Use a more precise format for internal representation of images
ID0000103: This issue was created automatically from an original Mantis Issue. Further discussion may take place here.
Id | Category | View | Date Submitted | Updated | Due Date |
---|---|---|---|---|---|
ID0000103 | Siril | public | 2016-01-27 | 2017-11-20 | - |
Reporter | vinvin | Assigned To | vinvin | Resolution | open |
Priority | high | Severity | feature | Reproducibility | N/A |
Platform | - | OS | - | OS Version | - |
Product Version | - | Target Version | 1.0.0 | Fixed in Version | - |
Description :
Processing algorithms could really benefit a floating point representation, to avoid excessive conversions and to have a better range/resolution. This is obvious for background extraction, but would be very good for many others.
In this task, we will introduce a new native image format for Siril, probably double precision floating point and migrate all processing algorithms to handle it natively at least, and maybe pre-processing and sequence management as well.
We must have both this new format and the old one, unsigned 16-bit integers, supported natively, in order to keep Siril fast in many operations that manipulate many images (convertion, sequences, pre-processing). The SER format does not support more than 16 bits per channel, so we still need to support this format for this reason at least.
attachment(s) :
0