.pt2/.pt3 format specification?
.pt2/.pt3 format specification?
Hi,
I would like to ask whether NaturalPoint would be willing to provide the specification of the .pt2 and especially .pt3 file format.
The background is the following: I have a large number of tracking errors in my data recorded with Arena 1.7.1, an example of which can be seen in the screenshot below.
This kind of error looks like it should be easy enough to fix automatically, so I would like to write a program that reads the .pt3 data, fixes these errors, and writes a valid .pt3 file again, which I can then load into arena for further processing.
Thank you,
Dietmar
I would like to ask whether NaturalPoint would be willing to provide the specification of the .pt2 and especially .pt3 file format.
The background is the following: I have a large number of tracking errors in my data recorded with Arena 1.7.1, an example of which can be seen in the screenshot below.
This kind of error looks like it should be easy enough to fix automatically, so I would like to write a program that reads the .pt3 data, fixes these errors, and writes a valid .pt3 file again, which I can then load into arena for further processing.
Thank you,
Dietmar
-
- Posts: 59
- Joined: Tue Nov 20, 2012 12:40 pm
Re: .pt2/.pt3 format specification?
Dietmar,
We are discussing this internally right now and will get back to you shortly.
We are discussing this internally right now and will get back to you shortly.
Re: .pt2/.pt3 format specification?
Hi,
I would also be highly interested by this.
Best regards,
Huseyin
I would also be highly interested by this.
Best regards,
Huseyin
-
- Posts: 59
- Joined: Tue Nov 20, 2012 12:40 pm
Re: .pt2/.pt3 format specification?
At this time we are unable to give out specifications to our proprietary file formats.
-Brent Mason
-Brent Mason
Re: .pt2/.pt3 format specification?
That's a pity. Thanks for considering though!
Do you have other suggestions, what one can do whithin ARENA to deal with these kinds of errors?
Settings for trajectorization, or anything else?
Do you have other suggestions, what one can do whithin ARENA to deal with these kinds of errors?
Settings for trajectorization, or anything else?
-
- Posts: 1365
- Joined: Fri Jun 27, 2008 11:29 am
- Location: Corvallis, Oregon
Re: .pt2/.pt3 format specification?
If you're looking to batch process your captures, it should be reasonable to handle in MotionBuilder via scripting. You could just export the C3D from Expression and process it in MoBu.
Marketing Manager
TrackIR | OptiTrack
TrackIR | OptiTrack
Re: .pt2/.pt3 format specification?
Yes, that's what I'm doing afterwards anyways. However, as far as correcting the capture data goes, it is very useful to use Expression's feature to project the marker data onto the grayscale video.
I'm also wondering if maybe I'd just need to set some parameters for trajectorization differently. If you have a look at this short video (15 MB):
https://dl.dropbox.com/s/bu3o4s2su45z8g ... s.avi?dl=1
Strange things happen here in the tracking, but if I display the 3D markers only, they look correct. They are "only" confused among each other. I know there is the "swap tracks" tool that one can use after trajectorization, but I didn't succedd in really repairing the data with that. Maybe I'm not using it right.
Does anyone have any suggestions what I could try to get better trajectories out of this data?
Thanks, Dietmar
I'm also wondering if maybe I'd just need to set some parameters for trajectorization differently. If you have a look at this short video (15 MB):
https://dl.dropbox.com/s/bu3o4s2su45z8g ... s.avi?dl=1
Strange things happen here in the tracking, but if I display the 3D markers only, they look correct. They are "only" confused among each other. I know there is the "swap tracks" tool that one can use after trajectorization, but I didn't succedd in really repairing the data with that. Maybe I'm not using it right.
Does anyone have any suggestions what I could try to get better trajectories out of this data?
Thanks, Dietmar
-
- Posts: 1365
- Joined: Fri Jun 27, 2008 11:29 am
- Location: Corvallis, Oregon
Re: .pt2/.pt3 format specification?
Oh, wow. That's a highly unusual amount of track swapping--especially with the solid marker data you're working with. I'm going to pass this along to one of our developers to examine.
