[Motmot] fview -> FastImage -> Framewave problem

[Jack Keegan]

Hi Andrew,

Thanks for your reply....

On Mon, May 24, 2010 at 4:42 AM, Andrew Straw <astraw at caltech.edu> wrote:

>
> I apologize for the difficulties you've been having on Mac.

Thank you for open sourcing it :)


> A couple of thoughts that are hopefully helpful:
>
> First, if you don't need flytrax or other plugins that use it, you don't
> need FastImage. FastImage is only used for realtime image analysis. Also,
> there would be other ways to do realtime image analysis (e.g. OpenCV and
> their new Python wrappers).
>

I was actually hoping to use it to track a set of moving LEDs attached to
the end of a boom in order to monitor deflection. I was thinking flytrax
would be a good place to start with this.
Thanks for the OpenCV recommendation also. I skipped over it yesterday but
will have a look if I fail with Motmot.


> Second, although I have had FastImage working on Mac, I believe it was with
> the Intel IPP rather than Framewave. I have two branches of FastImage -- one
> that uses the open source Framewave and the other than uses the closed
> source (but inexpensive) Intel IPP. If you start having real troubles with
> Framewave on Mac, try IPP. (By the way, I agree with your assessment of the
> issue -- your first error in the output you included is indeed that the
> compiler can't find a framewave header file.) You can download the source
> for the IPP version of FastImage at
> http://github.com/motmot/fastimage/tarball/release/0.5.4.ipp
>
> Thanks for that. I'll certainly keep it in mind. It's a student project
though so cash is pretty tight :-/


> Third, to figure out if this software is suitable for you without spending
> all this time compiling from source, I recommend trying my precompiled
> Ubuntu binaries. At least this will give you an idea of whether the software
> will be useful for you before investing too much more time in the
> compilation dance.
>
Yes I was hoping to do this today but have other things to get through. I'm
hoping the target host for the experiment will be Ubuntu but I develop on a
mac. I suppose I could try a VM first, provided it recognises the camera.

The thing is, I actually need to track with two cameras so I can
more accurately measure deflection. I get the feeling I may not be able to
do this with flytrax? I could possibly use ctrax on prerecorded files but it
would be good to be able to just store the data rather than the video files.

I'll hopefully be able to give the Ubuntu version a go tomorrow then. As I'm
new to python it's good to know at least I wasn't going off on the wrong
track.

Thanks very much for that.

Cheers,

Jack
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://code.astraw.com/pipermail/motmot/attachments/20100524/64025005/attachment.htm