Hi,
I'm after a product similar to Smartnav/TrackIR (any tracking device really) that will allow 2 objects (2 dots) to be tracked simultaneously.
That is - imagine two people looking at a screen, and each having there own independant point tracking on screen (Note I am aware of the windows limitations of dual mice.)
Can any of the Trackir/smartnav products do this?
Track multiple dots simultaneously?
Re: Track multiple dots simultaneously?
Hello:
Technically both the SmartNAV and TrackIR can track mulitple dots at the same time and report their position. We have a specific product geared towards this type of custom application called OptiTrack. You can easliy develop your own applications, including controlling two cursors at the same time with the free SDK. Check it out at [url=http://www.optitrack.com,]www.optitrack.com,[/url] or just click on the link at the top of all our web pages.
Technically both the SmartNAV and TrackIR can track mulitple dots at the same time and report their position. We have a specific product geared towards this type of custom application called OptiTrack. You can easliy develop your own applications, including controlling two cursors at the same time with the free SDK. Check it out at [url=http://www.optitrack.com,]www.optitrack.com,[/url] or just click on the link at the top of all our web pages.
Re: Track multiple dots simultaneously?
Hi,
Thanks for the feedback - I've had a look round at optitrack and it looks very nice, if even overkill. It would be great if you could clarify a couple of points for me:
I'm essentially after two pointing devices being used simulataneously, pointing at the same screen. As you mention, technically this seems very straightforward for almost any of your hardware solutions (ignoring software). However, I notice in the demo videos of people using the systems that they tend to be very static - ie seated in front of a screen.
I'm curious if there are calibration issues when using just a single dot to track? For example, if someone was wearing a dot on a baseball cap and they left the room and returned, and moved the chair they were sitting in or even decided to stand up (ie now pointing at the screen from a different angle) would the system have to be recalibrated? Or is this the type of problem the vector expansion solves?
One of the reasons I am asking is beacuse I've seen several less customised solutions were software is used to analyse video from a standard usb camera to track a pointer, and it struggles if the pointers origin is moved in respect to the screen.
Thanks.
Thanks for the feedback - I've had a look round at optitrack and it looks very nice, if even overkill. It would be great if you could clarify a couple of points for me:
I'm essentially after two pointing devices being used simulataneously, pointing at the same screen. As you mention, technically this seems very straightforward for almost any of your hardware solutions (ignoring software). However, I notice in the demo videos of people using the systems that they tend to be very static - ie seated in front of a screen.
I'm curious if there are calibration issues when using just a single dot to track? For example, if someone was wearing a dot on a baseball cap and they left the room and returned, and moved the chair they were sitting in or even decided to stand up (ie now pointing at the screen from a different angle) would the system have to be recalibrated? Or is this the type of problem the vector expansion solves?
One of the reasons I am asking is beacuse I've seen several less customised solutions were software is used to analyse video from a standard usb camera to track a pointer, and it struggles if the pointers origin is moved in respect to the screen.
Thanks.
Re: Track multiple dots simultaneously?
Hello:
Good questions. When you track a single dot, the relative X-Y movement of it is translated to a cursor position, so when the user moves, the cursor position will also move. This can be an issue, but it depends a lot on the desired application. We have a "recenter" feature that will put the cursor in the center of the screen, we also have Absolute mode, where the screen edges are "hard", but what you may want is more of a Vector solution. Using Vector you can remove translation from any cursor motion, which can be quite helpful in setting an absolute center, you can also find a true vector intersection with the screen. In practical use we have found that just using rotation only data to position a cursor, using Vector, is quite nice and is a good method for control.
Hope this info helps, feel free to let me know more about your application and how you want the users to interact with the screen/cursor and perhaps I can narrow the recommendation.
Good questions. When you track a single dot, the relative X-Y movement of it is translated to a cursor position, so when the user moves, the cursor position will also move. This can be an issue, but it depends a lot on the desired application. We have a "recenter" feature that will put the cursor in the center of the screen, we also have Absolute mode, where the screen edges are "hard", but what you may want is more of a Vector solution. Using Vector you can remove translation from any cursor motion, which can be quite helpful in setting an absolute center, you can also find a true vector intersection with the screen. In practical use we have found that just using rotation only data to position a cursor, using Vector, is quite nice and is a good method for control.
Hope this info helps, feel free to let me know more about your application and how you want the users to interact with the screen/cursor and perhaps I can narrow the recommendation.
Re: Track multiple dots simultaneously?
Hi,
Thanks for the feedback. It sounds like tracking a single dot would definately not be sufficient, as it would require recalibration everytime somebody new used it or some one moved.
The very rough idea is an educational program to be used in training courses, aimed at kids. The idea being 2 kids on each station (in front of each monitor) each with a pointer. A fun element of competition where they solved problems on screen using the pointer as fast as possible. I'm looking at the hardware to get it working.
I originally looked at tracking with usb cameras and open source software, but these turned about to both inaccurate and the high usb datarates (full video being passed down the usb1 connection) and analysis in software took too much cpu time. I've seen reports of people using your system with tiny cpu overhead.
I also looked at adapting 'light guns' - however, pc versions of these are both hard to come by and extrememly poor quality, and tend to demand tv-out signal is used, whereas we are using 32" VGA CRT's. I did find a succesful version the worked with VGA, but sadly they do not offer real time tracking at all (they flash the screen to locate the cursor) nor did they work with out monitors.
As essentially we need to identify where something is being pointed, a bare minium of 3 locations (camera, two dots) would be needed to triangulate the aim. Looking at your vector addition I see you have something like this, if not better - can the software calcualte where the unit is being aimed at the screen?. We don't need the 6 degrees of freedom, but if each one could be calibrated once, and then always know its location/aim when moved in and out of the cameras field of view that would be promising. Is this currently possible? And can two vector clips be tracked simulataneously by 1 camera (prefereably without getting confused!)?
I appreciate your time in answering all these questions.
Thanks for the feedback. It sounds like tracking a single dot would definately not be sufficient, as it would require recalibration everytime somebody new used it or some one moved.
The very rough idea is an educational program to be used in training courses, aimed at kids. The idea being 2 kids on each station (in front of each monitor) each with a pointer. A fun element of competition where they solved problems on screen using the pointer as fast as possible. I'm looking at the hardware to get it working.
I originally looked at tracking with usb cameras and open source software, but these turned about to both inaccurate and the high usb datarates (full video being passed down the usb1 connection) and analysis in software took too much cpu time. I've seen reports of people using your system with tiny cpu overhead.
I also looked at adapting 'light guns' - however, pc versions of these are both hard to come by and extrememly poor quality, and tend to demand tv-out signal is used, whereas we are using 32" VGA CRT's. I did find a succesful version the worked with VGA, but sadly they do not offer real time tracking at all (they flash the screen to locate the cursor) nor did they work with out monitors.
As essentially we need to identify where something is being pointed, a bare minium of 3 locations (camera, two dots) would be needed to triangulate the aim. Looking at your vector addition I see you have something like this, if not better - can the software calcualte where the unit is being aimed at the screen?. We don't need the 6 degrees of freedom, but if each one could be calibrated once, and then always know its location/aim when moved in and out of the cameras field of view that would be promising. Is this currently possible? And can two vector clips be tracked simulataneously by 1 camera (prefereably without getting confused!)?
I appreciate your time in answering all these questions.
Re: Track multiple dots simultaneously?
Hello:
Thanks for explaining your application, I think I understand what you will need better now.
My recommendation is for two OptiTrack units, one for each monitor / user. The Field of View is not that large for each camera, so fitting in two users is not a good option at this time. OptiTrack does come with a Vector license, so you don't need to purchase that seperately. Also, it should allow for a single calibration and then it will be very repeatable. You could even mount the Vector clip on a "gun" type device. Vector is capable of detecting the intersection of the screen with the pointing of the clip, if that makes sense. I think this is a good fit for your application.
Let me know how I can help further.
Thanks for explaining your application, I think I understand what you will need better now.
My recommendation is for two OptiTrack units, one for each monitor / user. The Field of View is not that large for each camera, so fitting in two users is not a good option at this time. OptiTrack does come with a Vector license, so you don't need to purchase that seperately. Also, it should allow for a single calibration and then it will be very repeatable. You could even mount the Vector clip on a "gun" type device. Vector is capable of detecting the intersection of the screen with the pointing of the clip, if that makes sense. I think this is a good fit for your application.
Let me know how I can help further.
Re: Track multiple dots simultaneously?
Hi,
Thanks for that - it sounds very promising. It sounds like the vector unit could fit the bell....
I'm interested to hear about the field of view - 32 degrees - what sort of range would the vector units work at? The design is for 1 monitor to be used by 2 people (although in theory we can duplivate the monitor output to 2 screens). Obviously close up this is an issue, but if the vector units were further back would it fit within range?
Thanks for that - it sounds very promising. It sounds like the vector unit could fit the bell....
I'm interested to hear about the field of view - 32 degrees - what sort of range would the vector units work at? The design is for 1 monitor to be used by 2 people (although in theory we can duplivate the monitor output to 2 screens). Obviously close up this is an issue, but if the vector units were further back would it fit within range?
Re: Track multiple dots simultaneously?
Hello:
With the 32 degree horizontal FOV it is best to have the camera about arms lenth from the user, but it can go up to 6 feet.
I would still recommend some sort of setup that involved two cameras, or perhaps two montiors so that each user has their own setup. It would also be best if the Fields of View did not overlap.
With the 32 degree horizontal FOV it is best to have the camera about arms lenth from the user, but it can go up to 6 feet.
I would still recommend some sort of setup that involved two cameras, or perhaps two montiors so that each user has their own setup. It would also be best if the Fields of View did not overlap.
Re: Track multiple dots simultaneously?
Well, I think two monitors are out, simply for cost and space reasons (these are currently large 28" CRT's).
Two cameras are possible if this is required, although it would be very hard to avoid any overlap.
From your suggestion I take it there would be an issue with tracking two vector units from 1 camera?
Two cameras are possible if this is required, although it would be very hard to avoid any overlap.
From your suggestion I take it there would be an issue with tracking two vector units from 1 camera?
Re: Track multiple dots simultaneously?
Hello:
Two Vector units from one camera will not work well, at least in the current configuration. I don't think you would have enough field of view to make it work well for 2 people anyway. I do think you can have two cameras, pointed slightly away from eachother and have the fields of view overlap while using Vector, it should be fine. The software will lock onto one clip and if parts of the other one show up it won't cause an error. Worth a try for sure.
Two Vector units from one camera will not work well, at least in the current configuration. I don't think you would have enough field of view to make it work well for 2 people anyway. I do think you can have two cameras, pointed slightly away from eachother and have the fields of view overlap while using Vector, it should be fine. The software will lock onto one clip and if parts of the other one show up it won't cause an error. Worth a try for sure.