| ipisoft.com https://forum.ipisoft.com/ |
|
| 8 cam and high marker misdetection https://forum.ipisoft.com/viewtopic.php?f=5&t=9130 |
Page 1 of 1 |
| Author: | mizgo [ Wed Aug 06, 2014 4:22 am ] |
| Post subject: | 8 cam and high marker misdetection |
Hello. I am student of VFX. I'm working on my motion captured short film. I am thinking about buying your system. Because I have 10 PS3 eyes, my own gyroscopes, ... I used them for my own capturing system. I have downloaded trial version of ipi studio. I love possibility to record 8 cameras in trial version, without ability to export or save to full format. Its great for testing. To my problem: I cant get proper calibration of 8 cameras. Marker misdetection are about 50%. On some cameras higher, on some lower. Marker is always visible and tracked on all cameras. I dont have checked Auto detect initial camera positions. I have changed environment, movement of marker, lenght of sequence and camera positions. (full circle/half circle - clockwise/counterclockwise/random). Ive tried multiple calibrations. Manual position of cameras, ...I have only one "success" calibration with highest misdetection about 24%. When I use same recorded sequence for 6 cameras, calibration is much better. Maximum misdetection about 18% for first try, witout my intervention. Same sequence for four cameras is heaven. Max misdetection about 8%. I ask you for help or advice on how to proceed to calibrate the eight cameras. Thank you for your early reply. Michal. |
|
| Author: | pas [ Wed Aug 06, 2014 7:03 am ] |
| Post subject: | Re: 8 cam and high marker misdetection |
Hello, yes, for 8 cameras you should not check "Autodetect initial cameras positions" option. But you need to manually set approximate mutual positions of cameras. The most important - order of cameras should be exactly the same as it reality. The closer initial cameras position to actual, the higher the probability to get the correct result. We'll make this fully automatic in the next version (v3) which hopefully will come out in September. But meanwhile you need to take care of this manually. If you are sure that initial position is good, but still no luck with calibration, please send us your calibration video file via any file-sharing service. |
|
| Author: | mizgo [ Wed Aug 06, 2014 7:15 am ] |
| Post subject: | Re: 8 cam and high marker misdetection |
Thanks a lot. I was not sure, if manual set of initial position is proper way to go. I now "do not checkt Autodetect initial cameras positions" from other posts here in forum. But i dont now, if is good idea to position it by hand and try to best manual match. I am working on it. If I cant solve that, I send recorded file and project file. Thank you once again. |
|
| Author: | mizgo [ Wed Aug 06, 2014 3:37 pm ] |
| Post subject: | Re: 8 cam and high marker misdetection |
I do not have satisfactory outcome. It is better to have more complex calibration sequence (little longer time with more space coverage) for 8 cam with 6x6 meter shooting place? Or "cage method" is good enough for all scenarios? |
|
| Author: | pas [ Thu Aug 07, 2014 10:45 pm ] |
| Post subject: | Re: 8 cam and high marker misdetection |
3000 frames sequence should be enough for calibration. The marker motions ideally should cover the whole capture area, but it is more important to ensure that marker is visible in all cameras as much as possible. Please share your calibration video with us (e-mail to tech support address ipi@ipisoft.com) |
|
| Page 1 of 1 | All times are UTC - 8 hours [ DST ] |
| Powered by phpBB® Forum Software © phpBB Group http://www.phpbb.com/ |
|