irwjager
06-07-2010, 11:09 AM
Hi all,
First off, I'm very grateful for the support for StarWipe and StarBright I've been getting from you all. Thanks! :thumbsup:
I'm working on a third tool in the StarTools series, which I use frequently called "StarDev". It's a replacement for the various debayering/demosaicing routines out there (e.g. converting RAW files into regular color bitmap files).
I had been getting frustrated with the (free) tools available, mostly based on DCRAW. Especially in astrophotography images, I find they can introduce inordinate amounts of noise, or create detail (artifacts) where none exists.
Therefore I created a replacement solution that can be trained to obtain better results for specific user-defined classes of images (e.g. DSO's, planetary, moon, etc.), instead of a sub-optimal one-algorithm-fits-all solution that is pervasively used. Using Artificial Intelligence, it uses separate neural nets for each class of image. The idea is that you train the program with some sample images, after which it will do a better job at debayering that specific type of image.
I find it does a good job, however I use a pretty obscure RAW format (CHDK), that, I suspect, most of you don't use. I don't know what RAW format other people use. So, before I can release StarDev, I was hoping someone could tell me what format they use and maybe send me some sample RAW files to test with.
Thanks in advance!
First off, I'm very grateful for the support for StarWipe and StarBright I've been getting from you all. Thanks! :thumbsup:
I'm working on a third tool in the StarTools series, which I use frequently called "StarDev". It's a replacement for the various debayering/demosaicing routines out there (e.g. converting RAW files into regular color bitmap files).
I had been getting frustrated with the (free) tools available, mostly based on DCRAW. Especially in astrophotography images, I find they can introduce inordinate amounts of noise, or create detail (artifacts) where none exists.
Therefore I created a replacement solution that can be trained to obtain better results for specific user-defined classes of images (e.g. DSO's, planetary, moon, etc.), instead of a sub-optimal one-algorithm-fits-all solution that is pervasively used. Using Artificial Intelligence, it uses separate neural nets for each class of image. The idea is that you train the program with some sample images, after which it will do a better job at debayering that specific type of image.
I find it does a good job, however I use a pretty obscure RAW format (CHDK), that, I suspect, most of you don't use. I don't know what RAW format other people use. So, before I can release StarDev, I was hoping someone could tell me what format they use and maybe send me some sample RAW files to test with.
Thanks in advance!