Resolution Coversion

[rummyPPG]

I'm looking for some formulas/methods (very explicit) to convert resolutions to smaller sizes.

Specifically I want to know the formula(s) scan converters use to get from 1280x1024 at 85Hz (and lower) down to the NTSC format. I.e if i had a screen frame in a buffer from a 1280x1024 85Hz signal from my computer, how would i make it into a frame that i could output to a NTSC TV?

Assume that I already did the A/D conversion and now have either 8 or 10 bit values (eahc) for RGB in my buffer. Also assume that I can take care of the D/A conversion and encoding of the RGB values into composite, s-video, etc ... All I want is how to take the larger frame and interpolate it (bilinear or higher) to get a smaller resolution image.

Thanks,
Peter

 

[Bothware]

Why would you want to do that - surely you would use TV out capabilites????

 

[rummyPPG]

Perhaps yes, perhaps no.

TV Out on older cards requires the screen to be set at the same resolution as the picture out to the TV (namely 640x480 or 800x600), whereas I'd rather have nice graphics on my monitor and have them display as best they can on the tv. Also I have a laptop with no TV out capability, but would still like to be able to display it on a TV. Finally, its fun to build things, so I'm also looking for a project ;^)

 

[Bothware]

I take it your an electronics engineer then, with appropriate access to suitable manufacturing facilities???

How on earth are you planning on implementing a resolution conversion algorithm in self built hardware?

 

[rummyPPG]

A microcontroller will be responsible for implementing the correct algorithm to match each VGA resolution/freqeuency to the NTSC standard.

I'm in school for computer engineering, so i have plenty of access to both parts and facilities for assembling/testing, but most of the parts can be gotten at radio shack -- they have make-your-own PCB which is easy to use, then just add your discrete components. The rest of the things, like the microcontroller, can either be gotten as free samples, or spring the few bucks for them.

What I'm missing is not how to put it all together or implement the algorithms, but what the algorithms are them selves, which I'm still looking for the answer to? How do I take a frame(s), combine if necessary and resize to the correct resolution, and achieve a descent quality?

 

[zephyrprime]

I've implemented one such algorithm. The Bresenham algorithm (not to be confused with the Bresenham Line Algorithm which was made by the same guy). I don't know if this is the algorithms resolution convertors use. There are other faster algorithms out there too. Just look at the resize algorithms available in photoshop and you'll find what you need to know.

Also, you should know that the sort of thing you want to do requires a lot a processing power so you can't just use a wimpy little pic microcontroller or something like that. You'll need something with more power. And something that 32bit is nice because it easiest to work with RGBA color components that way (even if the 'A' is blank.)