• Please could anyone help me with MultiVid – I have a patch working perfectly, generating a 5x5 grid of images in scan mode, which builds cumulatively and then changes the image in each position sequentially as different images arrive from a picture player. All I need to do is replace the images with a new set, but now it will only show one image at a time. It cycles through the grid correctly, but the grid doesn't fill up as before. I'm using same file size .jpgs and all images are cropped to the correct proportion. It will work in Simultaneous mode, but not Scan mode. And when I add one of the new images to the previous media list, the grid fills up until the new image is selected and then all others disappear and only the new image shows.

    I just can't see the logic to this behaviour and I need to fix it so I can finish my final MA project. I'm using Isadora 2 (though only with v1.x functionality) with Mac OS 10.10.5.
     Any help would be greatly appreciated!
  • Tech Staff

    I just ran a test on Windows.. and I see an incorrect behaviour as well.

    I am submitting a bug report.
    Perhaps you can create a 5x5 grid user actor, that routes the incoming image to correct location (projector sized and positioned, or maybe the Sprite or Matte actors)
    You can reproduce the effect in this way.

  • Thanks for that. Unfortunately I don't understand how to do what you've suggested. I just wish MultiVid would work correctly again – when it does, it's brilliant.

  • Tech Staff

    I was having trouble with it in 'scan' mode only. It seemed to wipe the history of the images on a random basis.. so the background would black out while the images continued to tile along.

    In anycase, there does seem to be an issue.
    Take a look at the 'matte' and 'matte++' actors. These will allow you to build up the same 5x5 grid. Then you just need to figure out how you want the images coming in. You could use a 'router' to switch between the 'matte' actor inputs to allow you to create the same scanning effect.

    Building your own this way would allow much more control. For instance if you wanted to be able to update a while 'row' at once you could set that up, or a column. Perhaps update a 'X', followed by the remaining blocks...

    If you tell me your image resolution, and your output resolution, I would be happy to build a user actor that you could use ( I think I might use it too... sounds fun)

  • Hi, thanks again. I looked at 'matte' but as I didn't need a movie on a movie I opted for 25 picture players and projectors, using a router with 25 outputs, plus two shuffle (+ comparator) actors linked to 'select' and 'in' on the router. That enables me to send a random selection from 32 images to the projectors. I've no idea if that's the most efficient alternative to MultiVid, but it worked for me. Shows how much work MultiVid saves if you need a large grid....

  • Tech Staff

    That sounds like a perfectly effiecient method if building the effect. In fact if it is running as vid-GPU, then its likely faster than the MultiVid actor.

    I have made a quick vid-cpu actor that works much like MultiVid actor.. allows setting a grid size dynamically. (2 or larger seems to work)
    - its basic, but features would be easy to add.
    Requires a Image (vid-cpu) input, and it moves thru the grid with every new image sent.
    Maybe its useful to you.


  • Thanks very much! Could you answer another query – I've got another patch which generates five lines of text moving left to right, but the motion stutters. Is there a way to fix that? I tried motion blur for one of the projectors, but that made the stutter worse on all lines...

  • Tech Staff

    Are you using the 'Text Draw' actor and GPU video output?
    That is what should allow for the smoothest animation.

  • I'm using Text/ure actors, each loaded with a different word list, and I don't know how I tell if I'm using GPU video output

  • Tech Staff

    The 'text/ure' actor currently supports only CPU video.

    (You can tell how it is running by looking at the text in the (video/rgb inputs and outputs))
    If you are using the 'horz pos' and 'vert pos' to animate the text, you are likely best to use the 'Text Draw' actor (not the 'classic text draw') to do the text drawing on the GPU.
    Otherwise you could continue with the 'text/ure' actor, and control the animation via the 'Projector' actors 'horz pos' and 'vert pos' inputs. (the Projector is the only actor that will do GPU based manipulation of a CPU video input. This is due to the Projector moving the data to the GPU for output).