2.15
#...
·
·
2
likes
|
|---|
|
Hi there, I just became a member of this group - not because GHS made a substantial difference but because I am using GHS together with other techniques. First I should maybe point out that I achieved very good results already with other techniques. I should also say that I am not using PI but Siril and that my main tool for improving images was in the post-processing phase with Rawtherapee. Rawtherapee offers the tone mapping tool and RGB color correction. The results were similar to applying GHS on an image. With GHS there are some differences, though. First, it is clear and reproducible at any time what exactly GHS is doing. Second, you can apply it to the color channels of the image separately. Third, you can subtract the stars, then apply GHS and then add the stars again. That's why I begin to like it. In my latest image I tried just that: I subtracted the stars, applied GHS only to the blue channel, and added the stars back, but only with a factor of 0.7. I would'nt say that the result was spectacular - but it was very easy to do the jobs. That's why it will probably become part of my standard workflow. Just my two cents stars Götz |
9.03
#...
·
·
1
like
|
|---|
|
Hi, I am using Pi and GHS integrates well. I used it after I use StarXterminator to get rid of stars and after NoiseXterminator. I then put the stars back in using Photoshop as a "lighten" layer. Jerry Y |
10.10
#...
·
·
2
likes
|
|---|
|
Hi Gotz, Both @Mike Cranfield and I were consulted a fair bit by @Adrian Knagg-Baugh and @Cyril Richard (two amazing developers) on GHS implementation in Siril. I have to say that the developments in Siril have been astounding and it really is a rock solid astro-image processing competitor - open source and adding new, and sometimes unique features all the time. Without getting into this software versus that, it is coming down to personal preferences - I have Siril on my machine and used it a few times- thanks for the reminder to fire it up again to see what's new. I too have gone down a huge rabbit-hole of stretching individual channels both for my monochrome narrow-band imaging and OSC dual narrowband filter images. - using PCC on linear just for stars; manually aligning histogram peaks - by overstreching with GHS and bringing back down into alignment using the HP parameter; stretching the weaker channel highlights more (after initial stretch) to bring more colour into the image; adjusting channel contributions to an extracted luminance - but I am a long way from settling down to my own "best practices" technique. As for star subtraction - it has become embedded in my workflow. Typically I remove them at the linear stage. The colour option stretch of GHS (like arcsinh and there are other ways) is good to bring more saturation into the stars, leaving pallette adjustment and other colour manipulations with the starless image. One thing I have noticed is when stretching an image with stars using GHS (typically with a large b), GHS can give the desired result by being very gentle on the stars and avoiding both clipping and bloat. However, neither of the two most popular star removal utilities seem to have been well trained on the more timid GHS stars and I found can occasionally miss stars when applied to non-linear. This isn't a disaster, if you are replacing stars anyways but I find it better to do my star extraction at the linear stage. Another thing I have to suggest, is that if you haven't tried deconvolution on a starless image yet - I highly recommend it. Create the psf from the image before star removal, and then use that psf with the starless image. You can then avoid any ringing artifacts and leave de-ringing completely off - which actually hinders the deconvolution iterations when implemented with a starry image. With deringing off, you only have to consider oversaturation of bright areas and noise generation in the background as artifacts. I guess this is a bit off-topic though. Thanks for getting the discussion going, Dave |
