Output for DeBayer2 with Bilinear type seems to be incorrect.

[shenxinyu1999]

The result of G-R order using bilinear has a similar color to G-B order using DDFAPD and vice versa.
I have implemented a bilinear demosaicing program so I think the problem lies in the bilinear type.

[jpsacha]

can you provide steps to reproduce the issue?

[shenxinyu1999]

OS: Windows 11
ImageJ version: 1.53t + Java 1.8.0_345(64 bit)

Raw Image I use: https://drive.google.com/file/d/1zvVSZ7u18qcHDGW6ZOo-nqcTXdAYwiey/view?usp=share_link

1. Open the raw image. Image type: 8-bit, width: 420, height: 288
2. From plugins, choose Raw Processing > DeBayer2
3. Set the options to G-R and Bilinear, click OK
4. In the result the color of the house is more greenish
5. Repeat with demosaicing type set to DDFAPD with Refining
6. In the result the color of the house is yellowish.
7. Repeat with order of first row set to G-B, and demosaicing type set to each. The results are opposite.

[jpsacha]

Thanks for a clear example of the issue. Looks that there is a bug in encoding demosaicing type from a name, like "G-R" to an integer used by the Bayer1 part of the code implementing: replication, bilinear, etc. "G-B" and "G-R" have they values switched. I will fix it soon.

[jpsacha]

New release with the fix is published: https://github.com/ij-plugins/ijp-DeBayer2SX/releases/tag/v.1.3.4