This project is trying to fully reproduce described movements in funscripts by Hismith (not only speed) with video synchronization in VLC video player.
After all steps required for it to work you will need only Start application and it will automatically detect what video do you play in VLC and execute related to it funscript with related scene synchronization.
To be able to do this it uses computer vision with using your Web Camera which tracks Hismith rotation (angle position at each moment) and tries by manipulation with speed changes to achieve all required movements.
For this moment it has very good results when speed is not too high, especially in slow movements (it can track very good all changes). In case of high speed 3-4+ strokes per second now it also has not so bad results.
For minimize complexity of used computer vision algorithms, also as for increase accuracy and speed for making decision, it use color markers which you will need to place on Hismith (more details below)
What is required for its work
You will need:
Hismith with remote control support + Web Camera + backlight (Ring Light as very good solution) + PC + Add color markers to Hismith (which will be tracked)
I see a man of culture as well, cheers for the contribution! This community is like 99% made for the Handy, which kinda sucks tbh… so seeing more people willing to create stuff for fuck machines makes me happy!
Though… i can’t help but notice that it’s all vastly more complex for kinda diminishing returns. A purely rotary drive machine like the Hismith and others in the price range will never be able to control depth to the speed and precision of servo drives or linear motors. While I appreciate the hard work and effort you went through to create this, I don’t really know how many people will have both the hardware and patience required to make this work properly.
Especially when running at higher speeds, say 3-4 hertz rotations on the machine, you will most likely need at least 60fps of image processing to get smooth position tracking results. And there’s quite a bit of extra stuff you need to procure to get it to work well as you described.
I would offer a counter-argument in the form of my own extension for OFS (which i will be also turning into a python app at some point).
It effectively does all that processing beforehand by analyzing the whole script. And it lets you modify it quickly afterwards as it is an OFS extension. Which makes it a lot safer because you will know what to expect when you eventually run it on the machine. And it requires zero additional hardware, and works with all other rotary machines. I made a nice GUI for it that allows you to make separate profiles for your machines, all it needs is a few initial measurements of the device in operation to get a pretty good linear approximation of its power-to-timing curve.
I don’t mean this in a competitive way or anything like that, i just feel the need to give people more choices in this space. Ultimately the goal is to automate as much of this work as possible so that we can spend more time fucking our brains out in sync with videos
1.10 version was released. It is more accurate now. Also it can automatically generate required current device statistic data for automatic control of device.
In most cases now it has very close to zero deviation from what it should be:
