Hi.
I am inquiring about fit compression. My initial idea was to enable lossless compression but in 32bit it doesn’t seem the best method. I saw that there was a quantization parameter but I don’t know what value to choose. Can you advise me ?
Thanks in advance
Soulearth
Hello, we have a new documentation online, this may help you: FITS — Siril 1.2.0 documentation
The default parameter works well in general, then it’s a trad-off between having smaller files and how many bits are actually used to represent your data.
Hello.
I did read the documentation and it’s great to have it. But it’s a little imprecise.
For exemple, the sentence: “If it is too large, one undersamples the pixel values resulting in a loss of information in the image. If it is too small, however, it preserves too much of the noise (or even amplifies the noise) in the pixel values” explains the principle very well but leaves the user a bit in the dark.
Finally, what number choise for delete only noise ?
It is difficult to make the right choice between not enough and too high.
Maybe examples corresponding to “strong”, “medium” and weak compression might help.
In any case thank you for all you do and thank you for the help.
Regards.
it depends on what the image data is. If you just convert a 16 bits image to 32, it will still have 16 bits of dynamic range, so losing a lot of bits in the compression will not be a problem. If your image is a stack or a processed image, it may need a lot more precision.
The default works in most cases, but for calibrated images you can probably increase it more without losing data.
To test it, you can set a value, save an image, open the image and compare pixel values with the original or their statistics. Repeat until you find the best for your images if you really need to…
Ok. I understand. I use and stack with a imx533. So after stacking, dynamique use probable 18bits but no very more.
Thanks you for the method. I will test.
Regards.