[fluxus] feature request: pipeable ppm image output

[Dave Griffiths]

> I've done basically the following with Pd+Gridflow, it works quite well,
> so I think it would be good if it were possible with Fluxus too :-)
>
> In Fluxus:
>
> (start-framedump "myfifo.ppm" "ppm")
>
> In a parallel shell:
>
> cat myfifo.ppm |
> ppmtoy4m -F 25:1 -S 420mpeg2 |
> mpeg2enc -f 8 -a 2 -o video.m2v
>
> Result:
>
> a video encoded without a vast amount of temporary files and all the
> problems that entails.
>
> Drawbacks:
>
> the usual fifo thing of blocking until the other end reads (which made
> me think Pd+Gridflow had crashed, when I tried it there, but it was
> merely waiting for the fifo to be opened for reading)

a couple of things...

ppm support looks fairly easy to add, I quite like the asciiness of it -
it would also be good for reading and writing to pixel primitives in
scheme.

is ppm the only file format you can do this with though? tif's would
probably be faster as they will be a lot smaller, and the data won't have
to be converted to ascii.

would it be possible also pipe images in using the same manner?
(force-texture-load "myfifo.png")

generally though I don't like this way of passing data around, as it's
syncronous, and it hits the filesystem which is going to slow everything
down. but other than writing a lot of code, and until video jack (which I
should really start looking at properly) I guess there is no better way of
doing this.

> Other wishlist items:
>
> lock window/frame size when piping ppm, as i'm pretty sure ppm's don't
> like frame size changing mid-stream.

this could be a commandline option - to set the size and then lock it.

cheers,

dave