19 LTX 2.3 LoRAs collected with video samples and download links

A new catalog rounds up 19 community-built LoRAs for LTX 2.3, the open-weight video diffusion model from Lightricks. Each entry includes a description, video examples, and a direct download link, consolidating what had been scattered across individual posts and model cards on HuggingFace and Civitai.

LTX 2.3 is a 22-billion-parameter video synthesis model that runs locally and supports fine-tuning. The LoRA collection spans motion control, style transfer, character consistency, deblur techniques, and aesthetic fine-tunes. Hosted on Stable Diffusion Tutorials, the single-page format includes embedded video previews for each LoRA, letting practitioners compare outputs before downloading weights. The roundup pulls from releases over the past month and appears to be community-maintained.

Lightricks released the base weights in late May 2026, and community fine-tunes began appearing within days. The model's open-weight nature and local-inference capability make it a natural target for unrestricted video generation, a use case that closed APIs like Runway and Pika explicitly block. The LoRA ecosystem around LTX 2.3 mirrors the trajectory of Stable Diffusion XL and FLUX, where community fine-tunes outnumbered official releases within weeks of the base model's debut. For practitioners running video workflows in ComfyUI, the consolidated list saves the manual work of tracking individual announcements and HuggingFace updates. Lightricks released LTX 2.3 under a permissive license allowing commercial use and redistribution—a stance that distinguishes it from earlier video models like Stability AI's SVD, which carried non-commercial restrictions at launch.