Hi,
thanks for the pic, it’s really nice!
Here’s my attempt, using RawTherapee.
Happy new year!
11 Likes
Sorry for not citing correctly, I am writing from my phone. I guess you are correct about the convolution, but I think it is by no means blind, but maybe not so simple because of errors along the way. First, we have to get the coordinate system right. It is given by the star trajectories and may be described by conformal mapping. To get the parameters, hough transform or some flow analysis technique, maybe? In the transformed coordinate system, the deconvolution should be possible since the original kernel is known (a line of length of the star movement). What is missing is some options for the fix landscape. Of course, all of this is harder for the milky way than for regular star fields. Does this make sense? I am a noob at image processing but extremely fascinated …
1 Like
I hope you get somewhere with this Chris. Any software I seen that controls for star trails is Microsoft based. I would love to see convolution editing of night skies for the Linux community too.
I’m not an expert on image processing either.
What you say does make sense, and it has been tried:
I don’t know of any success story though. I guess in practice there are a lot of factors that play into this. Lenses are far for perfect. Cameras are far from perfect. The atmosphere is (at least at higher magnifications) far from perfect and so on.
What could also be interesting would be training a Deep Neural Network on the problem. After all that seems to be all the rage these days.
@harry_durgin do you have a link to something for windows that actually works well for correcting star trails?
hi @agriggio, did you do your version wholly in RT? I couldn’t work out how to get something out of the hills but keep the sky under control. How did you do it, please?! Maybe you could post up your .pp3?
I used two different Rawtherapee conversions as layers in McGIMP. Masked the hills using the scissors selector.
2 Likes
Hi @RawConvert, yes, I did everything in RawTherapee. I kind of suck at GIMP, so I try to keep it simple and use only RT 
Here’s the pp3. Basically, I used tone mapping, graduated filter and custom L* and CC curves to try to brighten the lower part as much as I could while keeping the sky reasonable. I also applied quite a bit of noise reduction, using Contrast by Detail Levels to keep the details. Finally, Defringe to fix the purple stars, and auto WB, increased contrast and vibrance, and a Fuji Velvia film simulation (from the HaldCLUT collection) for toning.
2016-07-26-23-40-37-0.ARW.pp3 (9.5 KB)
1 Like
I used darktable for editing, RawTherapee for noise removal and GIMP for scaling down and sharping the image.2016-07-26-23-40-37-0.ARW.xmp (8.2 KB)
2 Likes
Nice job on the milky way, that’s the colour and look I prefer and was trying to get myself in my efforts.
Isn’t the “Lavender” tinting ironic? 
1 Like
Here is my first attempt, using PhotoFlow and G’MIC:
The foreground is still too dark for my taste, but I am trying to avoid any visible sky/mountains transition.
For the sky/mountains separation, I have used a combination of a bilateral-blurred luminosity mask and few gradients… of course, that’s the most tricky part of the edit.
1 Like
Thanks @agriggio, I’ll have a go with this, though possibly not for a couple of days.
I didn’t think of purple fringing, doh!
I was a little bilaterally blurred on new year’s eve.
It’s many years for me already, and things are not improving… 
Appropriate?
Yeah. I think.
I thought RegiStax had been used to remove or reduce star trails by some users. But, alas, my research is not encouraging. I did read many reports of doing a translation of a few pixels with a minimum blend mode. I also saw references to using a Richardson Lucy deconvolution in GIMP. But I didn’t find enough information to be helpful.
R-L deconvo. can be had by loading the GMIC module, opening the ‘details’ tree, and selecting that sharpening mode. My edit above uses that sharpening mode (on both original and output) as made available in Rawtherapee My Shot of the Year - #38 by HIRAM
1 Like
I’ve had some good success with the Vixen Polarie tracker, which has sidereal (star), solar, lunar, and 1/2 siderial tracking rates. Runs off AA batteries or mini-USB input. I’ve been using their earlier version polar alignment scope. It’s run by a stepper motor and precision brass worm drive gearing.
I can also supply power with a Makita LXT 4 amp/hour tool battery and ADP05 USB adapter with two USB outputs.
1 Like
Looking forward to giving this a try
Got my edit online. If you’ve got the time to watch, I’d love feedback. Especially this is by far the most knowledgeable community I belong to.
2 Likes