 Schmidt (on loan to the Parks department by James Roe). Images were taken every 15 seconds. 26 were selected for this initial animation test. Finder and comparison images follow below.
All charts were created via Virtual Moon Atlas (http://www.ap-i.net/avl/en/start)
Finders:
Finders:
field of view. From this, the remaining features of interest can be identified.
A section of the chart was scaled to fit the full size image. The chart section was used to provide a higher accuracy registration test. Registration tests were made in both the N/S and E/W axis. This was done by first scaling the image such that features in the N/S axis lined up in both images. The error in the perpendicular axis was noted then the image was rescaled for the E/W axis. In this case, the error in the East/West axis was smaller than the error in the North/South axis. In both cases, the errors in the perpendicular axis were on the order of 10percent. The difference between errors was less than two percent. This image shows the results of scaling to fit the North/South axis.
I'm not sure if these alignment differences are due to camera to eyepiece alignment, charting error or projection errors.
Equipment:
Equipment:
Comparison fields and registration validation:
Comparison fields and registration validation:
The rotated image was rescaled to match a new
This is an image taken at 02:10:26 Local Civil time. Three features were identified and labled.
The image was rotated, flipped and scaled to fit a full face image. Then layered to provide placement reference.
What next
What next
The process to create the 26image animation was a bit tedious to say the least. Several steps were needed for each of the images:
Image selection: Select the sharpest image for the specific period. Significant amounts of blurring was caused by seeing but also blooming caused by traffic to the South of the observatory.
Image drift: Re centering of each image on a common point was required to remove frame to frame "jitter". This jitter is
caused by the fact that perfect tracking of the Moon could not be accomplished. Even though the mount is exceedingly well aligned to the Earths' axis, the Moon does not move at a sidereal rate. In fact, there are two independent motions with respect to the sky (and this adds up to a significant deviation over this six hour period). The Moons movement in its' orbit with respect to the rotation of the Earth. The East/West motion was closely taken into account by setting the controller tracking frequency to 56hz. Even then, a small amount of error occurs in this direction over a matter of minutes. A North/South component is caused by the inclination of the Moons orbit. Manual corrections are made about every 5 minutes or so and that causes each image to be just slightlly off with respect to the others. When stitched together these differences apear as image jitter. This jitter is reduced by individually re-registering each image to a common point in the image. In this case, it was the peak of the shadow as indicated at right. Sharpening: The images were sharpened just a tad to improve the overall quality of the animation.
Resizing: To maintain some reasonableness in processing time, run time and resulting file size, each image was resized from the original 3264x2448 pixels to 800x600 pixels. Even then, 26 images resolves to a final 4meg file size.
Each image took approximately 5 minutes to process. Not something I want to do across the entire 1100 image set.
Once the images were processed, they were imported into JASC animation shop. The animation only took several minutes to complete at this stage.
I need to find a program that will allow me to automate the registration process for all 1100 images. Once that happens, a smoother animation can be made. Some kind of method to automate the sharpening process would also be nice but that is secondary to the registration process.
All in all, I'm pretty happy with the equipment, the image acquisition process and the initial test image. Since the present technique is so tedious, I probably won't get back to this specific image set unless I can't automate it - and then that will be to evaluate and refine the automation process.
Page updated
Report abuse