Back to the main page.
Bug 1625 - ft_databrowser: plotting components is terribly slow
| Status | CLOSED FIXED |
| Reported | 2012-08-02 15:52:00 +0200 |
| Modified | 2013-01-17 10:40:01 +0100 |
| Product: | FieldTrip |
| Component: | plotting |
| Version: | unspecified |
| Hardware: | PC |
| Operating System: | Windows |
| Importance: | P3 normal |
| Assigned to: | Roemer van der Meij |
| URL: | |
| Tags: | |
| Depends on: | |
| Blocks: | |
| See also: |
Roemer van der Meij - 2012-08-02 15:52:00 +0200
Even when there are just 10 to plot, maybe the code can be optimized.
Eelke Spaak - 2012-08-02 15:58:22 +0200
I think this might be because of ft_checkdata, see bug 1520.
Roemer van der Meij - 2012-08-02 16:04:04 +0200
Interesting bug, the illustrious checkdata. It's only called once at the start in the databrowser though.
Johanna - 2012-08-02 16:06:49 +0200
even if speed-up can't be found, at least to avoid confusion to the user, can you set it so that the time-series appear in their final position from the start (rather than displacing/squeezing, once the topos appear)?
Roemer van der Meij - 2012-08-02 16:27:13 +0200
That can be easily achieved by removing the 'drawnow' call, which I assume was only there for development purposes. (removed it now). I also found the main reason for the delay. In plot_topo, there is a bit of code to prevent the mask from being recomputed, but that bit fails because the position changes. I'll make it into a new bug. Oh no, so many bugs! :)
Roemer van der Meij - 2012-08-02 16:29:18 +0200
Bug 1626 for reference.
Roemer van der Meij - 2012-08-02 17:16:56 +0200
I improved ft_plot_topo a little, the calls inside the databrowser are now much fast (mask-previosness-detection slightly improved). However, it was still slow, most of the time is being consumed by interpolating the data. I lowered the requested gridscale/resolution that the databrowser uses a bit, it's much faster now. The change in resolution is barely detectable on my 24" monitor, so it should be fine?