Jumping points
Jumping points
Hello,
during my first capture test with ARENA expression I encountered the following problem:
The point cloud shown during live capture mode seems incorrect.
Some of the points are jumping around and the arena software is unable to connect them properly.
The camera calibration was successful, but the marker detection during the last step of the face wizard, which was also said to be successful, looked wrong.
I recorded a short video clip to show you what I am talking about:
[video:youtube]http://www.youtube.com/watch?v=QnzFfJSNNw4[/video]
As you can see, the markers are shown correct in the cameras views.
I tried to recalibrate a few times, but it doesn't help - the confusing result of the marker detection remains.
We attached the markers to the face exactly like in the example - here is a photo:
Can you give me any advise on how to solve this problem?
Thanks a lot in advance!
Jan
during my first capture test with ARENA expression I encountered the following problem:
The point cloud shown during live capture mode seems incorrect.
Some of the points are jumping around and the arena software is unable to connect them properly.
The camera calibration was successful, but the marker detection during the last step of the face wizard, which was also said to be successful, looked wrong.
I recorded a short video clip to show you what I am talking about:
[video:youtube]http://www.youtube.com/watch?v=QnzFfJSNNw4[/video]
As you can see, the markers are shown correct in the cameras views.
I tried to recalibrate a few times, but it doesn't help - the confusing result of the marker detection remains.
We attached the markers to the face exactly like in the example - here is a photo:
Can you give me any advise on how to solve this problem?
Thanks a lot in advance!
Jan
Re: Jumping points
How do you have the cameras setup?
How far away from each other?
It looks like some of the camera views, have the camera setup behind the ears, correct?
How far away from each other?
It looks like some of the camera views, have the camera setup behind the ears, correct?
Re: Jumping points
Hi,
I'm having exactly the same problem. Calibration results are great to excellent for each of my 6 cameras. I tried 2 different setups, one where the cameras were about 40-50cm apart and another one where they were closer, about 25-35cm. Both give me the same very disappointing results. I also looked at each of the cameras and usually I have get pretty good images with almost all of the images all points identifiable in at least 4 cameras - it's just the resulting point cloud that's so bad, it doesn't even look like a surface anymore.
Is there anything I can do to improve this? Is my setup wrong?
thanks a lot
Chris
I'm having exactly the same problem. Calibration results are great to excellent for each of my 6 cameras. I tried 2 different setups, one where the cameras were about 40-50cm apart and another one where they were closer, about 25-35cm. Both give me the same very disappointing results. I also looked at each of the cameras and usually I have get pretty good images with almost all of the images all points identifiable in at least 4 cameras - it's just the resulting point cloud that's so bad, it doesn't even look like a surface anymore.
Is there anything I can do to improve this? Is my setup wrong?
thanks a lot
Chris
Re: Jumping points
OK, the cameras should be atleast 2-3 feet apart (61-91cm), in an arc where the cameras can see the actors face.
Also, if you can send in a sample data capture (support@naturalpoint.com) then we will take a look and see if there are any clues.
Also, if you can send in a sample data capture (support@naturalpoint.com) then we will take a look and see if there are any clues.
Re: Jumping points
You are right the cameras are too close together.
Here is a picture of my test setup:
The lower left and right cameras are on one axis with the ears I guess.
The table I grabbed for testing is obviously to small.
I will change the setup using a wider arc, I have to integrate the system into the mocap studio anyway.
I thought about building a rig for the cameras with fixed positions, so the students cant't do anything wrong
Thanks for your advice!
Jan
Here is a picture of my test setup:
The lower left and right cameras are on one axis with the ears I guess.
The table I grabbed for testing is obviously to small.
I will change the setup using a wider arc, I have to integrate the system into the mocap studio anyway.
I thought about building a rig for the cameras with fixed positions, so the students cant't do anything wrong
Thanks for your advice!
Jan
Re: Jumping points
Hello,
we're 2 students working with a 6 Camera Setup. We're encountering the same jumping points as HAW posted. It seems that the calculation of the 3D Space seems incorrect, although we achieved excellent calibration (at least 5 out of 6 cameras). We tried altering the calculation settings but it showed no difference, so we keep trying it with 100 Data Points and min 5 Cameras. Tracking works fine with the wand, but when we place our "Face-Track Model" in the Capture Space, ARENA shows a lot of jumping points.
The Cameras are placed in a semicircle about 2 feet away from the face.
[img:center]http://www.imagebanana.com/img/6m97mzd/IMG_3943.JPG[/img]
[img:center]http://www.imagebanana.com/img/yloi6wrx/IMG_3949.JPG[/img]
First we tried applying the facial markers on an actor, as we recognized that we had to do lots of calibrations, we took a doll.
We hope you can help us, so that we can get on with our work
we're 2 students working with a 6 Camera Setup. We're encountering the same jumping points as HAW posted. It seems that the calculation of the 3D Space seems incorrect, although we achieved excellent calibration (at least 5 out of 6 cameras). We tried altering the calculation settings but it showed no difference, so we keep trying it with 100 Data Points and min 5 Cameras. Tracking works fine with the wand, but when we place our "Face-Track Model" in the Capture Space, ARENA shows a lot of jumping points.
The Cameras are placed in a semicircle about 2 feet away from the face.
[img:center]http://www.imagebanana.com/img/6m97mzd/IMG_3943.JPG[/img]
[img:center]http://www.imagebanana.com/img/yloi6wrx/IMG_3949.JPG[/img]
First we tried applying the facial markers on an actor, as we recognized that we had to do lots of calibrations, we took a doll.
We hope you can help us, so that we can get on with our work
Re: Jumping points
Can you email support with copies of your PT2, calibration, skeleton files, as well as a screen shot of the wanding page, with the camera views.
Re: Jumping points
Here's our screen capture of the jumping points.
FYI: We use a doll with applied head markers, so that the markers should be displayed non moving and still they jump around.
[video:youtube]http://www.youtube.com/watch?v=jkMhoRZeLFs[/video]
FYI: We use a doll with applied head markers, so that the markers should be displayed non moving and still they jump around.
[video:youtube]http://www.youtube.com/watch?v=jkMhoRZeLFs[/video]
Re: Jumping points
I'm having this problem, has there ever been a fix or solve?
Re: Jumping points
Which version of the Arena software are you using?
Are you using thwe 3-point calibration, or single point calibration?
How large is your setup area?
Which lenses do you have installed on your cameras?
Are you using thwe 3-point calibration, or single point calibration?
How large is your setup area?
Which lenses do you have installed on your cameras?