s-video encoding?

[Chuck McManis]

Ok, I'd like to get an s-video output from an FPGA. Seems like most people
have gone to DSP's attached to some reasonable fast DAC for this sort of
application these days. I s'pose I can build the necessary I & Q generators
in the FPGA but it seems like it will take up more space than I should have
to. What I want is something that takes 15 bits of RGB in (optionally with 1
bit of alpha) and generates S-video out. A search of the TI web site has
lots of decoders, but no encoders, same with Phillips.

Any suggestions from the rest of the world?

--Chuck
cmcmanis at my website.

 

[Zak]

Chuck McManis wrote:

What I want is something that takes 15 bits of RGB in (optionally with 1
bit of alpha) and generates S-video out. A search of the TI web site has
lots of decoders, but no encoders, same with Phillips.

Most VGA cards with TV out also have S video out. Google finds the
Bt868/869 as the first hit. It does all kinds of format conversions,
color space conversion, resolution conversion and handles quite some
input formats.


Thoms

 

[Chuck McManis]

Excellent! That is just what I needed.

"Zak" <spam@jutezak.invalid> wrote in message
news:LCn7c.5819$EV2.48104@amstwist00...

Most VGA cards with TV out also have S video out. Google finds the
Bt868/869 as the first hit.

 

[Mac]

On Sun, 21 Mar 2004 19:11:32 +0000, Chuck McManis wrote:

Ok, I'd like to get an s-video output from an FPGA. Seems like most people
have gone to DSP's attached to some reasonable fast DAC for this sort of
application these days. I s'pose I can build the necessary I & Q generators
in the FPGA but it seems like it will take up more space than I should have
to. What I want is something that takes 15 bits of RGB in (optionally with 1
bit of alpha) and generates S-video out. A search of the TI web site has
lots of decoders, but no encoders, same with Phillips.

Any suggestions from the rest of the world?

--Chuck
cmcmanis at my website.

I still don't get what you want. One point of major confusion for me is
that RGB and S-video are both analog signalling schemes. No FPGA that I
know of can accept RGB inputs or produce S-video ouptputs without
additional circuitry.

By the way, I'm assuming that by RGB, you are referring to the typical
video interface between home computer video cards and monitors. This has
analog red, green, and blue channels along with horizontal and vertical
sync timing signals. I can't remember if th hsync and vsync are encoded in
with the green or if there are separate wires for them.

Furthermore, given the bandwidth of video, a digital conversion from RGB
to S-video makes little sense. You would need three very fast ADC's, some
high-speed processing, and, what, three DAC's (or is it two, I can never
remember). Then an FPGA or DSP(or two, or three) in between the ADC and
DAC.

I'm not a video expert, buy you seem to be totally lost on the subject.
Maybe you need to read up on this stuff a bit before you go too much
farther.

--Mac

 

[Chuck McManis]

"Mac" <foo@bar.net> wrote in message
newsan.2004.03.21.21.55.08.820349@bar.net...

I still don't get what you want. One point of major confusion for me is
that RGB and S-video are both analog signalling schemes. No FPGA that I
know of can accept RGB inputs or produce S-video ouptputs without
additional circuitry.

You are correct sir. However you can produce composite video with nothing
more than two resistors, an a capacitor. Further, a number of the FPGA "demo
board" providers such as XESS (www.xess.com), BurchEd (www.burched.com), and
Digilent (www.digilentinc.com) proivde DB15 connectors wired up with some
R2R pairs to give 6 bit (2-2-2) VGA outputs that you can hook to any
multisync monitor.

Furthermore, given the bandwidth of video, a digital conversion from RGB
to S-video makes little sense.

Nominally we're generating RGB in the digital domain. The math to convert it
into luminance and chrominance is straightforward but consumes a large
number of logic blocks in an FPGA.

I'm not a video expert, buy you seem to be totally lost on the subject.
Maybe you need to read up on this stuff a bit before you go too much
farther.

Hardly.

So for the curious, I'm contemplating a "generic" video game. The latest
FPGA prices have come down to the point where you can easily build things
like a video unit, processor, and controller input into a single FPGA. My
thought was to build an "open source" video console that would be
inexpensive to produce yet sophisticated in its capabilities. It wouldn't
rival a playstation or an Xbox but it would be good value as a TV attached
entertainment device. It could also be leveraged in a number of interesting
ways.

The last time I worked on the design of a video card was in 1984 when "1024
x 768" was workstation resolution and required a $5,000 BARCO monitor to
display. "Multi-Sync" was a joke us EE types tossed about like the CS guys
did "bug free." So, I'm a bit out of date

The BT868 is a good chip (it's also gone "end of life" as of last year
however :-() The Connexant replacement appears to be workable. Now to find
out what the 100K piece price is (always a bit scary).

--Chuck

 

[Mac]

On Sun, 21 Mar 2004 22:11:54 +0000, Chuck McManis wrote:

"Mac" <foo@bar.net> wrote in message
newsan.2004.03.21.21.55.08.820349@bar.net...
I still don't get what you want. One point of major confusion for me is
that RGB and S-video are both analog signalling schemes. No FPGA that I
know of can accept RGB inputs or produce S-video ouptputs without
additional circuitry.

You are correct sir. However you can produce composite video with nothing
more than two resistors, an a capacitor. Further, a number of the FPGA "demo
board" providers such as XESS (www.xess.com), BurchEd (www.burched.com), and
Digilent (www.digilentinc.com) proivde DB15 connectors wired up with some
R2R pairs to give 6 bit (2-2-2) VGA outputs that you can hook to any
multisync monitor.

Furthermore, given the bandwidth of video, a digital conversion from RGB
to S-video makes little sense.

Nominally we're generating RGB in the digital domain. The math to convert it
into luminance and chrominance is straightforward but consumes a large
number of logic blocks in an FPGA.

OK, I think I see what you're saying. You have the video frames stored
digitally in a RAM or inside the FPGA, and you want to output S-video. And
I guess you are saying that if you can run your clock at an appropriate
speed, you can generate S-video from a digital output using a 1st order
filter. Is that right?

I'm not a video expert, buy you seem to be totally lost on the subject.
Maybe you need to read up on this stuff a bit before you go too much
farther.

Hardly.

My apologies. I misunderstood what you wanted to do, and as a result, I
mistakenly thought you didn't know what you were talking about.

Anyway, good luck to you. Since I've only recently started reading up on
the details of video (mostly component video) I'll be of no use to you.
;-)

[snip]

--Chuck

--Mac