I had to process 967 sub frames for this! 

Glad it helped :)

Glad it helped!

Glad it helped :)

Thank you! :)

Thank you :)

Have you seen my new video yet? You can download the tool now and upload larger files :)

@@deepskydetail Not yet, will do.

Apologies, when I commented it didn't look like the comment went through so I resent it.

Thanks! So right now, the tool can only accept 600kb images (which is pretty small). They also have to be black and white with no alpha channel. That's the limitation of the free web server I'm using. Hopefully in the next week I'll have a standalone app that can be downloaded and you can upload any size image (as long as you have enough ram). It will still be limited to monochrome images (probably).

@@deepskydetail Can't wait!

It was over a couple nights if I recall correctly.

That is a good question. You can make AI tools that create new data. I think some denoising tools can do this because they are too general. Take an AI tool that was trained on cat pictures, and you might start getting whiskers in your astro images. The thing about this tool, is that the training data is very constrained. Meaning, during training, the input is a blurred set up pixels, and the output is a sharpened one. After training, learning stops. The neural network is now just one mathematical formula, and the functions will be applied to any image in the same way. It's like a deconvolution matrix, but implemented in a way that takes into account patterns of blurring. But you don't have to just take my word for it. If you like, you can try it on your own astro images (if you have them). Compare the original and the new image and look for things that might be off. Then, let me know what you find! If you think it works, great! If you think it's creating things that aren't there, that's still great if you tell me about it :) I can then try to improve things!

I just realized you might be also talking about the processing in general (not just the sharpening tool I used). I think this deserves a really detailed answer. I might make a video on it sometime. But the short(ish) answer is this: 1) Digital cameras measure photons linearly. Meaning, a pixel that is 2x brighter than the other probably had 2x the number of photons hit it. 2) Our eyes do not see linearly. Our brain automatically adjusts for brightness. Sometimes in pictures, you'll see that part of the image is blown out even though the rest of the image is fine. You can see both areas in a way that isn't blown out. 3) This is why we need to stretch the image. It helps bring out the fainter details that (if our eyes were sensitive enough) would look kind of normal 4) Many of the techniques used are indeed subjective (adding high pass filters; color calibration techniques etc). But that is true of other photography too. 5) I don't think most astrophotographers' goal is to produce a scientifically accurate image. Rather it's to produce a pleasing image that brings out the underlying structure of what they are looking at. So, do I think it is computer generated art? Not really. Not like what DALL-E is doing. The picture has real data in it, and the processing helps us see those data.

haha! It is :)

You're right. Thanks for watching.

That's a lot of files :)

@@deepskydetail and the result was not great i think single long exposure 2-5min is better

@@superearth5256 Yeah, I think that if the exposure is too short, then there just isn't enough signal to pull out detail. What kind of setup were you using if you don't mind me asking?