I was finally able to spend some time doing a proper test of the "fish-eye" lens. And as long as I was making an adapter for the lens, I figured I might as well make the "Perfect Shot" more perfect for sight accurate use as well. It's quite a bit better now than stock
Near as I can tell, the alignment of the lens is right on. After calibration, things track nice and straight in the center of the screen and show similar curvature distortion on both the top and bottom, so that's a good sign.
So, does it work? Well, it does...but not without a few caveats. The first one is the obvious distortion issue. The lens will completely screw up the calibration and might make it difficult to use the wizards. This is to be expected, as the lens will make the ratios look a little off. Manual calibration isn't too difficult if that's what needs to be done. If enough people use this kind of lens (and the author happens to be one of those people
) we may see some built-in correction at some point.
Toward the central area of the screen, I didn't notice much of an issue, if any. The cursor seemed a bit jumpy, but I'll cover that later in. As far as accuracy goes, there really isn't much drift until you get toward the edges of the screen. It's not huge, but it's likely enough to adversely affect performance for a really demanding title like Point Blank. But VC and HOTD:R seemed to play perfectly fine.
In fact, in the first scene of VC, I was able to hit 95% accuracy on the first test and where I missed I knew it was something I did. And that brings me to a serious benefit of using the lens: Distance. This add-on allowed me to be closer than half of the distance I would otherwise need to be at. If I didn't know that these games were meant to have players this close to the screen, I'd have felt like I was cheating. It makes that much of a difference. My CRT is 37" diagonal, but there is a wide, angled bezel around the tube, so it's more like 43" based on where the LEDs needed to be. Even with a screen that large, 24 to 30" seemed to be the sweet spot. I haven't tested with anything smaller, but what I have seen leads me to believe that just over a foot would be doable on a typically sized 4:3 arcade monitor.
Now for what could be the biggest issue for some, at least those who have larger screens. LED beacon brightness. There's no such thing as 100% transmissive lens, and as the size gets smaller, and the lens system gets cheaper, the more that transmission gets reduced. This means that there is a smaller zone the gun must be in to work, as too far away means not enough energy from the LEDs, but too close, and the pattern gets lost. At first, I thought it was something which might be addressed by angling the LEDs, but even positioning the muzzle directly on-axis with one of them seemed to have no effect on the ability of the system to pick it up, past a certain distance, even with the sensitivity cranked to maximum. I also noticed a bit more jitter from the cursor, likely due to the reduced brightness. I was able to effectively make use of the "bad LED" and smoothing modes to minimize a lot of this, which normally isn't necessary without the lens. I may completely remove the IR filter to see if this has any positive effect, but probably not right away.
At the end, it's a bit of a mixed bag. If absolute accuracy is important to you, you may want to skip the lens, wait for the curvature corrections in the software or maybe try an intermediate power to see if it has less distortion. But if you tend to play the standard fare and want more of a "real light gun" experience where distance is concerned, I would give it try. I didn't find it so bad that I immediately removed it, so that probably says something.
*Edit*
Did a little more testing with a cell cam and the fish-eye. I think the FE may have some even larger issues. With a regular camera, there seems to be quite a bit of internal reflection and loss of focus at the edges, both of which could be causing some major problems. I think I'm going to downshift and try out the less extreme wide-angle lens. Some simple testing has showed that this has the potential to be nearly as as good distance-wise, but with less negatives.