StarTools 1.9 preview

[Carles]

Sounds great, @admin !

Sorry I am a little behind on the Spanish translastion, been busier lately!

Nice set of improvements there, specially offloading more work to GPU seems interesting too

Clear skies!

[hixx]

Thanks Ivo for the background...

I tried Pi, 0, 1, sqrt(2) and some of the most important physical constants (don't know where to place the units?! ) as well, but they all have one thing in common: nothing happens. Maybe my PSF samples are not distorted enough ... ?

Wow, Dietmar, I think If You have circular stars from the get-go, this parameter is obsolete. I just tried e as well and it did a decent job, but left some long stars in the extreme corners.
My M31 widefield has ellyptical stars (some 3:1 aspect ratio) in the borders (I need to work on adjusting the working distance between field flattener and sensor, I guess), so I was just curios and used the extreme setting 5.99 (or whatever). With just 11 iterations the corner stars now would convince even the pickiest pixel-peeper, being circles with a halo, even @ 400%,but what the heck....
It's a Decon abuse to correct for any optical distorsion but the result is just really mind blowing , however, Euler or smaller just does the safety car job pretty well......

Clear Skies,
Jochen

[decay]

I think If You have circular stars from the get-go, this parameter is obsolete.

Hi Jochen, I double-checked this because I usually have terribly deformed stars. But actually, this parameter has no impact on my dataset. With higher values, the image gets just a bit blurry. Maybe the stars are too fat at all or I used unsuitable samples? And there seems to be nearly no impact on the fat stars, only the cores are tightened ... ?

Before, After, Spatial Error set to 3.51:

2022-12-06 19_17_04-Window.jpg (240.4 KiB) Viewed 18171 times

But as discussed with Mike, it's probably better to wait until Ivo starts beta phase. Holiday is holiday and Christmas is Christmas.

Best regards, Dietmar.

[admin]

Sounds great, @admin !

Sorry I am a little behind on the Spanish translastion, been busier lately!

Nice set of improvements there, specially offloading more work to GPU seems interesting too

Clear skies!

Thanks Carles. Good to hear you're busy and hope you're well!
I am myself behind on the translations as well (changing hardcoded strings etc.), so no stress.

[admin]

I think If You have circular stars from the get-go, this parameter is obsolete.

Hi Jochen, I double-checked this because I usually have terribly deformed stars. But actually, this parameter has no impact on my dataset. With higher values, the image gets just a bit blurry. Maybe the stars are too fat at all or I used unsuitable samples? And there seems to be nearly no impact on the fat stars, only the cores are tightened ... ?

Before, After, Spatial Error set to 3.51:

2022-12-06 19_17_04-Window.jpg

But as discussed with Mike, it's probably better to wait until Ivo starts beta phase. Holiday is holiday and Christmas is Christmas.

Best regards, Dietmar.

That's odd... You should definitely see marked changes. You do typically need to increase the number of iterations however...

[hixx]

Hi Dietmar,
Please try 10-13 iterations, some highlight headroom 1.05, Resampling & Centroid Tracking, Deringing around 30% and crank up theSpatial error and use some of the deformed stars as samples (in each corner the shape might be different...
For me, it changes the star shape. Values around "Euler" leave some halo with a tightened center, which looks more natural for slightly deformed stars, but strongly deformed stars require values between e and the top end - these really squeeze the stars into points. It's a matter of taste when one think its looks too much...
And happy X-mas time everyone....
Jochen

[decay]

Hi Jochen, thanks for all your replies

Please try 10-13 iterations, some highlight headroom 1.05, Resampling & Centroid Tracking, Deringing around 30% and crank up theSpatial error and use some of the deformed stars as samples

I did all that (and some more iterations, like Ivo suggested), but the result regarding the 'Spatial Error' (that was what I meant) is as before: No significant change from value 1 up to 6. Just a bit more blurry. But maybe that's totally fine, since the stars are not that much deformed?

But I have the impression, that the overall 'clarity' with your settings is better than before?! I will have to double-check that, but thank you!

(And to repeat it once more: the impact on the smaller stars is unbelievable! )

Best regards, Dietmar.

[decay]

That's odd... You should definitely see marked changes. You do typically need to increase the number of iterations however...

Thank you, Ivo. I tried some more iterations (16), but there was no change regarding this "Spatial Error' parameter setting (please see post above). But as said, that's probably perfectly fine.

Could you please give us/me a hint if you would like to have feedback already now or better not until next year, maybe starting with a beta version?

Best regards, Dietmar.

[admin]

Could you please give us/me a hint if you would like to have feedback already now or better not until next year, maybe starting with a beta version?

Any feedback related to the quality of the output/images is definitely helpful. Any quirks regarding UI may be a little early, as there is plenty to do... Thank you!

[admin]

Could you please give us/me a hint if you would like to have feedback already now or better not until next year, maybe starting with a beta version?

Any feedback related to the quality of the output/images is definitely helpful. Any quirks regarding UI may be a little early, as there is plenty to do... Thank you!

Saying that, I am actually looking for feedback on the SVDecon workflow/UI. The automatic determination of sample size and automatic apodization mask generation in particular (any failure cases). Note also that you should now be able to select stars that have overlapping sampling areas (if need be).

Clear skies!