Quick question: When you decode to DNG, are you using any type of standard decoding method? Im trying to compare it to raw2cdng decoding quality and see which is better and I was curious if you decoded to DNG ? Long story short, should I execpect any quality difference when decoding using methods?
MlRawViewer has two different data paths for converting into DNG. In both cases a DNG container is created with tags extracted from the source file metadata.
1. When stripe reduction is NOT in use, the source raw data is copied and, if needed, padded to 16bit values with zero bits. In addition, (bad) pixels with zero value are hidden with surrounding data. This path should have very minimal impact on the data.
2. When stripe reduction IS in use, the data is loaded to the GPU and 3 shaders are run to estimate and correct any stripe patterns in the data due to different amplifier settings, as well as hot pixels. The result is written out as 16bit values. This path can touch all pixel values, but is intended not to cause an overall visible change in the image, other than reducing stripes and spots.