StarTools

I was thinking about this the other day, and I'm not sure it gets enough pub. I didn't know about it myself for the longest time, and so could perhaps be an issue that particularly beginners ought to be aware of from the get go - especially if comparing to programs that downsample their previews.

Every now and then it has popped up in a forum thread. And it actually is also noted in the docs, but is sort of buried and you have to go hunting for it (under "Interface"?) with a some clicks. Details are also a little thin, which is where I had some questions...

Is the algorithm always on, for all modules and stage of tracking/non-tracking?

Is the algorithm for the preview image only, and never saves out?

Does it only kick in at preview levels below 100% "zoom," and 100% would be more of a WYSIWYG, and above would be some kind of straight zoom or upsampling?

Thanks!

Mike in Rancho wrote: ↑Sat Oct 08, 2022 6:09 pm I was thinking about this the other day, and I'm not sure it gets enough pub. I didn't know about it myself for the longest time, and so could perhaps be an issue that particularly beginners ought to be aware of from the get go - especially if comparing to programs that downsample their previews.

Every now and then it has popped up in a forum thread. And it actually is also noted in the docs, but is sort of buried and you have to go hunting for it (under "Interface"?) with a some clicks. Details are also a little thin, which is where I had some questions...

Is the algorithm always on, for all modules and stage of tracking/non-tracking?

All modules show your dataset in this way.

Is the algorithm for the preview image only, and never saves out?

It's just a scaling algorithm for the UI. So, I guess it "saves out" in so far that if you make a screenshot, it will be in your screenshot.

Does it only kick in at preview levels below 100% "zoom," and 100% would be more of a WYSIWYG, and above would be some kind of straight zoom or upsampling?

Correct, correct and correct!

Really, there is not anything terribly mysterious about it - it's just a scaling algorithm that is intentionally "dumb" and simply "skips" pixels when squeezing down the image, rather than computing some sort of average for the range of pixels it is skipping. The latter would give a false impression of the noisiness of the image, and is really quite undesirable when gauging the impact on noise propagation of various tools/modules. The other benefit of "stupidly" skipping pixels in this way, is that it generates aliasing patterns if any latent repeating patterns exist in your data.

You can performing this sort of "dumb" scaling in, say, GIMP by choosing "None" for Interpolation Quality in the Scale Image tool.

Finally, if you really need to see some form of scaling that does averaging or outlier rejection (latter coming soon in 1.9), you can fire up the bin module.

Hope that helps,

Thanks Ivo, that's useful.

I was just wondering what I was looking at and when. As well as if the Bin module was included, since we use that for evaluation of SNR and how much the image cleans up.

With only an FHD monitor, I think I'm almost always in a below 100% mode, even after binning. It has been useful for digging out hidden patterns I didn't know were there, however.

I may need to upgrade to QHD or UHD, though all the choices seem to start getting a bit baffling.