convolve effect
[1/6] from: greggirwin:mindspring at: 14-Nov-2001 12:06
RT, et al
A while back, I made a brief attempt at writing an edge detection filter in
REBOL but it was hideously slow so I gave up.
I don't know how the compositing engine effects work, but would it be
feasible to expose a generic 'convolve effect that took a 3x3 kernel so we
could create new effects with it? That way RT wouldn't have to fill our
requests for new compositing effects.
--Gregg
[2/6] from: media:quazart at: 14-Nov-2001 14:26
YESS!
please....
This would be usefull for preparating images used interactively later on...
-MAx
[3/6] from: petr::krenzelok::trz::cz at: 14-Nov-2001 23:02
Gregg Irwin wrote:
>RT, et al
>A while back, I made a brief attempt at writing an edge detection filter in
<<quoted lines omitted: 3>>
>could create new effects with it? That way RT wouldn't have to fill our
>requests for new compositing effects.
Well, long time ago I asked about the feature. My friend suggested me
that general native matrice operations could be used for many
computational stuff, but I am no math folk, so I don't know if it would
help.
However - here's link Rebol is talking about upcoming effect pipeline:
http://www.escribe.com/internet/rebol/m15951.html
taken directly from Holger's description:
- Added "func" effect. This allows you to call a REBOL function from
within the effect pipeline, manipulate the current image (including
its alpha information) and pass an image back to the effect pipeline.
Is that what you are asking for? Or do you mean operations upon image
datatype directly? Now another question popped in my mind - are Core and
Command users left behind? What do I mean? Core and Command both know
image datatype. But they don't know of face composition. How can we work
with images using Core and Command? Hmm, shouldn't datatype manipulation
functionalities be part of the kernel of the system?
PS: just don't tell me to use View ;-)
-pekr-
[4/6] from: carl:cybercraft at: 15-Nov-2001 20:41
On 15-Nov-01, Petr Krenzelok wrote:
> Core and Command both know image datatype.
Core knows the image datatype? Are you sure? I get...
>> make image! [1x1 #{000000}]
** Script Error: Feature not available in this REBOL
** Near: make image! [1x1 #{000000}]
Perhaps it's just Command?
--
Carl Read
[5/6] from: cyphre:seznam:cz at: 15-Nov-2001 11:04
Hi Pekr and all,
> Well, long time ago I asked about the feature. My friend suggested me
> that general native matrice operations could be used for many
> computational stuff, but I am no math folk, so I don't know if it would
> help.
>
Native built-in matrice operations will be very handy not only for creating
2D effects.
We need them also for making 3D vector gfx. If you remeber Carl's answer in
interview at osnews about adding 3D engine into /View...I believe RT would
need them too ;)
regards
Cyphre
[6/6] from: greggirwin:mindspring at: 15-Nov-2001 11:01
Hi Pekr,
<< Well, long time ago I asked about the feature. My friend suggested me
that general native matrice operations could be used for many
computational stuff, but I am no math folk, so I don't know if it would
help. >>
Yes, that's the idea. I'm not a math person either, which is why I'd prefer
to have RT do it. :)
<< - Added "func" effect. This allows you to call a REBOL function from
within the effect pipeline, manipulate the current image (including
its alpha information) and pass an image back to the effect pipeline.
Is that what you are asking for? >>
That's a great feature too, but I think for the matrix operations, REBOL
functions will still be very slow if we have to pick/poke values.
<< Or do you mean operations upon image
datatype directly? Now another question popped in my mind - are Core and
Command users left behind? What do I mean? Core and Command both know
image datatype. But they don't know of face composition. How can we work
with images using Core and Command? Hmm, shouldn't datatype manipulation
functionalities be part of the kernel of the system? >>
It should work for Core and Command as well. It has nothing to do with
displaying the images, just manipulating the pixel values in them. I haven't
looked into it, but I wonder if there's a similar way to deal with audio, to
create audio compositing functions? It doesn't matter what the data is, it
may just be raw data (GIS or simillar) that you want to perform the
operations on.
--Gregg
Notes
- Quoted lines have been omitted from some messages.
View the message alone to see the lines that have been omitted