I'm considering a project where I need the same video being processed by multiple processes. The initial concept I have is to stream the video to some sort of multiplexer and then capture the video stream input.

Does anybody have any starting points?

I'd also accept APIs to help splitting AVI files.

um

i have no idea what exactly your talking about?

are you talking about accesing the same video with different processes and it won't allow or different applications?

what type of project is it? what are you useing?

help us help you.

last edited by WhiteWolf at 22:19:38 17/Apr/06

I am writing a computer visualisation program that uses uses computerised vision to modulate some visualisations. (i.e. if you move your hand in front of a camera cool s*** happens).

To combat the CPU intense nature of the computer vision (50% cpu load on a 3.2gig processor) I am considering using seperate processes on seperate computers. That way I can do multiple of things at the same time with out having to do low level concurrency.

To achieve this, I need to multiplex/stream my core video feed (which is either an uncompressed AVI or a video feed from a camera) to each of those processes.

Then, in theory, each process can get information from each feed and send the results to a final process which will create the visualistions.

Does that make sense?

#edit, 'cool s*** happens', not 'a cool s*** happens'

last edited by typo at 23:18:49 17/Apr/06

webcam + windows media encoder then access the video stream on each computer using mms:// or something.

or you could just use a hardware splitter to split the signal (assuming it isn't via usb or firewire).

webcam + windows media encoder then access the video stream on each computer using mms:// or something.

I should have mentioned that my webcam at home is circa 1995 or something and thus is s*** house. But I'll try your work at Uni. However, at home using an AVI file is my best way to work with the media. I'll look up windows media encoder.

or you could just use a hardware splitter to split the signal (assuming it isn't via usb or firewire).

That's a great idea and it is probably what I'd go for in a release model. However, for testing every cam I have access too is a usb cam and I'd also like to test with already recorded AVI files.

:D

uh, for motion detection you don't really need any more than 32 pixels or some s***, even an old web cam would be just fine. Although I suppose it depends on exactly what you're trying to do like sign language to speech or something would require a bit more resolution I guess.

uh, for motion detection you don't really need any more than 32 pixels or some s***, even an old web cam would be just fine. Although I suppose it depends on exactly what you're trying to do like sign language to speech or something would require a bit more resolution I guess.

The webcam is so s*** house that it draws motion markers on its own grainy quality. Also, I'm not just doing motion detection, there is some feature recognition and some other wicked stuff going on.

I've got about 4-5 core vision elements that I'd like to combine concurrently. Some of them are as high as 50% cpu (feature reognition), with some being as low as 20% cpu (motion detection). So, as you can guess, even low level concurrency programming wouldn't be that effective over all. Not to mention that threading in C isn't that fun.

i recommend buzzwords, and lots of them

i recommend buzzwords, and lots of them

I know it might be hard for you to understand this but; just because they are big words doesn't mean they are buzzwords.

synergise something if you can

haha infi :)

Where abouts in your house will you be setting this up Typo?

:/

um.. web cam and flash 8!

Where abouts in your house will you be setting this up Typo?

Why bear, it’s in my bedroom …

It’s actually a part of a framework for a project I am doing. So, other than testing, it isn’t really going to be used in my house at all. Although, the underlying framework would make a wicked security system that could be used in homes.

um.. web cam and flash 8!

The webcam I have at home is s***, really s***. The granularity actually triggers false positives on all of the visualisation possessors. It was so bad that it would occasionally match faces, bodies or motion from thin air.

After getting a bad case of “white noise” flashbacks, I decided to look else where. So, instead of my old webcam, I’m using AVI files to read from … hence my problem. How do I send the same frame of information to multiple processes on multiple computers simultaneously.

Finding something that worked well was actually much tougher than I thought it would be, so I’m writing something that does the job. It isn’t pretty, and it wouldn’t scale all that well, but it works.

#edit: In a final production system I’d use a camera splitter. How ever at the moment it is mostly proof of concept, and as such software hacks work well :)

last edited by typo at 23:14:05 25/Apr/06