A quick Compare of the new BlurXterminator AI

[Jockinireland]

This is a quick and comparison of a deep crop of a raw (left) BXT AI2 (Middle) and BXT AI4 output from my NGC7331 image posted on the DSO forum.

Frankly, I think the result speaks for itself.

[Juno16]

Wow!

That is really amazing David!

Maybe I will dig up some of those old images with my old mount!

[SkyHiker]

That's like cheating! Now I have to try it, my images are all like that. Well not the fracs.

Question: Are you supposed to use blurX on raw, unstretched/linear, data, or any? I just heard from someone on the Losmandy forum that he tried it both ways and the new version works well either way.

[Jockinireland]

That's like cheating! Now I have to try it, my images are all like that. Well not the fracs.

Question: Are you supposed to use blurX on raw, unstretched/linear, data, or any? I just heard from someone on the Losmandy forum that he tried it both ways and the new version works well either way.

Yes there is already blowback from the "purists" that this is unacceptable as an AP tool. I tend to take that as so much sour grapes. For me, the whole post processing process requires huge manipulation of the data already. Think, background extraction, deconvolution, colour calibration, sharpening etc. BXT is just more of the same. It's not adding anything into the image, simply manipulating the data to present it as we want it to be seen. I will continue to try to get the best, cleanest data I can, RC states clearly that even with AI 4, the better the data the better the image. But its nice to know that when something goes wrong it may be possible to salvage something.

Thats a very severe example but it really is incredible. I've only tried it with a few images so far so please report back as you get results.

So what I've read is that the AI was trained only on linear so that is going to give the optimum result. But there is also an expectation that it will still do a good job at non linear. Some people have reported positive results on non linear. Let us know if you try it.

[SkyHiker]

Finally, BlurXTerminator does what I wanted it to do when I bought it! Here is a comparison, BXT to the left, non-BXT to the right. The data is an Autosave.tif from DSS. I ran BXT on it, then edited in Gimp with the same parameters. Gimp has a nice feature that lets you save stretching curves so you can replicate the exact same processing. We can now throw our coma correctors and field flatteners in the trash. I am not sure if I was using a field flattener on the scope (and ED80) but it looks like I wasn't.

[SkyHiker]

BTW Here are the full images at 33% (BXT first, non-BXT second):

[SkyHiker]

Here's another version of the differences that is easier to flip back and forth (BXT first, non-BXT second):