Main > Main Forum
A Closer Look At The AimTrak (with video goodness)
AndyWarne:
The calibration assumes that the 4096x4096 screen map for the device is exactly fitting the screen, and that when you point into the corner, you are actually pointing at screen location 0,0 for example.
The video shows horizontal overscan and that is the reason for the error, although its a small overscan.
There is a way around this. I could change the calibration so that the mouse cursor moves to a known position near the corner or edge and stays static, and you aim at the pointer instead of aiming at the screen edge. This would eliminate this issue but I am not sure if its really necessary. The downside of this would be on some screens the mouse pointer might be difficult to see if static. On others with a very large overscan it might be off the screen.
I will think about this, but in fact its really quite easy to allow for this. Afetr a couple of calibrations you get the idea of where to aim when calibrating. It can easily be seen whether you need to re-calibrate by looking along the sight after calibration to check it against the pointer.
Of course if the gun does not have a proper sight then all bets are off and you dont then need it to be calibrated accurately at all since the sight would not be any use anyway in gameplay, you are just shooting blind.
Drakin:
--- Quote from: AndyWarne on October 12, 2009, 01:35:29 pm ---The calibration assumes that the 4096x4096 screen map for the device is exactly fitting the screen, and that when you point into the corner, you are actually pointing at screen location 0,0 for example.
The video shows horizontal overscan and that is the reason for the error, although its a small overscan.
There is a way around this. I could change the calibration so that the mouse cursor moves to a known position near the corner or edge and stays static, and you aim at the pointer instead of aiming at the screen edge. This would eliminate this issue but I am not sure if its really necessary. The downside of this would be on some screens the mouse pointer might be difficult to see if static. On others with a very large overscan it might be off the screen.
I will think about this, but in fact its really quite easy to allow for this. Afetr a couple of calibrations you get the idea of where to aim when calibrating. It can easily be seen whether you need to re-calibrate by looking along the sight after calibration to check it against the pointer.
Of course if the gun does not have a proper sight then all bets are off and you dont then need it to be calibrated accurately at all since the sight would not be any use anyway in gameplay, you are just shooting blind.
--- End quote ---
I opt for it to stay the way it is...or somehow allow for both types of calibration.
Xiaou2:
--- Quote ---Of course if the gun does not have a proper sight then all bets are off and you dont then need it to be calibrated accurately at all since the sight would not be any use anyway in gameplay, you are just shooting blind.
--- End quote ---
Maybe Im not interpreting this correctly... but this sounds Incorrect.
As an artist, I can tell you that I have a very good eye and feeling for
angles. My aim, is thus very good.. without the need for looking down
a sight.
In fact, plenty of good gun shooters (real guns) do not use the sights,
and have incredible accuracy. And no... that is not because they Saw
where the first bullet landed.
Sorry, but the Shooting Blind / Accuracy comment are not applicable
to all people.
--- Quote ---Looks like you won't be buying then. Oh well. Thanks for the input.
--- End quote ---
Ohh, thank you for your concern Frizzel. You are correct at this
point in time, as I do not buy Prototypes. I buy fully functional and fully
tested equipment, that works as I expect it to work.
If I wanted to play with something with poor accuracy, Id use the
CrapLabs guns I have. Instead, Ill stick with the Topguns until
a prototype can or should I say "IF" accuracy can be shown true.
Standing back a little further isnt so bad as having missed shots because
things are not accurate.
Im all for a good working and accurate gun that is able to track at
closer range... Ill keep waiting till I see it... And if it comes, I may buy
a few of them. Until then, Ill make my disappointment known. Im not
the only one interested in an accurate solution.
RandyT:
--- Quote from: AndyWarne on October 12, 2009, 01:35:29 pm ---The video shows horizontal overscan and that is the reason for the error, although its a small overscan.
--- End quote ---
For overscan to be the issue causing the gross deviation at the bottom of the screen, the vertical overscan would need to be very large. Even then, one would expect that a similar deviation would be seen at the top, or center, based on the screen mapping. Something else seems to be awry.
--- Quote ---There is a way around this. I could change the calibration so that the mouse cursor moves to a known position near the corner or edge and stays static, and you aim at the pointer instead of aiming at the screen edge. This would eliminate this issue but I am not sure if its really necessary. The downside of this would be on some screens the mouse pointer might be difficult to see if static. On others with a very large overscan it might be off the screen.
--- End quote ---
Move the cursor in a small box pattern with the center of the box inset from the edge of the screen by 10%. Get the resulting distance between the calibration shots, representing 80% of the screen distance, and add 1/8th of that distance to each side to intelligently guess where the actual screen edge is. This will account for overscan and therefore increase the effectiveness of the calibration step...at least horizontally. Vertically, you have a greater challenge. The error does not appear to be linear, which is likely due to to the curve of the lens in the camera. The further off-axis the camera lens is to the LED markers, the more likely it is that optical distortion will become a factor. Without a second set of markers to use as a reference, the only alternative is somewhat complex and time consuming math (or a factor table) to attempt to correct for it. The effectiveness of this is also somewhat questionable without the gun knowing, with very good accuracy, the physical size of the screen and / or or the distance between screen and shooter.
--- Quote ---Of course if the gun does not have a proper sight then all bets are off and you dont then need it to be calibrated accurately at all since the sight would not be any use anyway in gameplay, you are just shooting blind.
--- End quote ---
I'm not sure what you mean by "proper sights" (a definition of that would probably be helpful), but with real guns, a "proper sight" is sometimes nothing more than a ball bearing centered on the end of the barrel that can be aligned with the center of the barrel width when looking down the length of it. In reality, a "proper sight" is any static protrusion or physical feature on the gun that a shooter can use as a reference. So long as the shots always deviate the same way, the shooter can compensate. It's when the deviation varies based on location of aim that compensation becomes nearly impossible.
RandyT
ammitz:
--- Quote from: Xiaou2 on October 12, 2009, 02:00:18 pm --- As an artist, I can tell you that I have a very good eye and feeling for
angles. My aim, is thus very good.. without the need for looking down
a sight.
--- End quote ---
:laugh2:
Navigation
[0] Message Index
[#] Next page
[*] Previous page
Go to full version