Ivo ST Interview

[Mike in Rancho]

The Elf actually linked this Stargazine interview on CN, and I watched the whole thing last night.

https://www.youtube.com/watch?v=8Luu2-BMXzc

Wouldn't have known otherwise. I didn't see any link or post about it here, but sorry if there is one I just couldn't find.

Some useful extra description about what's behind ST and what some of the modules are doing. The processing walkthrough was also nice. I've seen a few silent ones, but the narration really helps to get a grip on what is happening and what to be looking for.

There were a couple open issues at the end and for a subsequent Part Deux, I think, my thoughts on those -

Indeed something other than beautiful observatory data would be great for a walkthrough, and much closer to reality. Well, for me anyway. I'd love a narrated walkthrough where Wipe really needs to be cranked up, or the balance between Wipe and the second AutoDev worked out in order to have the cleanest data going forward. Just from regular, properly calibrated, amateur data, not any kind of stacking or calibration errors.

I'm less interested in a save state than I was at the very beginning of using ST. For one, GPU makes things so much faster. And two, I'm just sort of used to it now and really the tracked period doesn't have to take super long time wise.

It would be kind of nice to have similar, short narrative videos as to each ST module, with some samples, what to be looking for, and perhaps some (light) discussion of the signal processing that is behind each one?

[admin]

Great to hear you found it useful Mike!

Indeed, I'd love to walk through a "mere mortal" dataset - I was actually hoping to do one during that same presentation, but time got away from me.

I'm less interested in a save state than I was at the very beginning of using ST. For one, GPU makes things so much faster. And two, I'm just sort of used to it now and really the tracked period doesn't have to take super long time wise.

That's precisely why I never implemented one - endless tweaking is not (should not be) a thing, so the time spent in "Tracking limbo" (quite literally - the signal stays in limbo until it is "collapsed" into the final untracked image) is comparatively short.

It would be kind of nice to have similar, short narrative videos as to each ST module, with some samples, what to be looking for, and perhaps some (light) discussion of the signal processing that is behind each one?

Absolutely.

I've been gradually working towards something in StarTools over the past ~3 years that, once implemented, should greatly help. Though it will not have the form of a video format, but rather a text/images format.

In a nutshell, the idea is for the software to better understand what is being processed and who is doing the processing.
It then uses this information to better set/suggest defaults and - much more interesting/novel - intelligently put together a personalised tutorial based on the object, dataset characteristics and skill level.

The pieces for this are almost in place, and I'm hoping to be able to start implementation for 1.9 or 1.10.

[Mike in Rancho]

Hmm, 1.10 sounds frighteningly like AI. Or maybe the MS paperclip?

Yes, mortal data would be great. And with some good clear outer space in it, not 100% nebulosity either so that's it is just "well, no need for an ROI."

[KlausKlaus]

Any news on 1.9 or 1.10?

Klaus

[admin]

Any news on 1.9 or 1.10?

Klaus

Version 1.9 is shaping up to be mostly about multi-language support (EN/DE/ES to start off with, with many thanks to @hixx and @Carles), improvements to SV decon, plus miscellaneous quality-of-life improvements.

I don't have an ETA yet, mainly due to other work commitments taking up a lot of time, but I will be seeking some feedback on decon-related improvements in the next few days or weeks via a private alpha.