Entangled: How Open-Source AI Made a Cinematic Short Film in 2026
The 3-minute sci-fi short Entangled is a landmark achievement in open-source AI filmmaking, using only local models for character consistency, voiceover, and music. No cloud services or proprietary tools were used in its creation.
Entangled: How Open-Source AI Made a Cinematic Short Film in 2026
summarize3-Point Summary
- 1The 3-minute sci-fi short Entangled is a landmark achievement in open-source AI filmmaking, using only local models for character consistency, voiceover, and music. No cloud services or proprietary tools were used in its creation.
- 2Just raw creativity powered by ComfyUI, Qwen LLM, and LTX 2.3.
- 3How Qwen LLM Crafted the Script with Human Oversight The film’s narrative, centered on spacetime singularity mechanics, began as a 7-minute script refined using the Qwen 3.5 35B local LLM.
psychology_altWhy It Matters
- check_circleThis update has direct impact on the Yapay Zeka Araçları ve Ürünler topic cluster.
- check_circleThis topic remains relevant for short-term AI monitoring.
- check_circleEstimated reading time is 3 minutes for a quick decision-ready brief.
Entangled: A Manifesto for Open-Source AI Filmmaking in 2026
The 3-minute sci-fi short Entangled has redefined what’s possible in independent AI filmmaking—proving that cinematic-quality video can be created entirely on a consumer-grade RTX 4090 using 100% open-source, local models. No cloud APIs. No Lora training. Just raw creativity powered by ComfyUI, Qwen LLM, and LTX 2.3.
How Qwen LLM Crafted the Script with Human Oversight
The film’s narrative, centered on spacetime singularity mechanics, began as a 7-minute script refined using the Qwen 3.5 35B local LLM. While AI generated the initial draft, emotional depth came from days of human editing—highlighting that AI assists, but doesn’t replace, artistic judgment.
Character Consistency Achieved with the JSON Flux Hack
To maintain visual continuity across 40+ shots, creator Psi-Clone developed a custom JSON Flux Hack. This system locked character parameters using hex color codes, pose descriptors, and cinematic camera angles, ensuring Elena and both versions of Leo remained identical—without Lora fine-tuning.
Voice Synthesis Using Qwen Voice Designer and VibeVoice
Base voices were generated with Qwen Voice Designer, then layered with emotional nuance using the 7B VibeVoice model. This two-stage approach captured subtle inflections—like hesitation or urgency—that generic TTS systems miss.
LTX 2.3 Delivered Unmatched Lip-Sync Precision
LTX 2.3 processed the refined audio to generate photorealistic lip movements. The creator described it as "like directing an actor"—with timing so accurate, viewers couldn’t tell the voices weren’t live-recorded.
Music, Rendering, and Efficiency with ACE and Flux.2 Dev
Original music was composed in under three generations using ACE, then manually edited for rhythm. Renders hit just 5 minutes per 5-second clip at 1080p, thanks to optimized ComfyUI workflows and FP8 precision in Flux.2 Dev.
Why Local AI Matters in a World of Cyber Threats
As the UK’s National Cyber Security Centre warns of state-sponsored spyware like Moonshine and BadBazaar exploiting cloud supply chains, Entangled offers a powerful alternative: ethical, transparent, and community-owned AI creation. Every frame, voice, and note was generated locally—no third-party servers, no data leaks.
The full pipeline—including custom JSON templates and ComfyUI nodes—is now open-sourced on ComfyUI’s GitHub and available on Arca Gidan. For creators seeking independence from corporate AI, Entangled isn’t just a film—it’s a blueprint.
How to Build Your Own AI Short Film in 2026
You don’t need a studio budget. You don’t need proprietary tools. Just:
- A powerful GPU (RTX 4090 or equivalent)
- ComfyUI with Flux.2 Dev (FP8) and LTX 2.3
- Qwen LLM for scripting and Qwen Voice Designer for audio
- A passion for precision and patience for iteration
Download the Entangled pipeline today—and start your own open-source cinematic journey.
