Sorry roa, it’s not released yet, still under heavy testing and development 
2 Likes
Update of June 3rd 2025:
Here’s what the current version in development looks like 
YOLO tracking, automated optical flow tracking, ROI fixed by the user, inside FunGen post editing etc.
If you feel like supporting the development:
7 Likes
Love it! Seriously, the level to which you guys have brought this project is amazing.
Is that second pic a hint at non-VR 2D support??
Haha, thank you, lost a few hours of sleep and a bunch of neurons in the process.
There is still a lot to do, and FunGen features makes it look more like a scripting assistant now than a pure AI autonomous funscript generator.
And yes, somehow, I added a feature for 2D where you define a ROI and the assistant does its thing. Still very raw.
Basically, you now have 3 built-in modes.
- Full autonomous scripting, mainly for VR and 2D POV. Mostly relying on YOLO detections and distance estimation, like the previous version. Generates Chapters (HJ/BJ, Missionary, etc., basic)
- Live autonomous combination of YOLO to build a ROI and Optical Flow to determine movement amplitude. Mainly for VR and 2D POV. Give pretty good results, manages occlusion nicely.
- User defined ROI and Optical Flow, for VR, 2D POV and 2D non POV.
Optical Flow delay can be calibrated to precisely match end points during generation.
Those modes allow you to either create a funscript from scratch or inject into an existing funscript.
Then you have the editing part, where you can move up, down, left, right, delete, clean, simplify, clamp points, or even insert new ones. Navigation frame by frame.
You can script, alter, delete only a range of frames.
Add a Chapter, Edit, Delete, Merge, Merge over a gap and track the gap.
Undo, redo actions.
Display or hide the funscript overview, heatmap.
Dispay or hide 1 or 2 editable timelines.
Zoom in/out the video, control the frame rate during playback and live scripting.
And there is a raw attempt at adding another axis (Left/Right) on top of Up/Down, but I did not spend enough time on this to make it relevant yet.
5 Likes
Now ready for alpha pre-release, looking for 5 to 10 testers with Python experience, no installation support.
First wave of bugs was identified by the crash-dummy Walton in discord, fixed one after the other.
1 Like
Hi k00gar, i’m interested
Reach out to me in the Discord please, will share the dedicated chat room and files
Thanks again for all the hard work everyone! This program once released will be a game changer 
Pm sent on discord
Great progress with the testing squad, new features incorporated, and we now have detection for Riding vs Thrusting to live inverse the pattern
6 Likes
I have a draft of new tag definitions:
- vr - 180+-degrees vr video
- vr-stereo - vr video that is not 180deg but rather a rectangle.
Must have ai-stereo if it was originally 2d - vr-2d - conversion from vr to non-vr
Questions:
Hi k00gar, I want to join.
reach out to me in discord please
Bot Upgrade Complete: Say Hello to a Smarter, Safer Server!
Get ready for a whole new experience! Your favorite AI Bot just got a massive brain upgrade and we’ve hired a new bouncer to keep things tidy.
The Genius Upgrade: A Real Conversation*
Interacting with the bot is now smoother and more natural than ever.
- Typo-Proof: Messy fingers? No problem! The bot is much better at understanding queries, even with spelling mistakes.,
- Just Chat: No more rigid commands needed for most searches. Just ask what you want, like you’re talking to a person!,
New Spam Protection
To keep our channels clean, the bot now automatically detects and bans users who spam, ensuring a better experience for everyone.
A Quick Note on Usage :
This is an experimental and locally hosted feature I’m excited to provide. I’ll be keeping an eye on its usefulness and performance. Please use it thoughtfully so we can keep this great tool running for everyone!
Enjoy the new and improved bot!
Opening registration for BETA Testing - THE BIG ONE IS ALMOST HERE!
This is turning the app into a full-blown production studio. We’re talking a complete GUI editor, a much smarter AI, and more power than you’ll know what to do with.
Release Highlights
Your All-In-One Studio!
That’s right, no more opening other tools! This is now a full-featured editor. Tweak every point, apply advanced filters, and sculpt your script to perfection right inside the app using the interactive timelines.
Drag, Drop, Done.
Starting a project is dangerously easy. Just drag and drop a video, a whole folder of videos for a batch job, or even an existing Funscript right into the window to get started.
AI Tracking That Handles Occlusions!
Our multi-stage analysis pipeline has gotten a serious upgrade.
It begins with high-speed object detection to find all the… points of interest. The real magic is in the tracking logic, which can now follow objects even when they’re briefly occluded (hidden), intelligently interpolating their path. The AI understands the scene’s context, automatically creating chapters based on the detected position. You have multiple tracking options, from a full 3-stage analysis with optical flow to a faster 2-stage segmentation-only run.
Intelligent Post-Processing!
Why use one-size-fits-all settings? You can now define unique post-processing parameters (smoothing, simplification, amplification) for each chapter type. The app applies them automatically for a perfectly tailored script every time.
Up/Down Not Enough? Now There’s Left/Right!
You asked for it! We’ve added a second, experimental axis for Left/Right motion to capture all the action.
Thank you to the incredible alpha stage testing team!!!
Apply, dive in and have fun! By the way, did you even check the latest crazy LLM update to the bot?
Support the development : https://ko-fi.com/k00gar Still a lot in the making ! (edited)
11 Likes
Just found this thread a few hours ago and would like to test out your AI funscript generation tool. I’m between jobs right now and have some extra time on my hands and wanted to improve my skills with AI with regards to training and incorporating it with OpenCV. I’ve got experience as a software developer (7+ years python + 2 years AI (torch)).
Tried to join the discord to download the models as indicated in the github repo README file, but I’m unable to join. I’m receiving a “Unable to accept invite” error from discord.
Are the models available for download outside of discord by chance? I’m using Nvidia GPU 4090 in windows and also linux.
Just also wanted to say this project is amazing and thank you spending so much of your free time on it.
Edit: I just read the installer.bat script and located where to get the models on GoFile. I’ll keep tinkering…
TA
Hi and thank you!
Get the v0.5.0 branch on GitHub, and the rest happens in the beta test channel of the discord. Not sure what the invite would fail, it is supposed to be permanently valid.
1 Like
I am not sure what is wrong, but I am not able to join Discord. The invite just will not work. 
I looked at branch v0.5.0 and it needs to have the model files that are only found on Discord. Is there another place I can download them, such as hugging face? I have two Nvidia 4090 (one wnidows machine and one linux machine).
TA
Try PM-ing me your discord username ?
PM sent.
1 Like