Learn how to generate funscripts from any video using AI-powered analysis. Step-by-step guide covering setup, processing, and device sync with HaptiQ.
Creating a funscript manually is one of the most tedious tasks in the haptic content ecosystem. A single minute of video can take 30 to 60 minutes of manual scripting. With AI-powered tools like HaptiQ, you can generate a synchronized funscript from any VR or 2D video in minutes instead of hours. This guide walks you through the entire process from start to finish.
A funscript is a JSON-based file format that maps haptic device movements to timestamps in a video. Each entry in the file specifies a position (0-100) at a given millisecond, telling devices like the Handy, Kiiroo Keon, or Lovense Max exactly how to move at each moment.
The file format was originally created by the open-source community and has become the universal standard for haptic synchronization. Nearly every major interactive device supports funscript playback, either natively or through middleware like Intiface Central.
Before AI-based generation, scripting a video required painstakingly placing keyframes by hand using tools like OpenFunscripter. The result was high quality but the process was unsustainably slow. A five-minute video could take three or more hours to script well.
AI-powered funscript generation solves the fundamental bottleneck: time. Instead of manually placing every keyframe, AI analyzes the video frame by frame to detect motion patterns and automatically generates synchronized haptic data. Here is what changes:
HaptiQ uses a multi-stage machine learning pipeline specifically designed for haptic script generation. Unlike simple waveform matching or audio-based approaches, HaptiQ performs actual visual analysis of the video content.
The pipeline begins by extracting frames from your video at configurable intervals. For most content, HaptiQ analyzes 5-10 frames per second, balancing accuracy with processing speed. Higher frame rates yield smoother scripts but take longer to process.
Using OpenCV optical flow algorithms, HaptiQ tracks pixel-level motion between consecutive frames. This stage identifies the dominant movement vectors in the scene, distinguishing between camera motion, subject motion, and the specific movements that should drive haptic output.
Not all motion in a video is relevant to haptic output. HaptiQ's guided tracking system lets you define a region of interest (ROI) so the AI focuses on the specific area that matters. For VR content, the system can automatically detect the optimal tracking region.
The detected motion vectors are translated into position values (0-100) with millisecond timestamps. The algorithm applies smoothing to prevent jarring movements and normalizes amplitude to fill the full range of device motion. The result is a raw funscript that captures the movement pattern of the video.
The raw script passes through HaptiQ's 14 transformation plugins for refinement. Smoothing removes noise, the anti-jerk filter eliminates unrealistic acceleration, and the RDP simplification algorithm reduces file size while preserving the script's character.
Visit haptiq.app/register and create your free account. The free tier includes 3 projects, 5 video uploads, and 10 batch jobs per month, which is enough to test the pipeline and decide if it fits your workflow.
Create a new project and upload your video. HaptiQ supports MP4, MOV, AVI, MKV, and WebM formats. Both VR (side-by-side, over-under) and flat 2D content are supported. Video files are encrypted at rest and never shared.
Click "Generate Script" to start the 5-stage pipeline. You will see real-time progress via WebSocket updates as each stage completes. A typical 10-minute video processes in 2-3 minutes.
Once generation is complete, the funscript loads in HaptiQ's interactive timeline editor. You can zoom, pan, select keyframes, adjust positions, and use rubber-band selection for bulk edits. The editor supports 20+ keyboard shortcuts for power users.
Fine-tune your script with HaptiQ's plugin chain. Common refinements include the Savitzky-Golay smoothing filter for natural-feeling motion, the amplitude plugin to increase or decrease intensity, and the speed limiter to cap movement rate for specific devices.
Export your finished script as a standard .funscript file, or connect your device directly in the browser via Web Bluetooth or Buttplug.io for instant playback sync. HaptiQ supports 870+ devices including The Handy, Kiiroo Keon, Lovense Max, and many more.
AI generation is not a complete replacement for manual scripting in every case. Here is an honest breakdown:
| Factor | AI (HaptiQ) | Manual |
|---|---|---|
| Speed | 2-5 min per video | 30-60 min per minute of video |
| Accuracy | 85-95% with refinement | 99%+ (depends on scripter skill) |
| Consistency | Highly consistent | Varies with fatigue |
| Learning curve | Minimal | Steep |
| Best for | Volume, prototyping, quick sync | Artisan quality, specific feel |
The optimal workflow for most users is to generate with AI, then manually refine the sections that matter most. This hybrid approach gives you 90% of the quality in 10% of the time.
HaptiQ processes all major video formats. The pipeline handles codec detection and frame extraction automatically.
