Problem with Register starting in v1.076 Beta 4

Hi Mike @mestutters,

Please upload those frames as soon as you can and I will have a look 😉

In bet4 relative to beta3 nothing changed with respect to registration. What changed however is how the reference frame is chosen, maybe it works if you use another reference frame? I will test 😉

https://ariesprodstor.astropixelprocessor.com:7001/

username & password: appuser

Mabula

Hi Mabula,

The frames should now be available.  My recollection is that with Beta3 I had an error message that one of the frames failed to register.  I have not seen this when processing the subs with Beta 4.  

On Calibration tab I set the cosmetic corrections check-boxes but otherwise used the default options during processing.

fyi I have loaded the integrated channels that I had previously from Beta 3 and processed these with Beta 4 RGB Combine presets HOO1 and HOO2.  Both worked fine but I am undecided at the moment which output I prefer.

Regards

Mike

Hi Mike @mestutters,

Thanks for uploading.

Yes I see the problem, the registration gets stuck in an endless loop, because of the new reference selector which is not working on this data. It tries to switch to another reference but no one works because of a data issue clearly.

What did you do with this data? Is it calibrated by iTelescope or did you do it? The data is full with bad pixel and bad columns still... do you have the original raws and calibration masters?

Mabula

Hi Mike @mestutters,

I have fixed the first issue with the reference frame switcher, It will first try the best H-alpha frame and then if that fails, it tries the best O3 frame as reference. This works correctly now:

Now, 3 frames, 2 Ha and 1 O3 aren't being registered. And something add is happening with the data. The data is flipped... this is very strange behaviour and MaximDL is reponsible, MaximDL has calibrated the data and is messing things up here...

Look at this, this is the reference: 

and this is one of the frames that can't be registered. The data is flipped in the Y-axis, left is right, right is left. This is not 180 degrees rotated as with a pier flip...

This is odd behaviour by MaximDL, if you can get the original raws and calibration masters, we should be able to get everything working and better 😉

Mabula

Hi Mike @mestutters,

Good news, registration will work if you:

- enable flip descriptors in X/Y

- disable same camera and optics

This is all due to MaximDL that is flipping the data ... (which it never should do if you ask me...)

(Oh, distortion margin can be kept at default allthough my screenshot shows a wider value which is needed...)

I will quickly release a new stable version with the fix for the reference switcher 😉

Mabula

Hi Mabula,

The images were captured about a week ago.  Thus far  I have downloaded only the calibrated frames.  I should still be able to download the calibration frames.  I will try to do this later today and upload the frames to your server.     I had not noticed the flipped images until you pointed it out.

Mike

Hi Mike @mestutters,

Nice 🙂

Mabula

Hi Mike @mestutters,

APP 1.077 has been released especially to solve the never ending registration loop on this dataset 😉

Thanks for reporting the issue so quickly.

Mabula

Thats a great amazing jobe Mabula 😎 

I will have to study this weekend for the new version 1.077 and process my new project on M-42 😀 

Thanks for all

Best regards,

Hi Mabula,

First, thanks for investigating and fixing the looping Registration issue so promptly.   Also for spotting the flipped image issue that was preventing those three  frames from Registering in Beta 3.   I had not looked at them closely before and just assumed that they were of too poor quality to Register.   Three frames out of 19 gives me c15% more photons to integrate! 

Looking at the results from 1.077 I am currently thinking that I prefer the teal coloration of M1 obtained by using the HOO1 preset compared with the bluer result of the HOO 2 preset.  However the star colouration of the HOO2 preset seems to me the more natural.   I shall explore combining the two results in PS but overall a very pleasing outcome

Following up our dialogue from yesterday, I have uploaded the iTelescope Raw and calibration frames to your server.   I thought that by just loading these files into APP, creating a BPM and then integrating the frames as I would subs from my own CCD that I might get a slightly better result.   I was surprised however by the result of this approach which looks as if the MFs have been doubly applied:

By eye, I think I get the best result by deselecting the MD and MFs and just using BPM, MB  and cosmetic correction.  Do you know if Maxim is maybe doing something in the background to the Raws ?  How would you process these subs for best result?   Perhaps it is best just to follow the Christan Sasse videos and simply load the iTelescope calibrated frames into APP and using just cosmetic correction for hot/cold pixels and columns.

Best Regards

Mike

Hi Mabula

I find it's very easy to spend hours tweaking colour, contrast and sharpening but I'm really pleased with this result.

Once again, thanks for your input.

Best regards

Mike