Help in image processing

[orasheed]

A Newbie question,

I am having difficulty in stacking pictures taken through a manual lens, the PI seems to stack them differently than the ones taken through an AutoFocus lens. Leading to a blurry picture.

Is there a process flow that can be used to stack them properly. These are planetary pictures, and there may be some clouds in some of them.

A few pointers will be helpful.

Thanks,
M

 

[nico1038]

Hello @orasheed

Your message surprised me for 2 reasons:
Planetary images are generally taken with high focal length telescopes and not with photo lenses.
And, while Pixinsight has no problem stacking planetary images, it doesn't have a tool to align them (as far as I know).

So can you be more specific about the equipment you used (the focal length of your lens and the camera used) and about your worflow to align and stack your images in Pixinsight?
Maybe you can also post a picture so we can see what it is?

 

[orasheed]

Hi,

Thanks for your reply. I am using a canon DSLR with a catadioptric lens 500mm F/8 to take the images. Plus using a star tracker mount. Quite a simple setup.

Workflow was quite simple. Take raw images and run the FFTRegistration script since the planet was in the same position in every frame. I got about 20 images.

I was able to get the images stacked when I was using a 200mm lens, with no problems.

Thanks,

 

[nico1038]

Hi,

Thanks for your reply. I am using a canon DSLR with a catadioptric lens 500mm F/8 to take the images. Plus using a star tracker mount. Quite a simple setup.

Workflow was quite simple. Take raw images and run the FFTRegistration script since the planet was in the same position in every frame. I got about 20 images.

I was able to get the images stacked when I was using a 200mm lens, with no problems.

Thanks,

Ok I see, It's unusual material for this kind of images but there's no reason why it shouldn't work. However, with this sampling you should not expect high resolution.
I'm not familiar with the FFTRegistration script but if the images are well aligned then maybe the problem is with the focus of the individual frames (It can't be easy to focus with this kind of lens)?

Can you maybe share an individiual frame and the stacked result?

And, by the way, what planet are we talking about?

 

[orasheed]

Ok I see, It's unusual material for this kind of images but there's no reason why it shouldn't work. However, with this sampling you should not expect high resolution.
I'm not familiar with the FFTRegistration script but if the images are well aligned then maybe the problem is with the focus of the individual frames (It can't be easy to focus with this kind of lens)?

Can you maybe share an individiual frame and the stacked result?

And, by the way, what planet are we talking about?

Hi Once again,

All 20 of them were quite large attachment. I am sending 3 of them along with the stacked image (background removed). As well as another image taken through 200mm lens and stacked. both lenses were manually focussed.


Thanks,

 

[fredvanner]

Small JPG images just don't have the resolution for this sort of analysis. You need to upload the full resolution images to a cloud server site (e.g. Google Drive; DropBox; MS OneDrive) and post a link to this forum; the link must be shareable.
Could you post two or three frames that are failing to align / stack. A similar two or three frames from a sucessful stack would be useful for comparison.
I have never had much success with FFTAlignment - mainly because it is not good at handling atmospheric distortion ("seeing" shimmer). For lunar imaging I have used Autostakkert!3 successfully. Running the aligned images through Blink often shows up this sort of problem - the outlines are aligned but the body of the image wavers and shimmers.

 

[orasheed]

Hi All,

As an update to my quest for stacking the images. I made the following steps and achieved a good result.

Goal was to stack the solar images in a way that will yield a harmonious image.

1. Analyse the images - I did a visual analysis using the STF function, although I would have preferred to use a tool (batchstatistics didn't give me a meaningful result) that would tell me how much is the signal to noise ratio in a specified area. I suspected that the haze (fuzziness) by atmospheric conditions (thin layer of clouds) was creating dispersed light leading to confusion with (Pix, Autostakkert and Lynkeos).

2. Use the DBE function to remove the background information (in this case it was foreground) and I thought of using this function as ABE with standard settings was not able to get any meaningful information on the background. I had to do that for all the images with same selection points for DBE. For this I loaded the stack in Blink and went one by one for each image. It was time consuming process but I wanted to see the end results. Saved the results as XSIF format.

3. Next step was to stack them
3a. - FFTRegistration wasn't doing the job and as fredvanner (thankfully) suggested that its a poor choice for this and I concur.
3b. - Used Autostakkert (Under Wine bottle for Mac platform) next and the images weren't aligning properly, maybe due to selection of Align points or due to the background (foreground in this case) still having too much dispersion of light for it to give a pin point alignment. At least it was better than FFTRegistration but still far from what I wanted. Had to convert the images to FIT format as autostakkert doesn't read XSIF. Maybe a loss of information in the conversion as well.
3c. - Lynkeos (Mac based)- this provided the razor sharp alignment of the images of the Sun and Stack was very good.

4. Next I used the Unsharp, Histogram and Curve adjustment functions to arrive at the final picture. Which shows the sun spots in great number (much more than the images I shared earlier with just a yellow orb) and also some variations in the corona of the sun. these variations are visually perceptible, albeit they do not provide any details. However it was pretty good for the equipment that I was using (mentioned above).

Just sharing this as a process for my quest. Comments will be welcome as we as alternate ideas of processing those images in a much more refined manner.

My two cents: Would really be helpful if a process pipeline is built with the steps taken to arrive at final image. So that it is repeatable and predictable.

Thanks,
M