The Ultimate Mouse Thread - Page 97

On January 09 2013 05:49 MisterFred wrote:


Actually Predje is mostly incorrect. It's not really the accuracy of the sensor that's being displayed. What is being displayed is the presence (or lack thereof) of such "features" such as angle-snapping (prediction), acceleration, sensor jitter, that sort of thing.

The Savu doesn't really have any of these "features," which is considered a huge plus for most gamers, so the line tests are pretty uninteresting. And they're actually really hard to do correctly! So, anyway, I'm not a complete expert, but here's what the tests show.

Spiral, vertical, and horizontal lines: "angle-snapping," aka prediction
If you get a relatively regular spiral, then there is no angle-snapping/prediction. If you get a "squarish" spiral (relatively flat in the parts that are mostly horizontal or mostly vertical, with corners marking the transition between the two, then there is angle-snapping/prediction.

Why does it work? Some mice have software (or hardware!) built into them that make it easier to draw perfectly straight lines. The mouse assumes you want to go directly from point a to point b when you move your mouse - but people generally cannot draw completely straight lines because of inherent inaccuracy, shaky hands, etc, so the mouse "smooths out" the information sent to the computer. Gamers generally dislike this because this "feature" can theoretically interfere with accuracy - the mouse might put the pointer a few pixels off where the gamer actually wanted it to go.

So look at the spirals above for the Savu. They're hand-drawn, sure, but there's no consistent flat lines at any particular part of the spiral, nor repeated angles at the places one would expect given angle-snapping/prediction.

The vertical & horizontal line tests work in a similar fashion. People can't generally draw perfectly straight lines. So if mouse "angle-snapping/prediction" is turned on, the lines will have long sections where they are perfectly straight, not straying to either side by even a pixel. That's not the case here, so the undesired "feature" is not present.

Diagonal lines: this can measure angle-snapping/prediction as well, though not as well. Instead, the test is good for exposing DPI settings that are "software-enhanced" rather than "native". Take for instance my Mionix Naos 3200. As much as I love it, the 3200 DPI setting is software-enhanced. What does this mean? Well the 800 & 1600 settings, which are native, can measure each pixel. If a certain distance moved translates to an 1600-pixel shift of the mouse pointed, the mouse can also measure half that distance plus a tiny bit for a 801-pixel shift. The software-enhanced 3200 DPI setting cannot be that precise, because it is just the 3200 DPI setting with the pixel shifts doubled. It can measure a 3200-pixel shift or a 1602-pixel shift, but it CANNOT measure a 1601 pixel shift. The upshot is that there are certain pixels on the screen it's next to impossible to click on. The same as if the mouse was set to 1600 DPI & windows mouse sensitivity was used to double the speed of the mouse.

The diagonal lines test helps show this. Zoom really far into any diagonal line and it looks like a staircase. "Software-enhanced" DPI or high windows settings increase the size of the "stairs". So if you have smooth-looking diagonal lines, there isn't software-enhanced DPI. Of course people's hands still aren't perfectly steady, so you basically want to look at the size of the aberrations. For example:



There are more "bumps" in the 1600 DPI lines only because my hands aren't very steady & it's a faster DPI setting (harder to be accurate/shaky hands problems). But the "bumps" aren't much larger than in the 800 DPI setting. The 3200 DPI bumps, however, are of a more extreme quality and even the smoother parts of the lines are more "blockish" than the smoother parts of the 800 & 1600 DPI lines.

The Savu advertises all its DPI settings as "true," and as having no angle-snapping/prediction or acceleration or anything like that. Unsurprisingly, the line drawings are really boring and tell us only there are no problems.

Does this make the Savu the end-all, be-all of gaming mice? Well, no. It's a very, very good implementation of the Avago 3090 sensor (the newer version of my own Mionix Naos 3200's Avago 3060 sensor). But shape of mouse is a big deal too. Many people don't like it. Many people (who like smaller mice that are also easy to lift off & reposition), seem to love it. Personally, I'd avoid the mouse on shape issues. But maybe it's right for you.

**Hopefully I got the technical details right in this post. It's possible (problable?) I made some mistakes.