I've recently been pining for a version of dScaler for Linux and was wondering if someone from the dScaler group could tell me where the major hurdles are in such a task. It seems as though this could be a manageable project for me to work on, but I'm not familiar with the dScaler code.
It appears that the inner workings of dScaler are built as modules and that these modules could be built up into a workable Linux system as long as the glue to put them together was made.
I also read that a great deal of the modules are written in assembly. I'm not very concerned about speed because dScaler runs with low CPU usage on an average PC these days (Duron 700 or higher). As long as I could port some of the modules to slower C code, I'd be happy.
GCC does support inline assembly, just not with the same syntax as the microsoft compiler. I haven't looked at it, but in theory it ought not be that difficult to write a filter that converts MS syntax to gcc syntax.
I don't think you'd have to port the driver code. There is support in Linux for bt8x8 series cards. I have an old bt848 based capture card in my RedHat 7.2 server at home with a little webcam hooked up to it that I use to keep an eye on my dogs. I just plugged the card in and Kudzu recognized and configured it.
You would still have to port the dScaler code to Linux and make it talk to the driver though. Then there's the issue of making it talk to the display device. I don't know much about how Linux handles graphics, but it does handle openGL if that helps.
Thanks for the replies... To start, I wouldn't want a UI. I'd like to just get the data in, be able to use one of dScaler's deinterlacing modules, then shovel the data out.
Of course, I have yet to look at the code, though, so I'm not sure how modular things are. If they have a heavy reliance on a particular interface to the Bt848, then they aren't as modular as I would have thought.