I am still working on it but maybe not in the ways you'd expect.
In terms of emulation Impact is fully supported, MPU4 is ready for integration once Epoch is finished.
Epoch has has the H8 core almost completely rewritten from the iteration that got released, increasing accuracy and speed. How the original actually ran at all I am surprised, it was so broken. There is a couple of issues left to resolve but I've not really been concentrating on the emulation as much as the visual aspects.
There are various visual aspects I have been working on. The lamping now generates a brightness (calculated in lumens) and a colour (temperature colour in kelvin), both physically accurate making use of no less than 5 quadratic equation solutions to make the lamping look like the real thing. Stand it next to Casino Crazy and run it at 120fps and they look the same. Casino Crazy is special as it seems to have an error in the lamping which makes it much more flickery than other games, this is present on the real machine and emulation.
I make use of the graphics card and feed the values in as pixel values in a 16 * 16 textures for calculation to take the load off the CPU, for the GPU a 16*16 texture or even 2 of them for later techs is nothing.
I have also been working on integrating a ray-tracer, which was a much bigger idea than I thought it would be, I started this before nVidia's RTX was a thing, however it taught me the light physics required to get an accurate filament simulation.
The idea is to create a scene in 3D with the machine in it, then render the 2D view (like a regular MFME game) using raytracing and cache the lighting values, this allows for an accurate 2D view to be played with light reflecting of the cabinet, walls, mirror glass, etc.
Using several acceleration tecniques I have managed to get the render time for a 2560*1600 down to around a minute using an 8700k (6 core 12 threads) @5Ghz, which with full global illumination is not that bad, I am working on a GPU driven version as it has to be rendered for each lamp, so for an Epoch clubber with 512 lamps that takes quite a while. The files generated can be loaded and played in seconds but for editing that's still a bit long and 4k takes even longer. Looks gorgeous though.
I also disovered a local company that does fine art scanning at 10,000DPI and can scan through glass in sizes up to A0. I am preparing to get the IMPACT machines scanned but its £150 setup fee plus £50 per scan at 1,000DPI (10,000 costs more), so I need to strip all the glasses first to avoid doing multiple trips with extra setup fees. I have point cloud data for the Vogue cabinet that I had 3D scanned too. When I get round to finishing it, it will be the most visually accurate emulator it can be.
However I got a bit distracted. The priciples used in a raytracer can also be used for computational fluid dynamics, so I started putting together some basic demo's and went off on a bit of a tangent. Grab a model of an F1 car, stick it in the simulation and you can trace the airflow over the car and produce some fun images.
And then, I've been looking at using the same priciples again to look at model data for Tornado's, there is real life data available but at the moment only some of them create a Tornado when placed in the current simulators when they obviously all should do because it happened in real life. This is usually done on a supercomputer so I'm a bit behind the curve, but there are so few people looking at the problem I'm hoping I might have a breakthrough, there are some things I learned doing CFD that cross over and I have a couple of approaches that haven't been tried yet. I could be entirely wrong and barking up the wrong tree, but if I'm right it may bring us 1 step closer to identifying which storms will/wont create a tornado. If you're wondering why, I just loved the movie Twister, its a crap film, but the subject matter is fascinating and I was surprised to learn that even in 2020 utilising super computers we still aren't able to give accurate warnings. And the false warnings breed complacency.
So that's what I've been up to for fun. Professionally I am now an embedded Android developer, so pretty busy. But I still open up the emulator and do something when I get fed up of working with linux/java.