CommunityNSFWPlatform

Star Trek TNG LoRA extends LTX2.3 video with show-accurate motion

A new LoRA for LTX2.3 recreates Star Trek: The Next Generation's visual style in AI-generated video, trained on episode screengrabs and shared on HuggingFace.

May 15, 2026

Star Trek TNG LoRA extends LTX2.3 video with show-accurate motion

A Star Trek: The Next Generation-style LoRA for LTX2.3 is generating TV-accurate video clips from show screengrabs. The LoRA, released on HuggingFace by bionicman69, trains LTX2.3 to match TNG's lighting, color grading, and camera movement. Users feed the model still frames from the show and get short video clips that extend the scene with consistent motion and atmosphere.

More in Community

Watch: Star Trek TNG LoRA extends LTX2.3 video with show-accurate motion

The approach pairs LTX2.3's native motion capabilities with the LoRA's style transfer. Each clip starts from a single TNG screengrab—bridge shots, corridor scenes, close-ups—and the model generates 2–4 seconds of video that holds the show's visual signature. The LoRA weights are available at HuggingFace and run in standard LTX2.3 workflows.

Workflow and performance

The process is straightforward: grab a frame from any TNG episode, load the LoRA into an LTX2.3 ComfyUI or command-line setup, and generate. The model handles camera drift, character micro-movements, and environmental details like console blinks and background crew motion. Results vary by input quality—high-resolution DVD or Blu-ray frames produce cleaner output than compressed streaming stills—but the style consistency holds tight across different bridge sets, uniforms, and lighting conditions.

LTX2.3's open weights make this kind of fine-tuning accessible. The LoRA itself is small enough to run on consumer GPUs alongside the base model, and the HuggingFace card includes sample prompts and recommended settings. For TNG fans running local video generation, it's a direct path from screengrab to clip without API costs or content filters.