Once your funscript is ready, HaptiQ offers multiple ways to sync it with your device:
HaptiQ's free tier lets you process up to 5 videos. No credit card required.
Yes. HaptiQ's AI funscript generator produces a complete, playable funscript from any video with zero manual keyframing required. Upload your video, click Generate Script, and the 5-stage machine learning pipeline handles frame extraction, OpenCV optical flow motion detection, region-of-interest analysis, funscript generation, and post-processing automatically. For a 10-minute video, the full pipeline takes 2–3 minutes end to end. The output is a standard .funscript file ready to play on any of the 872 supported haptic devices. Accuracy is 85–95 percent on typical content — good enough that most users never touch the editor. If you want artisan perfection, the timeline editor is there with 20+ keyboard shortcuts, rubber-band selection, and 14 transformation plugins for refinement.
Processing time scales roughly linearly with video length and is typically 20–30 percent of real-time. A 5-minute video processes in about 90 seconds. A 10-minute video finishes in 2–3 minutes. A 30-minute video completes in 6–9 minutes. A full-length 60-minute video takes 12–18 minutes. Variables that affect speed include frame extraction rate (HaptiQ defaults to 5–10 fps for the optimal speed/accuracy balance), video resolution (4K takes longer than 1080p), and whether you have a region of interest defined (ROI-scoped analysis is faster than full-frame). You can watch live progress through each of the 5 pipeline stages via real-time WebSocket updates in your browser — no polling, no refresh.
HaptiQ-generated funscripts are standard format and work with all 872 devices in the IoST device database. This includes every major linear stroker (The Handy, Kiiroo Keon, Lovense Solace Pro, Lovense Gravity, OSR2+, SR6), every Lovense vibrator and plug (Lush 3/4, Hush 2, Nora, Domi 2, Max 2/3, Edge 2), Kiiroo Pearl 2, Kiiroo Onyx+, Satisfyer Connect devices, WeVibe Chorus and Sync, Svakom interactive toys, and hundreds more. Connection options include Web Bluetooth (Chrome/Edge direct, no software), Buttplug.io with Intiface Central (all browsers), The Handy WiFi (HAMP/HDSP/HSSP modes), and JoyHub cloud relay. You can also export the raw .funscript file and use it with third-party players like DeoVR, Heresphere, XBVR, and ScriptPlayer.
Yes. HaptiQ supports both VR formats — side-by-side (SBS) and over-under (OU) — along with 180-degree and 360-degree stereoscopic content. For VR, the 5-stage pipeline automatically detects the format and analyzes the appropriate frame region, so you do not need to pre-crop or convert the video. You can also use the guided tracking feature to draw a region of interest on the video, which focuses the AI on the specific action area and dramatically improves accuracy on VR content where most of the frame is background. 2D POV videos, flat 2D content, and tripod-mounted static scenes all work as well. Supported containers are MP4, MOV, MKV, AVI, and WebM with H.264, H.265/HEVC, VP8, VP9, and ProRes codecs.
Yes, the free tier is generous and requires no credit card. You get 3 projects, 5 video uploads, 10 batch jobs per month, and 3 basic client-side plugins (invert, clamp, amplify) — enough to fully test the AI pipeline, generate complete funscripts, and decide if the tool fits your workflow. HaptiQ Pro at $9.99/month or $79.99/year ($6.67/month effective) unlocks unlimited projects, unlimited video uploads, unlimited batch processing, all 14 transformation plugins (including RDP simplification, Savitzky-Golay smoothing, anti-jerk, speed-limiter), multi-axis T-code export for 8-axis devices, guided ROI tracking, priority processing, and priority support. There is also a $4.99/week tier for short-term needs. No tool is pay-to-evaluate — every feature works on the free tier at reduced quota.
HaptiQ's AI achieves 85–95 percent accuracy on typical content, measured against professional hand-scripted references. Manual scripting by an experienced creator reaches 99 percent or higher but requires 30–60 minutes of labor per minute of video. The AI does particularly well on stable camera shots with clear motion, and slightly less well on heavily edited content with rapid cuts or abstract motion. In practice, the 5–15 percent gap matters less than raw accuracy numbers suggest because the quality analyzer scores each script 0–100 and flags specific weak sections for optional manual polish. The hybrid workflow — AI baseline plus 5 minutes of targeted editing on the roughest scenes — consistently produces scripts indistinguishable from hand-crafted ones in blind listener tests, at roughly 10 percent of the time cost.
Create a free HaptiQ account to generate a funscript from your own video.