| ipisoft.com https://forum.ipisoft.com/ |
|
| Another way to do a calibration https://forum.ipisoft.com/viewtopic.php?f=6&t=5140 |
Page 1 of 1 |
| Author: | rosemma [ Mon Jun 20, 2011 12:57 am ] |
| Post subject: | Another way to do a calibration |
The environment where we are working is not a clear space, then we cannot trace the area with Maglite as you suggested. Is it possible to calibrate it using the misures to modify directly the xmls file? Rosemma |
|
| Author: | Greenlaw [ Mon Jun 20, 2011 8:42 am ] |
| Post subject: | Re: Another way to do a calibration |
While it's best to have an uncluttered space, in many situations you can use a area that isn't 'clear' if you have also recorded some clean video to help iPi separate the background from the performer. (Many examples of successful tracks using 'unclear' backgrounds can be found elsewhere in the user gallery section.) The main thing is to avoid reflective surfaces (windows, glossy floors, large stainless steel appliances and fixtures, etc.,) because they may confuse the tracking system with distracting reflected motion or lights. Even an active screen saver on a computer monitor can be mistaken for a 'maglite'. The point is, however, you can usually correct these types of problems if you know what to look for. It's also important to light the performer as evenly as possible, because dark shadows moving across the body may also confuse the tracker. That said, it is possible to construct the scene manually. An artist I work with had done this before, but the result was not as accurate as allowing iPi create its own 3D environment. IMO, I would try to adapt the physical shoot space to be more iPi-friendly as it's best to allow iPi Studio to generate it's own 3D environment. If this isn't possible, I would find a different space (I tried three spaces before deciding on an ideal space,) or I would switch to a Kinect which works in almost any small area even if it isn't 'clear' and isn't affected by shadows. G. |
|
| Author: | Michael Nikonov [ Mon Jun 20, 2011 2:29 pm ] |
| Post subject: | Re: Another way to do a calibration |
Your description of the problem is not clear. Can you please provide more details. What particular step of calibration is not working in your environment? Can you provide a screenshot to illustrate the problem? |
|
| Author: | Michael Nikonov [ Tue Jun 21, 2011 9:30 am ] |
| Post subject: | Re: Another way to do a calibration |
I received your illustration. Your environment is indeed very unusual and interesting. I've sent you some suggestions by email. For those of our forum members who are curious about what is going on, I am not allowed to go into details but in general Rosemma works on motion capture of a human interacting with some kind of mechanical equipment. |
|
| Author: | rosemma [ Wed Jun 22, 2011 12:19 am ] |
| Post subject: | Re: Another way to do a calibration |
Thank Michael for your discretion. Thank Greenlaw for your suggestion but we can't choose another area :-( Rosemma |
|
| Page 1 of 1 | All times are UTC - 8 hours [ DST ] |
| Powered by phpBB® Forum Software © phpBB Group http://www.phpbb.com/ |
|