Assuming you had a sufficient number of users running the client, I would
like to see this split single frames into different rendering jobs, then
stitch the image back together when it was done.

For example, I have a set of four machines (450, 500, 702, 822MHz), and I
have one scene which takes these machines three solid days to render, and
I've only done a low quality image so far: 640x480, no animation, no AA,
povwin set to highest priority (there's a lot of glass involved). That's
about 280 hours, about 30 minutes per line. With 480 clients out there on
the net, I could have my image in a half hour instead of 3 days...

I don't recommend this in place of animation rendering, but rather in
addition to it. In fact, it would be just as beneficial to animation tasks.

Elsewhere this conversation had turned to hacking, and you should pretty
much assume this will instantly fall under attack as soon as it is publicly
available. The world is full of deviants, and it looks like a lot of them
own computers now. As was already mentioned, file IO would be an issue, but
worse than that, you will see people purposely return false results. The
SETI project had a terrible time with that. Imagine, you send off your
wonderfully detailed Louvre script, and you get back an image of Miss
November. (Ok, maybe it wouldn't be completely bad, but...)

As far as I can tell, there wouldn't be a way to really detect a
false-results hack, unless the system maybe randomly submitted test renders
to clients (with a known one-row return image) to see who wasn't behaving
and killed their account if they sent back gibberish.

Also, denial attacks will be common -- particularly in the case of something
like POV, it wouldn't take much effort to submit a bunch of files that would
effectively prevent anybody from ever getting anything else processed. The
script I mentioned earlier takes about two hours to render as just a 160x100
thumbnail and lengthy rendering times definitely wasn't my goal...

Actually, to bring this commentary full-circle, I bet farming out rendering
as single-line tasks would do much to alleviate the false-results problem.
It simply wouldn't be much "fun" for a hacker to just mess up one row in a
1200-line image, and the script owner could perhaps re-submit a single line
for re-rendering.

--j.

Post a reply to this message