- Seth
- Seth
Marketing Manager
TrackIR | OptiTrack
TrackIR | OptiTrack
-
- Posts: 1365
- Joined: Fri Jun 27, 2008 11:29 am
- Location: Corvallis, Oregon
Re: .pt2/.pt3 format specification?
Dietmar--do you have your source files available to send us? Ideally the .pt2 (instead of .pt3), so we have it pre-labeling.
If so, you can email me at seth[at]naturalpoint[dot]com for FTP instructions.
If so, you can email me at seth[at]naturalpoint[dot]com for FTP instructions.
Marketing Manager
TrackIR | OptiTrack
TrackIR | OptiTrack
-
- Posts: 1365
- Joined: Fri Jun 27, 2008 11:29 am
- Location: Corvallis, Oregon
Re: .pt2/.pt3 format specification?
[quote=dsc]Yes, that's what I'm doing afterwards anyways. However, as far as correcting the capture data goes, it is very useful to use Expression's feature to project the marker data onto the grayscale video.
I'm also wondering if maybe I'd just need to set some parameters for trajectorization differently. If you have a look at this short video (15 MB):
https://dl.dropbox.com/s/bu3o4s2su45z8g ... s.avi?dl=1
Strange things happen here in the tracking, but if I display the 3D markers only, they look correct. They are "only" confused among each other. I know there is the "swap tracks" tool that one can use after trajectorization, but I didn't succedd in really repairing the data with that. Maybe I'm not using it right.
Does anyone have any suggestions what I could try to get better trajectories out of this data?
Thanks, Dietmar[/quote]
Hi Dietmar,
We had a chance to review your data. The problem was primarily due to the face template not aligning toward the actual front direction of the actor. The problems were amplified slightly by the residual setting being a bit too high, which we recommend reducing to 2mm stabilize the 3D points.
You can see how to address these issues here:
http://media.naturalpoint.com/ftp/npupl ... ection.wmv
Here are the steps:
1. Reduce the Residual in the Point Cloud tab to 2mm
2. Select the Rotation button from the toolbar
3. Select the head rigid body by selecting the marker connecting lines for the head
4. Inspect from the top. The z axis (red ring) is pointing off to the side, not exactly aligning with the face orientation. You'll need to adjust the 4 head markers so that the z axis better aligns with nose marker.
5. Hide 3D markers from the Display tab
6. Select the 4 head markers and rotate them to better align the z axis with the nose. In order to see the rotation result, advance one frame and reselect head rigid body. Repeat if necessary.
7. Save the corrected face template.
I'm also wondering if maybe I'd just need to set some parameters for trajectorization differently. If you have a look at this short video (15 MB):
https://dl.dropbox.com/s/bu3o4s2su45z8g ... s.avi?dl=1
Strange things happen here in the tracking, but if I display the 3D markers only, they look correct. They are "only" confused among each other. I know there is the "swap tracks" tool that one can use after trajectorization, but I didn't succedd in really repairing the data with that. Maybe I'm not using it right.
Does anyone have any suggestions what I could try to get better trajectories out of this data?
Thanks, Dietmar[/quote]
Hi Dietmar,
We had a chance to review your data. The problem was primarily due to the face template not aligning toward the actual front direction of the actor. The problems were amplified slightly by the residual setting being a bit too high, which we recommend reducing to 2mm stabilize the 3D points.
You can see how to address these issues here:
http://media.naturalpoint.com/ftp/npupl ... ection.wmv
Here are the steps:
1. Reduce the Residual in the Point Cloud tab to 2mm
2. Select the Rotation button from the toolbar
3. Select the head rigid body by selecting the marker connecting lines for the head
4. Inspect from the top. The z axis (red ring) is pointing off to the side, not exactly aligning with the face orientation. You'll need to adjust the 4 head markers so that the z axis better aligns with nose marker.
5. Hide 3D markers from the Display tab
6. Select the 4 head markers and rotate them to better align the z axis with the nose. In order to see the rotation result, advance one frame and reselect head rigid body. Repeat if necessary.
7. Save the corrected face template.
Marketing Manager
TrackIR | OptiTrack
TrackIR | OptiTrack