Vvvv to c

Is there a way to take the vvvv patch and convert it into c? Here is my current patch: http://nuigroup.com/forums/viewthread/162/

can you please help me?


as vvvv is basically made of very high efficient code:

what do you expect by converting the patch to c? which parts of your vvvv solution seem to be to slow? for which parts you need better (faster) solutions in vvvv?

here is the patch, http://nuigroup.com/forums/viewthread/162/, it lags, in the detection of blobs, and runs at 100% cpu.

vvvv has a useful function called debug mode, showing cpu consumption of any node involved.

the killer in your patch is asVideo. converting dx9 graphics into direct video is something you should try to avoid. windows is just not made for this kind of conversion…or the hardware ? well something makes this a crazy bottleneck. its NOT a vvvv problem.

stick with freeframe plugins for motion tracking, don’t do any video processing with a shader if you want to analyse the stream with a freeframe plugin later on.

looking at your patch you used a difference shader, probably to do background subtraction. you can do the same with the freeframe plugin trautner, which comes with vvvv.

true. the AsVideo node can be a bottleneck as u7angel describes, but i think it depends heavily on the graphikcard in use.

which graphikcards are you using build18er and u7angel?

i have used the videotexture-shader-asvideo-countour combo successfully on an nvidia 7800gt card without significant performance troubles caused by asvideo. i had two cameras with 640x480@60fps combined to one texture of size 512x256 and running that through AsVideo to Contour which actually ran at around 100fps.

i have a ati x1900. with some recent machine and maybe an nvidia card it might be fine BUT if i can avoid this trouble maker, i would. especially, if the machine is occupied with more than just tracking.

whats wrong with trautner for background subtraction followed by contour ? that makes the patch not just more efficient but even clearer - less nodes.

never tested atis newer cards with AsVideo.

nothings wrong with trautner-contour if it works. actually i only used the way via asvideo only to combine two cameras in one image. not using AsVideo should always be faster. i was just surprised that it seems newer cards don’t have so much troubles with AsVideo anymore (at least at low resolutions) as older cards always had.

iirc correctly that should be the main advantage of pci-express, which can handle bidirectional data transfers quite fast, in comparison to agp, which is basically only a one way street. can anybody confirm this?

the reason i am using asvideo is because countour won’t work without it attached. also my gpu is a geforce 6200 agp8, and processdor is a dedicated machine running a celeron 3.2GHz

freeframe plugins dont work when nothing goes in…connect a valid videostream like the one coming out of videoin and it should work. asvideo outputs nothing else then the same shit ;)

and regarding your agp card you should avoid asvideo at all…this card and slot is dated. you better punish your cpu with a pure freeframe solution.

right. i’d definitely opt for a pci-express graphic card. in your setup surely the agp-slot is the bottleneck.

You have the link to my patch, what shouldi change in it?

the graphics card :)