FreeTrack Forum - 3-point cap callibration?

Home

Welcome, you're not connected. ( Log in - Register )

FreeTrack Forum > FreeTrack : English Forum > Support : Tracking System > 3-point cap callibration?

RSS > 3-point cap callibration?

Deimos	#1 08/11/2007 - 01h14 Top
Class : Beta Tester Posts : 120 Registered on : 07/11/2007 Off line	When i got freetrack 2.1 i was really pleased to see that it can use a 3 point cap. So i quickly converted my 4-led cap into a 3 led one by covering the middle diode. However after inputting the correct model dimensions into freetrack i found out that this setup doesn't work too well - when i turn my head in y axis, freetrack also generates a significant y axis movement. I noticed that the 3-point clip can be calibrated to eliminate this effect. Is there any chance of having such calibration option for the cap as well?
Kestrel	#2 08/11/2007 - 08h32 Top
Class : Webmaster (admin) Posts : 780 Registered on : 13/07/2007 Off line	Deimos @ 08/11/2007 - 01h14 a dit: So i quickly converted my 4-led cap into a 3 led one by covering the middle diode. This isn't recommended, the 4 point model doesn't have enough depth with the middle led disabled for good 3 point tracking. when i turn my head in y axis, freetrack also generates a significant y axis movement. Do you mean you get up/down movement when you yaw? Support FreeTrack
Deimos	#3 08/11/2007 - 12h33 Top
Class : Beta Tester Posts : 120 Registered on : 07/11/2007 Off line	Well, it isn't a typical clip-on setup: The front LED's are attached directly to the cap, and the side LEDs are closer apart (55mm, not 75mm like in reference design), and moved more to the front than in the cap shown in freetrack manual - the distance between them and the front led in z axis is 15mm, not 50 as recommended. So even after covering the central led and leaving side and top leds, my model has more depth than the reference design converted in the same manner, but less width. and less height. I belive that my problem comes from the fact that the entire rig is moved much more towards front end of the cap than designs proposed in the manual, and the axis of my head's rotation isn't where freetrack software expects it. But that's jus a guess of course. [quote]Do you mean you get up/down movement when you yaw?[/qutoe] That's right - small movements work fine, but when i turn the head more about 20 degrees or more, the game head turns all the way down, and sometimes even ends up upside-down. Also, the tracking doesn't seem to be as smooth as with the 4-point cap, especially with sensitivity turned down. Well, i think i'll just have to stick to 4-point mode now, and follow the dimensions from the manual when i build the next cap
Kestrel	#4 08/11/2007 - 13h28 Top
Class : Webmaster (admin) Posts : 780 Registered on : 13/07/2007 Off line	Deimos @ 08/11/2007 - 12h33 a dit: That's right - small movements work fine, but when i turn the head more about 20 degrees or more, the game head turns all the way down, and sometimes even ends up upside-down I think your model might have too much depth, I must confess I use the clip model and my experimentation with the cap models is limited. This person: http://www.youtube.com/watch?v=epF5xM43_d0 had a model that was too shallow, causing problems that sound similar to yours, they solved it by using the default measurements. Also, the tracking doesn't seem to be as smooth as with the 4-point cap, especially with sensitivity turned down. ??? Don't you mean with the sensitivity turned up? Support FreeTrack
Deimos	#5 08/11/2007 - 16h55 Top
Class : Beta Tester Posts : 120 Registered on : 07/11/2007 Off line	Don't you mean with the sensitivity turned up? Nope. When i use the 3 point model on the same sensitivity settings and response curves i used with 4-point model, it is way too sensitive - turnig my heat about 10-15 degrees makes the virtual head look all the way back, and bigger movements can cause weird effects like the virtual head turning upside-down. However after decreasing the sensitivity i found that while possible movement of virtual head is a bit more sane, it becomes highly inaccurate. Then again, since it works fine for quite a few people, it's probably caused by wrong LED positioning on the cap in respect to its axis of rotation.
Kestrel	#6 09/11/2007 - 01h40 Top
Class : Webmaster (admin) Posts : 780 Registered on : 13/07/2007 Off line	Deimos @ 08/11/2007 - 16h55 a dit: When i use the 3 point model on the same sensitivity settings and response curves i used with 4-point model, it is way too sensitive The sensitivities are different but are just constant multipliers that have nothing to do with the tracking itself. However after decreasing the sensitivity i found that while possible movement of virtual head is a bit more sane, it becomes highly inaccurate. It would be good if you could make a small video like the one I linked before. Support FreeTrack
Deimos	#7 09/11/2007 - 17h02 Top
Class : Beta Tester Posts : 120 Registered on : 07/11/2007 Off line	http://www.youtube.com/watch?v=WnaOu4oqywE There's the video - first it shows the 4-point cap behavior, which is just fine. Later i switched to 3-point cap mode. All head translations in this mode except forward/back cause the game head not only to translate, but also rotate. Pitching up/down works fairly well (except that it's way too sensitive). But yawing left/right is the worst as you can see on the video. Before the end i decreased the sensitivity, and i don't know if it's visible that well in the video, but the virtual head has very poor precision - it jumps between quite widely spaced positions. But now, that i think about it and look more closely at camera footage, it seems that you were right - the top led should be placed eighter lower, or moved much more to the back, because significant head movements cause only small led position shift on the captured video. However some kind of additional model calibration, even a rough one, like setting the distance between lower leds and the neck measured vertically (and therefore horizontal axes of rotation - for pitch and roll) and front led and the center or the head measured horizontally (vertical axis of rotation - for yaw) would be nice to see. It should prevent freetrack from confusing translation and rotation movements and therefore should allow for more precise control with many various setups, not only the reference designs. Edited by Deimos on 09/11/2007 at 17h04.

FreeTrack Forum > FreeTrack : English Forum > Support : Tracking System > 3-point cap callibration?

> Stats

1 user(s) connected during the last 10 minutes (0 member(s) and 1 guest(s)).