The noise generation is not random. The seed used seems constant
Closed this issue · 2 comments
carmeloarci commented
I tried in different conditions and verified that if I generate two consecutive images using the same parameter I obtain exactly the same image. The background noise and the illumination Poisson noise do not change their random seed.
Attached is a notebook.
Test random seed.zip
teutoburg commented
This is very much by design, to have reproducable results for testing purposes!
If you include "!SIM.random.seed": None in your configuration, you will see a difference between each image.
hugobuddel commented
The current behavior is kinda stupid though, because all detectors get the exact same noise, see #97. Perhaps we should create an issue about how we want the RNG to behave.