how to mirror the live video from the camera, help!
As the question goes, I want to build a mirror-camera effect in Jitter, in which anything moves from right to left, will display as the opposite: left to right.
Could any one know "cv.jit" can do that? and how?
Thanks!
just flip the matrix then
no need for cv.jit?
v a d e //
www.vade.info
abstrakt.vade.info
how to flip it?
What is the commend name?
HELP!
if you're using the jit.dx.grab or jit.qt.grab objects (importing images
from a camera) u can send the message 'settings' to it, which opens the
configuration dialog for the camera in question. Most of those dialogs have
'flip horz' an 'flip vert' checkboxes.
On 10/18/06, luckyforest wrote:
>
>
> how to flip it?
> What is the commend name?
>
> HELP!
>
jit.dimmap, jit.matrix with srcdim/dstdim, .... many other ways.
wes
On 10/18/06, luckyforest wrote:
>
> how to flip it?
> What is the commend name?
>
> HELP!
>
I think the command name is
'do the tutorials'
v a d e //
www.vade.info
abstrakt.vade.info
On Oct 18, 2006, at 5:52 AM, luckyforest wrote:
>
> how to flip it?
> What is the commend name?
>
> HELP!
Thanks every one here, I have done it with "jit.rota"
On 10/18/06, luckyforest wrote:
>
> Thanks every one here, I have done it with "jit.rota"
Nooooooooooooooooooooooo! Unless you need arbitrary rotations, don't
use rota. It's expensive. Use jit.matrix src/dst dim or jit.dimmap
or jit.resamp or ...
wes
speaking of expensive, is there a list , official or not, of optimum
methods for doing typical/useful things in jitter?
ie, is mxform2D or repos going to be faster than srcdim for this
particular function?
is there a coherent thought process behind choosing optimal methods,
or is it just trial and error with the objects?
v a d e //
www.vade.info
abstrakt.vade.info
On Oct 19, 2006, at 3:02 AM, Wesley Smith wrote:
> On 10/18/06, luckyforest wrote:
>>
>> Thanks every one here, I have done it with "jit.rota"
>
>
> Nooooooooooooooooooooooo! Unless you need arbitrary rotations, don't
> use rota. It's expensive. Use jit.matrix src/dst dim or jit.dimmap
> or jit.resamp or ...
>
> wes
i guess the coherent thought process behind it is trying to guess how
the object actually works internally
flipping consists of moving pixels around without much calculation,
rotation is much more complicated than that doing calculations per
pixel, so jit.rota makes less sense in this case than jit.dimmap.
I suppose it helps if you have some programming experience other than
jitter so that you can make better guesses as to what's under the
hood, and I'm certainly not claiming to really know what is best.
and I guess it's always good not to start optimizing before speed has
actually started being an issue (what's the guys name, Knuth ?).
Good thing about using jit.rota is that it might lead you to trying
other things.
ciao,
Joost.
On Oct 19, 2006, at 9:11 , vade wrote:
> speaking of expensive, is there a list , official or not, of
> optimum methods for doing typical/useful things in jitter?
>
> ie, is mxform2D or repos going to be faster than srcdim for this
> particular function?
>
> is there a coherent thought process behind choosing optimal
> methods, or is it just trial and error with the objects?
>
>
> v a d e //
>
> www.vade.info
> abstrakt.vade.info
>
>
> On Oct 19, 2006, at 3:02 AM, Wesley Smith wrote:
>
>> On 10/18/06, luckyforest wrote:
>>>
>>> Thanks every one here, I have done it with "jit.rota"
>>
>>
>> Nooooooooooooooooooooooo! Unless you need arbitrary rotations, don't
>> use rota. It's expensive. Use jit.matrix src/dst dim or jit.dimmap
>> or jit.resamp or ...
>>
>> wes
>
>speaking of expensive, is there a list , official or not, of optimum
>methods for doing typical/useful things in jitter?
Official? Naah.
But when you think of compositing, I encourage workshoppers
to consider the following, and to make the assumption that if
they only WANT one thing, to use the tool that best fits that
one need. Otherwise you're wasting perfectly good Oils when
a Magic Marker will do.
well, yes, but in this case, there is no obvious reason that repos,
or resamp, with a matrix normal map would be slower than a bunch of
pak srcdimstart/dim end messages, esp if you want to stay in matrix
land.
v a d e //
www.vade.info
abstrakt.vade.info
couldn't we start collecting max/jitter cpu test patches? i have a
bunch of benchmark examples that does the same thing repeatedly in a
few different ways. i bet there are a lot of these laying around.
it'd be interesting to see some and we could all learn useful tricks i
think. say we just mailed them here with a [cputest] in the subject?
(or maybe i just have a hook up on optimising my videopatches - not
being spoiled with new computers)
_f0
On 19.10.2006, at 09:11, vade wrote:
> speaking of expensive, is there a list , official or not, of optimum
> methods for doing typical/useful things in jitter?
>
> ie, is mxform2D or repos going to be faster than srcdim for this
> particular function?
>
> is there a coherent thought process behind choosing optimal methods,
> or is it just trial and error with the objects?
#|
fredrikolofsson.com klippav.org musicalfieldsforever.com
|#
Hey
I tried your method of jit.dimmap, but that doesn't produce a mirror image. Isn't used to rotate the matrix, not invert it?
All the best
J
jit.dimmap @invert 1 0