Saturn, May 26, 2015 - Stack of 748 video frames

This image was very challenging for several reasons:

 

First, I used "eyepiece projection".

 

Eyepiece projection uses a tele-extender tube to connect the camera (video in this case) a few inches behind the eyepiece of the telescope.

 

The result is more magnification, but a much dimmer image.

---------------------------------------------------------

Then, I used Canon's 640x480 crop video mode

 

The 640x480 crop mode, produces a 1:1 pixel resolution, resulting in no information loss, but a much smaller canvas to record the image on. This video mode, combined with eyepiece projection, produced videos that had the following challenges:

 

The first problem was just to FIND Saturn in the camera's field of view. In the eyepiece, it was plainly visible - even basically centered. But once I screwed the tele-extender on, then hooked up the camera, most of the time I saw... nothing. Using a tele-extender, along with a 640x480 canvas, produces a seriously small field of view. Just the added weight of the camera is enough to mess up the alignment.

 

And when I was lucky enough to center it over the Camera's sensor, Saturn was so large, that I had to record it diagonally, just to fit it in the 640x480 frame. Just the slightest drift caused part of the planet to be off frame. Every little shake, every footstep, every slight breeze, caused enough shaking to ruin the image.

 

Also, Saturn was so dim, that I had to use a video ISO of 3200, just to be able to see the planet. This created an enormous amount of digital noise in the videos. And even then, the videos were still dim.

------------------------------------------------------------

The next step was to run the video into Registax.

 

Registax is a beautiful program... when it works. It takes all of the individual frames of a video, and aligns and stacks them on top of each other. The random camera noise ends up cancelling itself out (since it's random), and the real information accumulates, resulting in a huge improvement in the signal to noise ratio. The end result is a final image that is far superior to any individual frame.

 

At least that's the way it's supposed to work. Problem number one is, Registax won't open most videos. It has a limited number of formats that it supports, and even then, the codec has to be supported. Personally, I've had about a 95% video failure rate.

 

I've had better luck using individual image frames, extracted from videos, and saved as .pngs or .bmps. But even then, if the planet is located towards the top of some frames, and towards the bottom of others, or left and right.. basically if there's any substantial drift in the planet's location, then the program fails. Since my polar alignment technique is to basically point the tripod to where I think north is, I usually end up with a lot of drift.

 

However, I've had 100% success rate (so far) using images where the planet was centered in every frame. The problem there is, it then becomes my responsibility to align 100's of individual video frames (over 700 for this image) onto a larger background frame, so it's not so claustrophobic, and then save them again as .bmps, and then run them into Registax. Doing 25 images at a time, was taking me about 25 minutes of time. Aligning 25... 50... 75... images was tedious. Aligning 100... 200... 300... was torture. So then I thought...

--------------------------------------------------------------

Why not make my own program to handle this chore?

 

Using Visual C++, I've always managed to find a way to create any type of program that I've ever attempted. Mostly little games and gadgets and doohickies - some of them pretty cool if I do say so myself.

 

So I thought, How hard can it be to create a program that:

 

1. Creates an 800 x 600 bitmap.

2. Accepts drag and drop images.

3. Loads those images into RAM.

4. Determines the brightness and color of the background of each frame.

5. Fills the background of the new image with that color.

6. Locates the planet in the old image.

7. Finds the geographical center of the planet.

8. Moves the old image out of RAM, and onto the new canvas, centering the planet.

9. Saves the resulting image as a .bmp in the same folder as the original, using the same name, but with a few characters appended onto the end.

10. Does all this automatically, with no input from the user.

 

I whipped it out in 2 nights and it works BEAUTIFULLY. At first I dragged and dropped a single video frame onto my new program (which I call AstroStax). Immediately, it generated a new, larger image - with the exact matching background of the dropped image, and with the planet perfectly centered. So then I tentatively dragged 4 images at once, and again it worked. 4 new images popped up in 1 second.

 

So, since I had about 220 images, still remaining to align, I figured I might as well go for it. If it crashes - it crashes.

 

It didn't crash. I dragged 220 images, all at once, onto my program, and in about TEN SECONDS, it spit out 220 larger images, all with the same background as the original, all with the planet perfectly centered. Every single image was how I wanted it.

 

Aligning and saving 220 images manually the way I had been doing it, would have taken me 3 1/2 hrs of tedium, and over a thousand clicks of the mouse.

 

Now that's been replaced by:

Drag

Let go

Wait 10 seconds.

 

I have hours of video that I've taken over the years, of Mars, Jupiter and Saturn that I've never quite gotten around to processing because I never had the time. Now I'm looking forward to seeing what I can get out of them.

 

Who knows, maybe I'll attempt to upgrade AstoStax to the point where I won't need Registax at all. It would be a lot of work, but I think I could do it.

---------------------------------------------------------------

Done