What do people think of Starnet version 2? - Forums

Just starting to try out Starnet version 2. It seems to be quite a big improvement. I'd welcome other people's thoughts?

Here is a small version of an original Crescent capture:

Here's a version after star removal with version 1:

and here's the same after using version 2:

and this is what the version 2 star mask looks like:

I think it's particularly impressive when it comes to big stars and diffraction spikes which we as Newtonian users live with and for some of us actually like

For me at least, it was a massive improvement. Not that V1 was bad, far from it. But it always left artifacts, specially in the place of small stars that made it really difficult to work with the starless image. In the end, I used starnet V1 almost exclusively to generate excellent star masks.

Starnet V2, however, is a whole different beast. The resulting starless image is almost devoid of artifacts, with the exception of very large ones. This makes processing a starless image separately from the stars extremely easy and effective. See my starless version of the Carina nebula:

Carina Nebula (NGC 3372) under the Full Moon

I think it's a huge improvement. I stopped using the original version and switched to Star Exterminator. After trying version 2, I was so impressed by the results that I'm going back and reprocessing some old data. Version 2 will be in my workflow going forward.

I have just one word for the results of StarnetV2: FLAWLESS!

This software is an amazing gift and has changed the way we process our images forever.

Chris

Thanks for the heads-up, yes its a huge improvement. I only imaged this the other night, OIII data, just over 3 hrs, Bortle 8. 1CMa wasn't removed but that's no big deal. Overall, cleaner image and artefacts are less obvious or easier to manage out, very happy indeed. I used linear data, convenient feature to have, worked well.

Really hope there will be a command line version for mac. Don’t want to buy pixinsight just for that tool 😅

I also switched from Starnet V1 to StarXTerminator, a huge step forward both in terms of picture quality and integration in Photoshop and PI.
I now re-processed some of my files and Starnet V2 has always created better results than StarXterminator, so for now Starnet is the new king and I am glad that there are now two very potent solutions to choose from.

Michael

Starnet V1 -> StarXterminator -> Starnet V2 Done.

Gotta try it. I have been very happy with StarXTerminator.

Is the version 2 just installed with the latest PI Version? I guess not? I hate downloading something from sourceforge.

I looked on CNs and I guess it does not work with linear images?

I looked on CNs and I guess it does not work with linear images?

It does also work with linear images

I looked on CNs and I guess it does not work with linear images?

It does also work with linear images

They say they stretch it- destar-then go back to linear.... anyway will look at it since pfile knows a lot about such matters and was involved, but I am very careful about SourceForge downloads. It should hit PI mainstream.

Good destaring is essential and I hope it is an advance.

I paid 60$ for StarXterminator and now I regret it because StarNet V2 is even better…

Very pleased with it. so far I've only done some tests on old data, But I can't wait to use it on a proper project.

Figured I'd show my results. Especially focussing in on a noisy part of the mosaic image/data.
Just standard/default settings, on pre-stretched data in pixinsight.
results might be more 'fair" if star removal is done later in the process, after some de-noising and such.. but.. here we are.

standard stretched data:

StarNet V1: (It doesn't seem to like noise at all, and can't deal with the noise in this specific part of the image. Other parts of the image were "better".. again: it might be more fair to Denoise the image first.. but.. meh, it's only a quick test)

Star Exterminator: (obviously, can't beat the speed of starxterminator! pretty ok result, still a bit noisy. And all removed stars have turned into blurry blotches)

Starnet V2 (Very happy with this. great improvement of V1.. Less noisy than StarXterminator, and the "filled in parts" are far cleaner)

Its scary how good the new starnet is. I basically stopped using starnet when I started using reflector telescopes because of how messy it made the image but this version 2 is unbelievably good.

Test image before starless

Starnet v2

Stars put back via pixel math, cropped. You cant even tell a starnet was performed. The diffraction spikes are still sharp and the stars lack ringing artifacts.

It’s great that V2 didn’t break CUDA GPU acceleration support.

I ran v2 on three images last night after downloading.
I run an oldish Ryzen 7 2700x cpu with no gpu acceleration (AMD card). In all three cases, v2 completed in less than one minute.

Wow what a difference -- even running V2 standalone on a 16bit tiff looks great!

side by side, V1 and V2 (pardon the sensor artifacting)

I started using StarXterminator, which works well in most cases, but in this particular case Starnet V2 kicks azz.
From top to bottom: original, starxterminator, starnet v2 (all in linear mode with standard STF)

Hi all,

I did the same with my Cocoon-Nebula last October and amended the old comparison of StarNet++ and APP with the new StarNet++ v2.
(Disclaimer: The StarNet++ v2 data is not processed as the other two. This is just about the star removal

)

Best regards
Mike

how StarNet v2 works for me is like black magic

V2 is a great improvement over V1.
There are instances, where SXT performs better than SN2 and the other way around.
I do not regret buying SXT.

I have also tested the new Starnet and have to say that I am really impressed! Even large stars and spikes are removed very well. The result also looks less noisy and very homogeneous in the areas where the stars were before.

A big thank you to the developer! To provide such a great and powerful tool free of charge for the community can not be appreciated enough!

Cheers,
Frank

Orion Redux

I used the new StarNet on this one and it did a really nice job. I struggle to see any artifacts that would of plagued the super bright stars before. I highly recommend taking the developers advise and checking off the up-scaling option as it really helps with the smaller stars.

Many thanks for posting the topic here, I had no notice that the new version was released on sourceforge.

Installation of Starnet V2 in PixiInsight is flawless, and apparently it does not interfere with the existing "official" V1.

I tested on a single image all three tools: StarNet v1 StarNet v2 and StarXterminator. SX is clearly superior to SN v1, but SN v2 has picked up. In my specific case SN v2 has managed some Newtonian spikes significantly better than SX, but on the other side has left behind a few small artifacts that SX has managed well.

There are some disclaimers in the SN v2 announcement in Cloudiynights that make me think that SN is going to become a pay product.

All in all a great improvement. For the time being I keep both SN v2 and SX in my drawer, and both are excellent.

Mau

There are some disclaimers in the SN v2 announcement in Cloudiynights that make me think that SN is going to become a pay product.

I don't think this is the case, even if I believe the developer has every right to do so. I suspect that something more sad has happened: Someone else used Starnet's output data in order to train his/her own software and probably he/she is making a profit out of it. It is sad that others exploit his generosity. Whatever Nikita decides, he needs our support.

Despite the bug on linear images, Starnet v2 is a great improvement. Maybe the temporary stretch applied is not invertible (in the mathematical sense)?

In my old PI installation, Starnet v2 can happily coexist with the older version, and surprisingly, the older Starnet is compatible with the newer tensorflow.dll file.

Thank you Nikita for this truly innovative tool!