2. Demosaicing the lights preprocess lights_ -debayer
Registering the stars (i.e., finding the stars in the files): register lights_
Stacking the registered lights: stack r_lights_ rej 3 3 -norm=addscale
Load the resulting stacked image in memory: load r_lights_stacked.fit
Optionally: postprocessing the image crop 30 30 5950 3970 (you have to change those numbers according to your files) log rmgreen 1 rl 10 0.6
It is also possible to autostretch the preview image, and with a right mouse click save as TIFF from the preview (I haven’t found a way to autostretch from command line).
Exporting the final image: save result.fit savetif result close
What I’m saying is that you can’t feed Siril with images of different sizes, or it will automatically resize them all by the smaller size (which seems to be fair). I mean, this is what I think it’s happening.
Why don’t you repeat the steps @sguyader draw with the raw files, this time checking the debayer option?
Hmmm, that’s strange. I was under the impression I was feeding full-sized tiffs.
I gave it another shot, feeding the raw files this time, the resizing issue is no longer present, but the output file is very very dark, way darker than the raw.
I was able to finish that, no automatic resizing from Siril.
This time (and the first, for me), I loaded the raws and simply selected Scripts → … No_flat_No_dark_No_bias script. It did all the job and produced a _result .fit image. While in Siril, it’s just a matter of file->save and select tif, so you can do the pp on a raw editor.
Here’s my take (basically, exposure to stretch the histogram, then filmic and tone curves)
15:07:39: Running command: cd
15:07:39: ‘lights’ No such file or directory
15:07:39: Error in line 18: ‘cd lights’.
15:07:39: Exiting batch processing.
15:07:39: Script execution failed.
I tried renaming the folder to lights, but no dice. What am I missing?
the script assumes you have the raws on a subfolder named lights. Do that and rerun it
But it won’t give anything much different from what you got manually.
I think it has to do with the fact that the debayer was done in linear space.
You have to pp the result in any raw editor and get back all the tones (they’re there, but hidden to our eyes)
EDIT: Also, I recommend you to take a look at the script output and note the script location. Mine is /usr/share/siril/scripts/ (yours is probably the same).
open a terminal (you’re in Linux, right?), cd to that directory, then issue a ls command. You’ll see all the scripts in there. Issue a cat DSLR_Preprocessing_NoFlat_NoDark_NoBias.ssf. You’ll see the steps it took to process the images.
No big deal.
EDIT 2: Regarding the result I got, I’m missing some color (aren’t there any red stars?). Maybe I did something wrong. Let’s see if @sguyader add some light here (pun intended).
Weird, now I have an issue.
I edit the resulting tif in darktable, then export it to jpeg, and the exported image is very different from the edited tif. The jpeg looses the big stars
@gadolf thanks for what you did. I won’t have time to try before tomorrow.
One difficulty is white balancing, so maybe you lose red stars because of a wrong white balance. I get quite good results by using auto WB in RT on the resulting tif.
Another risk of loosing star colors is if you clip the highlights. Make sure to monitor highlights clipping when you tweak exposure.
I think there’s a benefit from feeding Siril with raw files when you preprocess them with raw bias/dark/flat frames, as these steps are done before demosaicing. If you don’t preprocess in Siril, I think you can get good results by developing the raw files in you favorite raw processor, making sure you have don’t clip anything in the process. In RT, you can use the “unclipped” profile, and save 16 bit integer TIFFs (unfortunately Siril doesn’t read floating point TIFF).