I apologize for the delay -- one of your issues turned out to be a
tough bug to hunt down, but one that we had been seeking for awhile
now. There was a bug in Ctrax that would occasionally result in a
fly's position being reverted to 0,0 (or maybe inf,inf in older
versions), which is why the figure you sent earlier looked very
strange. Some of the flies would spontaneously move from their track
to 0,0 and back, leaving the jagged lines you noticed in the position
plots. This bug is fixed now, I think -- at least it works in the
movie you sent me. The fix will be released with Ctrax 0.1.6,
hopefully sometime in July (or anyone building from source can get the
fixed code from the repository now).
However, as you have realized, the arena setup you're using is not
particularly good for tracking. First, it's 3D, so flies will
frequently change shape and occlude each other, making them difficult
to track. Second, the lighting is inconsistent, so there are bright
areas and dark areas of the arena, and flies on camera will appear to
change size, shape, and orientation as they move from one lighting
condition to another. Our brains are so good at tracking that I
frequently find it difficult to remember how tricky of a problem it is
for computers, so we have to make it as easy as we can for them.
Fixing these lighting issues will lead to better tracking in general,
and perhaps the bug you were seeing would not arise as often in an
arena more conducive to Ctrax's tracking algorithm.
I couldn't actually replicate the problems you were seeing in
Fixerrors. Perhaps the tracking was so bad because of the bug I fixed,
that Fixerrors was not recognizing things correctly. In any case,
please let me know if you have those issues again and if you can help
me see the problem precisely in another dataset (screenshots might
help).
I believe the reason you couldn't open the AVI file is the same reason
other people reported, that some AVI reading was accidentally broken
in Ctrax 0.1.5.7 but is now fixed in 0.1.5.8. I had no trouble reading
the AVI you sent me.
Best,
JB
On Thu, Jun 16, 2011 at 5:16 AM, yukiko yamada ho
<yukiko.y...@gmail.com> wrote:
> Hi John,
> Thank you for your kind offer to have a look at our movie and related files.
> I have a movie segment and other related files that I uploaded at the MPI
> file sharing site. If you get a chance, please have a look. The info is at
> the end of this email.
> I've been trying to use the FixErrors GUI to correct some tracking errors.
> Sorry you have asked for fixed M-file, but due to the following problems,
> I'm only sending you the original tracking M-file.
>
> 1. for some reason I cannot open the avi file. So I've been opening sbfmf
> file.
> 2. Also, when I use the GUI, there are so many untracked flies. (the ones
> without colored circles?) from frame 1. For those, I cannot create a new
> trajectory since there is no error fixing option for that. Is that right? Or
> are there ways to handle this issue?
> 3. When I am setting up the tracking most flies are marked with circles so I
> assumed FixErrors GUI would show the same, but strangely no.
>
> Parameter setting I have for tracking must be the main problem. I've been
> playing around a bit changing some numbers in the settings, but not much
> improvement. Also, the arena design itself may be blamed for the poor
> tracking, as suggested by some people already. If you could have a look at
> the movie and let me know what you think, that would be super.
> Thank you so much for your help.
> Best,
> Yukiko
Hi Yukiko,
I or the delay -- one of your issues turned out to be a
--
You received this message because you are subscribed to the Google Groups "ctrax" group.
To post to this group, send email to ct...@googlegroups.com.
To unsubscribe from this group, send email to ctrax+un...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/ctrax?hl=en.
I have lots of AVI codecs installed on my machine, but ffmpeg says
your video is encoded as 'rawvideo bgr24', meaning an uncompressed
24-bit (color) format. Try encoding it in grayscale (8-bit format). It
may make a little sense that Ctrax will only read 1/3 of the frames,
since 8*3=24; I can't say for sure, but I suspect your problem is
there somewhere.
JB
On Tue, Jul 5, 2011 at 7:42 AM, yukiko yamada ho
I believe this problem will be fixed in Ctrax 0.1.6, but I'm still
testing that version to make sure that the fix doesn't break anything
else. Ctrax 0.1.6 also has other many changes and improvements over
0.1.5.x, and I'm hoping to release it by mid-August (though it could
be a few weeks later). I will announce the release on this mailing
list at the appropriate time.
It's quite possible that your MaxJumps problem is related to the 0,0
positions, but I can't say for sure. Unfortunately, the only thing I
can suggest is waiting until 0.1.6 fixes the 0,0 positions and then
trying those movies again.
JB