this is completly offtopic. Its not even a vvvv question, I am asking it here because I know there might be people around that have an answer to my problem.
I am having a hardware issue with a current project.
We are running 2 computers that calculate some '‘stuff’ 'independently and in the end we want to join the signals together on one of the computers.
Therefore we were thinking using the TV-Out (card: ATI x300) from one computer (computes some processing stuff) to go into the BNC-In of our Framegrabber card (IVC-200) on the other machine. Somehow at PAL-B the image sucks - its green/red interlacing terror (http://www.unshift.de/manuel/tmp/ivc-200-grab.png, http://www.unshift.de/manuel/tmp/ivc-200-grab2.png). At NTSC the quality is slightly better, no colorish artifacts but it is not crisp at all.
Probably its just stupid thinking that this might be an appropriate way - but in lack of my own knowledge, I have no other idea of how to get the picture in realtime from Processings visual output (maybe openGL offers something?) to the other machine - where we want to key and output the joined feed via vvvv or eyesweb.
strange idea of yours… but:
i’d say the colorproblems we can see in the images you provided are correct. pal-video is greeting! if your framegrabber and video-output can do SVideo try that. it should give you a noticeably better image. maybe even a better cable would do (which i doubt).
interlacing artefacts are also correct since pal comes in 50 fields and you are only capturing 25 (i guess) you always get two mixed. you could try capturing only half-pal resolution which would likely result in the capturedriver ommitting one field. deinterlacing is not so easy. how are you capturing? anyway have a look at www.dscaler.org.
maybe you could describe your situation in some detail then someone could come up with a better idea then processing the output of one pc on the other via video input. maybe vnc is an option?
Thanks for the fast reply, strange idea yeah!
Maybe I really need to get S-Video Framegrabbers. Damn. The IVC-200 was recommended by one of the EyesWeb guys and honestly I dont have too much of a clue about video stuff (not that you didnt notice so far).
Lets describe the setup a bit closer:
Basically its an gesture controlled installation.
Machine A is receiving a live video via Capture card (card: IVC-200, composite BNC) and is computing/analysing some _stuff_with EyesWeb. EyesWeb sends an analysed image (almost) every frame over to Machine B which is generating visuals in Processing to it. These generated visuals shall then be send back to Machine A, where they are keyed and mixed with the live video.
Everything works well so far, the only issue is now to have the full crisp image send to Machine A (plus the keying).
This all needs to be done in realtime of course hust.
A really stupid idea: Is there any card that can “capture” a VGA in? So I could just go via VGA-out to that In? I never ever heard of that, but who knows…
One of the fun features of the DeckLink HD card is its video desktop. When not being used for video capture and playback, DeckLink cards default to a fully featured desktop display which outputs as video. Using the DeckLink SDI video output, you can send any application window to the video output.
apart from that, your digital-to-analog-to-digital proce55 seems slightly awkward, may I add. Maybe do all of it in Jitter?
Strange I thought I had posted something…well here we go again:
Thanks again for your help. I checked the VGA grabber, it will ship in about 8 weeks and it also doesnt support what I am looking for, its rather ment to be a HD recording device for pictures but I guess messing with the SDK could be an idea…well its too late anyway since I need to be done in 2 scary weeks.
In the meantime I checked another capture card which works better in S-Video but still not the quality that we need. So I think I drop the basic idea of Video-out > Video-in and look for alternatives.
If I consider my output possibilities (VGA-OUT, Video-Out, Network) I think the only one that can could do it in realtime and good quality is the VGA, or what do you think?
I mean the output isalready there in great quality, I just need to get it IN somehow :)
But maybe there could be a network possibility…
Ok, I think Ill drop that idea, take some features out and let it run on one machine. This is just too much hassle with 2.
Now comes the next trap.
How can I join a live-video together with the processing output? Can I grab somehow the output-window of Processing? Any experiences?
Yes I guess thats the idea. I already tried that with Jitter which consumes about 40% of a 3.4 GHZ CPU @ 20fps. eek. Do you think vvvv will be faster? Actually I tried to put it together in vvvv but I couldnt find the block to set the resulting texture to a ex9-layer for the renderer input. Obviously Im a total noob with vvvv
When I wrote grab the output I was thinking of grabbing it from memory or something in that direction…
(still couldnt figure out that screengrab in vvvv…if anyone might be so kind on this beautiful late-summer day to throw that together, I would be really thanksful so I could compare the results of various approaches…)
ScreenShot (EX9.Texture) is pretty easy to use:
either set the last pin to “Rectangle” and use the pins x, y, width, height to select a rectangle on screen that will be output as texture.
or set it to “WindowArea…” and use the HWND node to get a valid window handle that you connect to the ScreenShots “Handle” input. HWND takes the title of a window as string.
seems we have introduced a bug to the ScreenShot node shortly before shipping beta9…i fixed that now but sorry, you’ll have to wait for beta9.1 for it to work.
anyway in your patch there have been some more problems/missunderstandings. probably you didn’t dig through HowTo DX Rendering. from there you should know that vvvv’s coordinate system goes from -1 to 1. therefore if you set the size of a rectangle or that of the ScreenShot node setting something like 640x480 doesn’t make sense.
i adjusted the attached patch accordingly. but you’ll only see it working with beta9.1
Ok just to give some final feedback to everybody: I have bought the Vision-RGB-Pro - and its amazing.
It captures easily 1280*1024 (didnt try higher so far) at 60 fps. CPU load relativly easy. I am also simultaneously capturing with a framegrabber at PAL resolution, and PCI bus is doing it easily.
The card is expensive, but definitly worth it.
Amazing piece of hardware.
There are some other cards/manufacturers like: www.mutech.com, www.unigraf.fi, www.alacron.com unigraf is supposed be as good, but a bit more expensive, mutech is cheaper (1000$) than the Vision-RGB-Pro (1300 €) I think, but I couldnt get it as fast as I needed it.
I’m a little bit late about your post but if you can read this one I’ve three questions.
Which card did you buy: Single VGA or Dual VGA?
Could you play a video full screen in 25fps (1280 x 1024 first PC configuration screen) and capture it with your card in a second PC (1280 x 1024 resolution also)? Do the video display on your capture PC is fluid?
2b. If you have the Dual capture card can you capture two videos like previously in the same time? How is it in your Screens? Cpu load?
Does the card work properly with vvvv?
I would like to khnow if I can buy this card.
Thank you.
