Hi,

I started playing with Ipisoft Version 2 this morning and i faced a problem with Refit pose. The problem is that when i hit Refit pose buttong, the skin and the bone can not fit the actor. It always appears above the real actor (Before hitting Refit Pose button, i place the skin nicely with the actor already). Here is the screenshot:



After that, i opened the depth layer to check, and i realize that the video and the depth does not fit each other. So is that why the skin can not fit the actor in the video, because of the depth layer? Please help me with this.

Here is the picture including the video and the depth:



However, i put the above issuse aside and just start Tracking process. As i mention above, the skin does not fit the actor but the tracking process works fine. Ipi still track the actor's movement very well.



In short, is there any way to get the skin fits well with the actor in the video ?
Note:
I tried version 1.3 before and Refit Pose works very well. The skin fits perfectly with the actor

I have similar problem - but single Kinect config works just fine but can't make two of them work - refit pose move character somewhere else.

Does it still actual for the v2.0?

Important! Initial two seconds of video should contain only background and static objects. Presence of actor or any moving objects in the initial two seconds of recorded video is unacceptable. It’s because initial seconds of video are used to calculate static background necessary to separate actor from other objects of 3D scene.
http://wiki.ipisoft.com/User_Guide_for_Dual_Kinect_Sensor_Configuration#Recording_Actor.27s_Performance

No. In v2 background is calculated separately at the Background stage of iPi Recorder. And then it is stored in each subsequently recorded video file. So in the video itself, there is no need for static image at the beginning.
We will update wiki documentation later to reflect changes in the software.

thanks for all your feedbacks, but is there any way to fix the issue i mentioned above?

You can do nothing about that. And also you should not worry, because RGB image from depth sensors is used now only for reference purposes, not for computations.

Exact matching depth and RGB pictures requires knowledge of internal characteristics of individual device. It is strongly affected by the least differences in actual position and orientation of infrared and RGB sensors in the device. In future releases, we'll try to use maximum of information provided by device's driver for specific device instance. But currently we can use only one-fits-all formula. In v2 release, we updated this "formula" to provide the best matching among all the Kinect and Xtion instances we have. Unfortunately, for your specific device, this formula is not working well :(

THanks, im using one Kinect right now and i will upgrade to 2 Kinects system in the future. Will this issue cause any problems when i have 2 kinects?

This issue will not affect tracking for two Kinects either.
It only will affect tracking when we start using RGB channel. But for that we must implement correct depth-RGB matching, and this issue will vanish.

With Recorder 2.0.0.15, precise depth-color matching info is recorded in video, so images are perfect aligned in Studio.