To keep things clear and avoid spamming that thread with updates, this post serves as the official, consolidated release of the CFALD method, including the whitepaper, workflow, and GitHub repository.
All future updates, examples, scripts, and user feedback will continue here.
What CFALD Is
CFALD (CFA-Domain Luminance Drizzle) is a new processing technique for OSC cameras that extracts true monochrome luminance directly from the raw CFA data before any demosaicing.
By combining the native R, G1, G2, and B samples in each 2×2 Bayer cell, CFALD produces a mono luminance frame with significantly higher SNR, zero interpolation artifacts, and cleaner micro-detail than standard RGB workflows.
When this luminance is dithered and drizzled, the reconstruction recovers fine structure far better than a normal debayer-then-stack pipeline.
In practice, CFALD provides an SNR boost comparable to a mono sensor binned 2×2, while preserving full OSC resolution.
Please check the GITHUB:
[ShaunisaGit/CFALD: CFALD: A CFA-domain luminance extraction and drizzle reconstruction method for dramatically improving OSC astrophotography SNR and detail.
If you have a DSLR or OSC Siril projects you’d be happy for me to test with CFALD, please pm. You’ll be given all credit for providing the images of course.