LTX Video 2.3 — V2V Style Transfer (ReStyle IC-LoRA)

Model: LTX Video 2.3 — 22B Dev (FP8) + ReStyle IC-LoRA Type: Video-to-Video

Video-to-Video style transfer using a custom-trained IC-LoRA on top of LTX Video 2.3 — 22B. Takes a content video and a style reference image, and re-renders the video in the reference style. The IC-LoRA guide is composited via LTXAddVideoICLoRAGuideAdvanced, with the distilled speed-up LoRA applied for fast 8-step sampling.

Preview

Requirements

ComfyUI: recent stable build
Model: ltx-2.3-22b-dev_transformer_only_fp8_scaled.safetensors — Kijai/LTX2.3_comfy
Speed-up LoRA: ltx-2.3-22b-distilled-1.1_lora-dynamic_fro09_avg_rank_111_bf16.safetensors — Kijai/LTX2.3_comfy (applied at strength 0.4)
ReStyle IC-LoRA: lora_weights_step_08000.safetensors — Cseti/852654_LTX2.3-22B_ReStyle_IC-LoRA
Video VAE: LTX23_video_vae_bf16.safetensors — Kijai/LTX2.3_comfy
Audio VAE: LTX23_audio_vae_bf16.safetensors — Kijai/LTX2.3_comfy
Preview VAE: taeltx2_3.safetensors — Kijai/LTX2.3_comfy
Text encoder: gemma_3_12B_it_fp8_scaled.safetensors — Kijai/LTX2.3_comfy
Text projection: ltx-2.3_text_projection_bf16.safetensors — Kijai/LTX2.3_comfy
Custom nodes:

Notes

Inputs: a content video (VHS_LoadVideo) and a single style reference image (LoadImage). The reference image and the first frame of the content video are resized to the desired resolution with nvidia_rtx_vsr and concatenated into the IC-LoRA guide via LTXAddVideoICLoRAGuideAdvanced.
Sampling: euler sampler, linear_quadratic scheduler, 8 steps, CFG 1.

Changelog

2026-05-05 — Initial upload

Training

This IC-LoRA was trained on RunPod cloud GPUs.

Support

Producing and sharing this kind of open-source work requires renting cloud GPUs, which gets expensive quickly. If you find it useful and would like me to keep contributing, your support is very much appreciated: