Main > Main Forum
A Closer Look At The AimTrak (with video goodness)
Neverending Project:
I made a few videos with hopes to show everyone what the device is like in real use. I should give a little background information on my setup, so you can compare if you need. My AimTrak is installed in an Act Labs gun, with the camera PCB right at the tip of the gun barrel. My screen is a 29" VGA monitor (running at 800x600) in a stand-up arcade cabinet, and the LED board is mounted in the cardboard bezel an inch or two above the top edge of the screen. For normal lightgun use, I would stand about 5 feet back from the screen. I am running Windows XP x64, SP2.
Here is a short description of the videos:
Chapter 1: Software
This is a quick look at my test rig and the configuration utility that Ultimarc provides, with a focus on how to change the button assignments of the AimTrak. In reality, I needed to set these, so I thought I would capture it on video.
Chapter 2: Calibration
This video shows the calibration routine, preformed in Windows at the desktop with no other software running. My goal here is to show the simple process of calibrating the AimTrak. You will need a cursor visible (which excludes calibrating with MAME on pause, or in a front-end unless it has been modified) and if you have the AimTrak buttons set to Mouse buttons, this may click your desktop while you are calibrating.
Chapter 3: Accuracy
In this video I take a look at moving around after the gun has been calibrated at the center of the screen, but without re-calibrating in each position. I take a look at how the calibration (and hence, accuracy) changes as the gun changes positions.
I hope this helps. Let me know if there are other things you want to see.
ammitz:
:applaud:
Great videos, specially video nr 3!
Looking forward to the sequels ;)
Xiaou2:
Thanks for the Vids.
Very sorry to see... but not surprised... that the accuracy is poor.
Basically, you can shoot most things in the center fine. But anything on the edges
and you will be off Over an entire CM! Heck, I think I saw over an inch off in the vids.
Sorry.. but thats just unacceptable to me. I actually like to do well in the games I play.
RandyT:
This is an opinion based on what you are showing in your videos. I don't have one of these (yet), so your videos (and Syph's) are what I am commenting upon.
I'm impressed with the great job you did on these! My only criticism is that you tend to use the phrase "a liiiittle bit" where I would probably find "a country mile" to be more appropriate ;). To each his own.
On the bright side, it looks to be very accurate at the center and top center, which is encouraging, and the tracking seems very smooth and fast. Additional firmware tweaking could solve some of these issues. Possibly through a change to the calibration process and the way the data is used. The thing that is difficult is that the gun probably can't tell with any great precision if the shooter is holding the gun in the center and pivoting it around the X and Y-axis, or if the shooter has physically changed the position of the gun in the X and Y (and Z)....or a little of both. It looks to me like the camera sees the markers the same way at 90 degrees to the screen as it does at a different angle, regardless of the fact that the tip is actually pointing at a different location on the screen when at that different angle. This seems to be confirmed by the fact that accuracy is better when pointing at the side of the screen the gun is offset toward. With only one set of markers, it may not be possible to do much better as averaging and checking for distance changes between marker sets, and any other trick associated with multiple markers situated at opposite sides of the screen, isn't possible.
The above video on YouTube from Syph007 shows it in action. It seems fine toward the center of the screen, but at around 1:20 on the video, you can see that he misses nearly every target at the edge of the screen. It's perfectly understandable as to why; he became very good at shooting the targets in the middle of the screen, because his brain was learning and connecting that logical extension of his arm to the target. When he went to the edge of the screen, things changed enough so that what his brain learned in the center of the screen, no longer applied. Even with the cursor enabled, he was apparently unable to compensate. This aspect is the area of greatest concern to me. Hopefully a new version of the firmware will have some new tricks to address these issues by the time I get to play with one.
RandyT
AndyWarne:
--- Quote from: Xiaou2 on October 12, 2009, 04:17:56 am ---
Thanks for the Vids.
Very sorry to see... but not surprised... that the accuracy is poor.
Basically, you can shoot most things in the center fine. But anything on the edges
and you will be off Over an entire CM! Heck, I think I saw over an inch off in the vids.
Sorry.. but thats just unacceptable to me. I actually like to do well in the games I play.
--- End quote ---
The videos show the effects of moving standing position which will affect the accuracy if not calibrated. But I am concerned about the first section of the accuracy video. This gives the impression that the gun is not accurate at the outset, towards the edges. It is. So something is not correct with the calibration. If the gun is not accurate then the calibration would need to be repeated, because it should be.
One issue might be screen overscan. When using an LCD monitor, the edges of the screen are the true edges of the picture so this is not an issue. On a CRT monitor with overscan it might be necessary to aim the gun outside of the screen edge when calibrating. It should be a very quick process to try some calibration cycles and get an impression of the best place to aim.