If there are specific raw files that I should process then if I am told how to download them I could process them. There are also other algorithms that do more to suppress color aliasing artifacts for cameras that are prone to creating them. That is because I didn't include non-linear ringing suppression, so the luminance interpolation has the usual trade-off of a linear filter between sharpening and ringing. The intent is to create a natural looking image that looks good as long as you don't zoom in too closely and pixel-peep. I just created this algorithm so it still needs evaluating- and it may benefit from some tuning: the look can be adjusted by changing the two filters. You can look at the example I posted and compare it to the output of the jpeg and raw of the same file in the comparison tool, or if you have Matlab you can process your own images- though the example script that calls it is rather crude (there isn't even provision for adjusting the white point except by changing the Matlab code). What are the benefits of using it vs the others. So, how does this algorithm perform compared to other demosaicing algorithms (e.g.
0 Comments
Leave a Reply. |