EWOCprojects proudly presents the all new and fully-featured Pure Data video mixer application called EWOCvj.

The application is quite a beast to set up, requiring a Linux setup, installation of Pd-extended and compilation of some custom Pd externals. EWOCvj is under continual development but can be used and is already used for variable VJ tasks. To use the program one needs one or two quadcore computers (PAL 25f 720x576 output) or dual core (15f 320x240 - not yet supported) and the application is specifically written to be used with the Vestax VCI-100 MIDI mixing'n'scratching deck, you won't be able to use it without one, maybe future will see a mouse-only interface? When using two computers, one can VJ with two persons, the video database being on both machines, with one preparing videos with loop and effect settings and sending all this data to the other machine, where the other person performs the live mix. EWOCvj features 46 realtime video effects with 5 effects per deck (two decks A and and per-deck saturation settings, 10 video wipes, a normal and inverse chroma key, one video+effect monitor per deck, Quicktime recording of monitors or mix, 32 video bank in 4 saveable 8 video parts, 8 videos in preset bank for direct triggering with Korg NANOpad2 and another NANOpad2 serves as loop input recorder/player. Videos and parts can be sent to other locations on the same or other machine. Most effects have a parameter, controlled by a turning knob. The video database consists of Quicktime movies 720x576 25f and one needs to transcode to the dv codec.

We know this program needs a load of setup and hardware, thats why we would really like to know if anyone is using it on any projects, also feedback and feature requests are always welcome. Also arrangements can be made if help is wanted getting the setup together. Consider this program "feedback-ware", if you use it, let us know at paleajed@gmail.com!

Project webpages at http://www.ewocprojects.be
Documentation still under construction. Let us know if you need help!

Looks like a pretty cool and also useful project, if a bit long-winded to get going. Have you considered put everything in place on a live CD?

Would be great to see some footage of it in action.

A live cd, thats a good idea, I didn't think of that yet! Thanks for the hint, I will most certainly look into it!

Video footage showing the program will be on site soon.
Also taking the oppotunity to stress this is a production-level project, it has already been used on hour-long VJ sets and in theater settings.

It took a lot of work, but the first version of the EWOCvj LiveCD is now available on the website (link above). If there is anyone out there with the required hardware, it'd be nice to have you testing it.

And soon some documentation will be on site.

You can contact me on paleajed@gmail.com or through this forum!

Weird problems testing the mixer...

Everything runs smooth on my Intel laptop quadcore, but when testing on my AMD desktop quadcore, I run into a very annoying bug.
When running the videomixer, a [metro 40] object that counts the frame count of the clip, shows very visible stuttering, namely it counts 12 or 13 frames then pauses a quarter second and so-on. The clip playback really suffers. Strangely enough it seems, when disabling the pdp_qt structure, stopping the clip doesn't resync the metro object. There seems to be something wrong with base Pure Data...

Anyone?

I have a Pure Data window onscreen with a load of bangs as a user interface. When I minimize this window, video and metro are silk-smooth again. So this GUI is totally unefficient... Would it help to use Pd0.43? I hear it has renewed the GUI?

It might be worth giving it a try. From what I understand, 0.43 had an overhaul on the Tcl side of things, which is what actually draws what you see on the screen. There's still some work to be done on the C side of things, which sends the Tcl messages to update the display. Apparently, there's still some bottlenecks there.

There's also Pd-L2ork, which is a fork of Pd-extended being developed at Virginia Tech. They've supposedly made significant progress getting the GUI to run smoothly and efficiently, as well as fixed some other longstanding bugs. I'm not sure how compatible it is with Pd-extended, though.

http://l2ork.music.vt.edu/main/?page_id=56

I separated the GUI part in its own pd process today (someone should write an external to automate this) and this solves all stutter problems!

How do you communicate between the two processes, by using netsend and netreceive over the localhost network?

Yes I use netreceive and netsend.

This carries me on to the next point. The videomixer relies on video being sent between processe on different cores using pdp_netreceive and pdp_netsend. This works well with anything except very high resolutions on my Sandy Bridge i7 2630qm quadcore, thus OK since I prefer to use standard PAL 720x576 anyhow. But on my older AMD Phenom quadcore, it seems the memory throughput to and fro the cores is not even enough for this standard PAL. I think its the send and receives since gkrellm gives low values for processor load, so it must be a memory bottleneck? Does anyone know a way to prioritize and make faster this throughput on Linux?

Any hints appreciated...

Also watch the net this night for a new EWOCvj release 1.1 which features most importantly support for native NVIDIA and ATI drivers. Also dual cores will be supported.

It seems the problems so-called due to memory bottleneck were due to bad performance by the open source ati driver. On the latest LiveCD is now support for the proprietary ATI driver and everything runs fine, also on the AMD Phenom.

Hello everybody:

@Maelstorm said:

It might be worth giving it a try. From what I understand, 0.43 had an overhaul on the Tcl side of things, which is what actually draws what you see on the screen. There's still some work to be done on the C side of things, which sends the Tcl messages to update the display. Apparently, there's still some bottlenecks there.

There's also Pd-L2ork, which is a fork of Pd-extended being developed at Virginia Tech. They've supposedly made significant progress getting the GUI to run smoothly and efficiently, as well as fixed some other longstanding bugs. I'm not sure how compatible it is with Pd-extended, though.

http://l2ork.music.vt.edu/main/?page_id=56

I installed the "Burrito Supreme" version of Pd-L2Ork onto Debian Wheezy. I wanted to see if it was really an improvement on the GUI. It installed right away without problems. First impression: it looks a little nicer. Something new: the console is detached from the main window. The magnifying tool for connection changed into a square aiming visor. Something really good: if you put a new atom below an existing one it automatically connects them. Next I loaded my current work-in-progress-monstrous-patch to see what it was up to. Ok, the names of subpatches go beyond their grayed area but there are no black lines around boxes. Let's go to performance behaviour. It gave the same amount of cpu usage for the same patch and setup. It can coexist, apparently, with pd-extended already installed, so what they claim in the website is true.
Bump! When changing desktops (I'm using gnome3) i hear glitches, not happening with current version of Pd-0.42.5-extended. It happened to me the same when I was using awesome2 as desktop environment.
Bump squared: Gem does not work at all, despite being included in the "extras" folder.

For a twenty minute test and review, I think it deserves to be better tested on what it was inteded for: laptop orkestra. I didn't check networking performance or Wii-mote interfaces. I think that the main advances had been put on that, so it's ok for me. Worth give it a test when I have some extra devices, he he.

Read you all soon, mates.

Sumidero