2.11
...
·
·
1
like
|
|---|
|
Hello, Does anyone have an idea why Starnet2 leave my stars with these patterns ?  Thanks |
8.51
...
·
·
2
likes
|
|---|
| With bright, overexposed stars it does that. So far I couldn't find a cure (other that not using it). It is less prominent is you do the star removal after stretching the image. |
1.51
...
·
|
|---|
|
I'd recommend looking at this tutorial, specifically the parts regarding "relinearization" https://www.nightphotons.com/guides/star-addition |
|
2.11
...
·
|
|---|
|
Thanks for your answer. You're right. If i run StarNet on an unstretched image it's fine. Do you know if it happends too with starXterminator? |
|
8.51
...
·
|
|---|
Thanks for your answer. StarXterminator does not leave artifacts But it might remove some part of the nebula as stars, especially in the linear phase. |
11.80
...
·
·
3
likes
|
|---|
|
If that really bothers you, you may try running StarNet++ a few times, each time with different stride parameters, and average the results. This is my default mode of running StarNet++. Or you can just circle the area in Photoshop and apply a Gaussian blur to wipe out those. There are no real data in near-saturated star cores anyway. StarNet++ tries to create something in areas where there are no real useful information. If that's no cheating, wiping out what it created by hand shouldn't be considered cheating either. Can be somewhat time consuming if you have many stars like that, but there is nothing wrong with doing such editing. My 2 cents. |
3.87
...
·
·
1
like
|
|---|
|
I have done what Mr. Wang suggests and used a small gaussian blur brush on the worst, largest stars before re-combining them. It works. This is the result of StarXterminator, I like it much better. It even puts the nasty halo on the star layer. This is 57 Cyg in the pelican nebula, a magnitude -1.23 star. You can see that it's a bit nasty in my telescope and had a halo that was removed by StarXterminator.  |
|
2.11
...
·
|
|---|
|
Thank you all for the help. For now i'll remove stars while in linear, not my favorite method but.... it is what it is ! Testing processes like starnet it's a computer killer, so for now i'll have to give some rest to mine. |
