British Astronomical Association
Supporting amateur astronomers since 1890

Secondary menu

Main menu

BAA Community Member Pages Chris Hooker
Image search

Mercury observation: comparing frame selection techniques

Image

Expand image

About this observation
Observer
Chris Hooker
Time of observation
09/10/2018 - 12:40
Object
Mercury
Observing location
Didcot, Oxfordshire
Equipment
254 mm F/6.3 Newtonian
TeleVue 3x Barlow lens
ZWO ASI120MM-S camera
Baader 610 nm & IR-cut filters
Exposure
1 msec
Tags
Like This Image
Share

This post is a response to Martin Lewis's comment on my previous post. Rather than reply to the comment, I thought it might be of interest to respond in more detail in a second post. Martin specifically asked how I select frames for stacking to produce my images of Mercury, and I have described this below. The image shows a comparison between a version of the previous Mercury image obtained by manually-selecting frames and one obtained by using the same number of frames selected using Registax. To give the software a reasonable chance, I combined six registered AVI files from the first stage of software processing, choosing the six that gave the largest numbers of manually-selected frames. I then processed the resulting file in Registax, selected the same number of frames for stacking as in the original image, and used the same wavelet and histogram setting to process the stack.

In my view the difference is clear. Although the software-selected stack shows some of the same albedo features, it is noticeably more blurred than the original. Selecting frames manually is very time-consuming, but it does produce better results.

For those who may be interested, the technique I use is as follows:

I capture video segments with a standard length of 5000 frames, which Registax is able to deal with. If necessary (which it was for this set of data) I first examine all the video files and edit out the sections where Mercury disappears out of the frame due to the telescope shaking in the breeze. I use a video editing program (VirtualDub) for that, but there are other pieces of software such as PIPP which can also do the job.

The seond step is to use Registax to align and sort all the files. I prefer Registax for this because it offers a variety of quality estimators, and also allows me to save the best frames as a new AVI file. Autostakkert! doesn’t appear to have that capability, or else I’ve failed to find the command for it. Generally I save about 10% of the frames after alignment and optimisation; sometimes it’s more and sometimes less, but 10% (i.e. 500 frames) is typical. This 10% typically contains about three-quarters of what I would consider the good frames, mixed with several times their number of blurred and distorted ones. The discarded 90% inevitably contains some good frames, but the time required to find them would be out of all proportion to the contribution they would make to the final image.

The third step is the time-consuming one of going through all the files saved from Registax and selecting the best-quality individual frames, for which I again use VirtualDub. I have made several attempts to do this in software, but my experience is that with typical Mercury data, none of the alignment and stacking programs is able to produce images that are as good as those obtained after manual frame selection. Depending on the amount and quality of the data, this manual selection can take quite a few hours, which is why there is often several days’ delay between the observing session and my posting the image on this website! The image above required selecting the best of about 15,000 frames. I save the selected frames as new AVI files, then concatenate them for final alignment and processing in Registax. Initially I put them into two groups for separate processing as verification images, then combine them for the final image.

Comments

Fossil Light's picture

Thanks for the details Chris- got to admire your patience esp with 260,000 starting frames as on your previous post.

PIPP can use a number of different quality algorithms and will export an aligned avi. If you export from PIPP as an SER format avi then you can use SER Player also available from Chris Garry's PIPP webpage. That will allow you to go through frame by frame as you do in VDub and save individual frames. Think you might find it more user friendly than VDub.

Thanks for being so open about sharing your methodologies.

Best wishes

Martin

Copyright of all images and other observations submitted to the BAA remains with the owner of the work. Reproduction of the work by third-parties is expressly forbidden without the consent of the copyright holder. For more information, please contact the webmaster.