Lighting and Performance

The Hexgears Gemini Dusk supports N-key rollover USB, which tested successfully using Aqua's test. Switch Hitter confirmed no chatter with these keys, and you can see above that the R. Win key is replaced effectively by an Fn key as usual with most mechanical keyboards today.

There's something I should point out here: R. Win is visible in the stock photos on the Hexgears product page, and the entire keycap set is in fact quite different from what you actually get. The Kono Store Hexgears sells the keyboard on reflects the correct set, but has Fn and Menu swapped. It is the third-party MechanicalKeyboards.com store that actually gets it right, which is just poor attention to detail on the first-party front. I can understand why Kono Store got it wrong, maybe the person taking the photo was assembling it and put the keycaps on how they usually go, in the order of R. Alt, R. Win, Menu, R. Ctrl, and Fn replacing R. Win. Either way, I too got confused initially and hit Menu when I meant to hit Fn, which wasn't great for all the times done testing and flashing the keyboard firmware.

When first connected, the keyboard lights up in a rainbow wave effect that is quite impressive to look at combined with the side lighting. The 32 LEDs help, but I wish the diffuser were thicker since you still have discrete LED lights showing up around the keyboard.


Seen above are a few other photos of the keyboard lit up, and as mentioned before, there aren't many lighting effects to choose from. The side lighting is brighter than the backlighting too, and those thick PBT keycaps aren't the most cooperative, either. A closer look shows some of the tiny gaps in the lighting of the legends, especially where you would otherwise have seams, so it does appear Tai-Hao took up where Ducky started and improved things to where you don't have the ugly bars, though it is still not perfect. There is some light bleed, albeit barely noticeable on this Gemini Dusk model. I suspect the Gemini Dawn will be brighter overall courtesy the more reflective white colors. Speaking of which, seen above is what #fffff looks like. It should be white, but has a red hue, which Hexgears acknowledged is a bug. The guide suggests using #B9EEFF instead, which made absolutely no difference.


Here is the one saving grace that can help avoid most of the software issues. Remember those seven layers available on the keyboard? In practice, only layers 3–7 are blanked out and good to go for customization purposes. The base layer is obvious enough, and the provided two replacement keycaps are indeed best used here if you don't plan on using any onboard controls since Scroll Lock and Pause/Break are replaced by the Vol - and Vol + keycaps. Layers 1 and 2 have a lot of pre-programmed functions tied to them, and at least layer 1 is easy enough to access, so you can have media playback and volume controls, NKRO to 6KRO change if needed, and PC power settings shortcuts, as well as other shortcuts to pull up programs and more language remap settings than I have seen before. Missing are some obvious ones, like QWERTY/Dvorak/Colemak or even the PC/macOS layouts, though. Layer 2 is more about the lighting controls but simpler, and only allows you to switch between the presets with minimal customization on the brightness and speed of the effects. You can also turn off the lighting altogether here, which is far more complex a procedure than the average consumer would like.



The key-assignment options in the Hexgears configurator allow users to also go with layouts other than QWERTY, and here, I used the source code from an older keyboard review in a layout tester (courtesy of Patrick Gillespie). The results above indicate that QWERTY is not even close to the most efficient layout for me, and very likely you as well. If you are willing, try out Colemak or Dvorak as they will result in lower finger travel and fatigue overall. As such, the software drivers offer a significant functional tool if you are willing to devote some time to adopting either Colemak or Dvorak. Here is the link to the full results if you are interested, including a detailed analysis on key presses and distances moved.


We saw before that there are three Kailh BOX switch options with the Hexgears Gemini Dusk, and I have the BOX Brown switches here. I will be doing a whole new "Building a Keyboard" article featuring five different Kailh BOX switches soon, including this one, if the design interests you. The actual feedback itself is quite good and pronounced, closer to the Cherry MX Clear than the MX Brown, so that is certainly a good thing. These are also quite smooth because of the decoupled actuation and the lubed plastic active block. They are rated similar to the Cherry MX Brown in terms of actuation at 2 mm (closer to 1.8 mm) but with a total travel of of just 3.6 mm and the tactile bump coming in quite early followed by two linear segments. Actuation force is 45 gf compared to the peak tactile bump needing 60 gf, so despite there being a substantial gap between the two, you definitely end up actuating the switch. Bottoming out peak force is again 60 gf, making this a light-to-medium force switch overall. The BOX Brown in itself is not my favorite of the Kailh BOX switches, mostly because I do find some inconsistency in the actuation and tactile feedback. A random test of 20 switches revealed tactile feedback to be the worst in consistency when it comes to switch travel, and actuation was more or less spot-on at 1.8–1.9 mm. Actuation forces were also a touch more varied as measured by a tension gauge, but it's not precise enough to take very seriously.


As always, the sound of a keyboard is based on more than just the switch type. So when comparing sound clips, consider the keyboard as a whole. In this case, I have provided above an example sound clip of me typing on the Hexgears Gemini Dusk sample at ~110 WPM as it comes out of the box with the Kailh BOX Brown RGB switches. For context, you can find sound clips from other keyboards here, including those with tactile switches. The slimmer nature of the case helps avoid reverberations, and the lubed stabilizers coupled with the thicker PBT keycaps also results in a bassier response than usual. The main source of the noise ends up being the switch bottoming out on the metal plate itself, although the space bar gets in there at times too.