I propose a python script to extend (extrapolate) images. The image is extended outside of its boundaries with noise having statistical properties equivalent to the noise in the image, and the boundary between the actual image and the noise extension is feathered for a smooth, invisible transition. There is no limit on how much the image can be extended. This script is a better alternative to padding with PI Noise Generator, as that module offers very limited control over noise statistical properties and does not implement feathering at the boundaries between the image and the padding. There are 2 motivations: - when doing a narrow band mosaic, one of the channels may have signal only in a few tiles of the mosaic. Then instead of acquiring this channel over all the tiles, one can acquire only those tiles with signal, and pad the empty tiles with the script.
- when doing 2D Fourier processing in Pixinsight, some edge effect (Gibbs phenomenon) can affect the boundaries. With the script it is possible to extend the image before FFT, and trim back to the initial support after IFFT, pushing the artefacts to the zone to be trimmed.
The script: https://github.com/erellaz/Fit_extrapolator/tree/mainExample of padding for the first use case:  Real life example: acquire 6 tiles of Ha, 6 tiles of SII, but only 2 tiles of OIII, only in those area where OIII is present. This allowed to triple the exposure of the OIII area and go much, much deeper. Then extended the OIII area by 4 tiles to have the same support as Ha and SII, and combine the channels.  |
You cannot like this item. Reason: "ANONYMOUS".
You cannot remove your like from this item.
Editing a post is only allowed within 24 hours after creating it.
You cannot Like this post because the topic is closed.
Copy the URL below to share a direct link to this post.
This post cannot be edited using the classic forums editor.
To edit this post, please enable the "New forums experience" in your settings.
Where is the line drawn?
I honestly have no idea but I'm interested in the discussion.
|
You cannot like this item. Reason: "ANONYMOUS".
You cannot remove your like from this item.
Editing a post is only allowed within 24 hours after creating it.
You cannot Like this post because the topic is closed.
Copy the URL below to share a direct link to this post.
This post cannot be edited using the classic forums editor.
To edit this post, please enable the "New forums experience" in your settings.
Have you tried any of the inpainting methods available in opencv?
|
You cannot like this item. Reason: "ANONYMOUS".
You cannot remove your like from this item.
Editing a post is only allowed within 24 hours after creating it.
You cannot Like this post because the topic is closed.
Copy the URL below to share a direct link to this post.
This post cannot be edited using the classic forums editor.
To edit this post, please enable the "New forums experience" in your settings.
Die Launische Diva:
Have you tried any of the inpainting methods available in opencv? Yes, Inpaint in Opencv is next in my list . I suspect, short of generative AI, diffusion algorithms are the way to go for this, to make the boundary even more realistic. cv2.inpaint, however, seems to be operating in 8 bits only , and that has stopped me for now. I would have to either get it to work in 32 bits, or implement the PDE directly, or upscale the 8 bit sampling to 32 bits.
|
You cannot like this item. Reason: "ANONYMOUS".
You cannot remove your like from this item.
Editing a post is only allowed within 24 hours after creating it.
You cannot Like this post because the topic is closed.
Copy the URL below to share a direct link to this post.
This post cannot be edited using the classic forums editor.
To edit this post, please enable the "New forums experience" in your settings.
My opinion, are you a scientist trying to get the most accurate photos? Yes, then take the time to do it properly.
Are you some Joe Schmo that is trying to take pretty pictures? Then do what you need to to fill in the other panels as long as you're not adding some sort of structures in the image that aren't actually there.
|
You cannot like this item. Reason: "ANONYMOUS".
You cannot remove your like from this item.
Editing a post is only allowed within 24 hours after creating it.
You cannot Like this post because the topic is closed.
Copy the URL below to share a direct link to this post.
This post cannot be edited using the classic forums editor.
To edit this post, please enable the "New forums experience" in your settings.
It is a cost opportunity problem: by imaging empty tiles, you implicitly forgo more integration time on the object.
In this example, with 6 tiles, by skipping the 2 bottom and 2 top tiles and focusing on the 2 central tiles only, the integration time on those is increased 300%, and the SNR on the object increased 173%.
With this technic, the imager now has more freedom to choose how to allocate his resources: where they make a visible improvement to the image, or where they don't.
|
You cannot like this item. Reason: "ANONYMOUS".
You cannot remove your like from this item.
Editing a post is only allowed within 24 hours after creating it.
You cannot Like this post because the topic is closed.
Copy the URL below to share a direct link to this post.
This post cannot be edited using the classic forums editor.
To edit this post, please enable the "New forums experience" in your settings.
Since all we want are pretty pictures, I say don’t bother with any of this and just get some AI software to make the picture as pretty as you like.
|
You cannot like this item. Reason: "ANONYMOUS".
You cannot remove your like from this item.
Editing a post is only allowed within 24 hours after creating it.
You cannot Like this post because the topic is closed.
Copy the URL below to share a direct link to this post.
This post cannot be edited using the classic forums editor.
To edit this post, please enable the "New forums experience" in your settings.
Just do whatever to feel like (except breaking actual laws). I don't understand why people must approve how you exercise your hobby…
|
You cannot like this item. Reason: "ANONYMOUS".
You cannot remove your like from this item.
Editing a post is only allowed within 24 hours after creating it.
You cannot Like this post because the topic is closed.
Copy the URL below to share a direct link to this post.
This post cannot be edited using the classic forums editor.
To edit this post, please enable the "New forums experience" in your settings.
Arun H:
Since all we want are pretty pictures, I say don’t bother with any of this and just get some AI software to make the picture as pretty as you like. I dont think adding OIII noise to unfuck the white balance is worthy of your sarcasm.
|
You cannot like this item. Reason: "ANONYMOUS".
You cannot remove your like from this item.
Editing a post is only allowed within 24 hours after creating it.
You cannot Like this post because the topic is closed.
Copy the URL below to share a direct link to this post.
This post cannot be edited using the classic forums editor.
To edit this post, please enable the "New forums experience" in your settings.
Thanks for sharing this tool.
|
You cannot like this item. Reason: "ANONYMOUS".
You cannot remove your like from this item.
Editing a post is only allowed within 24 hours after creating it.
You cannot Like this post because the topic is closed.
Copy the URL below to share a direct link to this post.
This post cannot be edited using the classic forums editor.
To edit this post, please enable the "New forums experience" in your settings.
