LTX Director CS2 — One ComfyUI Node to Rule All LTX 2.3 Workflows

In this video, I'll show you the ultimate LTX Director CS2 workflow for ComfyUI — built on the CGlide fork (WhatDreamsCost) custom node. This is the single most versatile LTX 2.3 workflow you can run locally: Text-to-Video, Image-to-Video, Reference-to-Video, First Frame Last Frame, Video Editing, Retake Mode, Prompt Relay shot sequences, and more — all inside one unified timeline-based ComfyUI node. If you do anything with LTX 2.3, this is the only workflow you need.

Featuring WhatDreamsCost-CSGlide (CGlide Fork)github.com

Timestamps

00:00Introduction to LTX Director CS and the CG Glide fork 01:04How to install the custom node and workflow in ComfyUI 02:32Deep dive into the workflow: Models, LoRAs, and technical settings 04:38Exploring the LTX CS panel: Inputs, presets, and timeline controls 08:11Managing modes: Ghost Mode vs. Lyken MSR for reference-to-video 09:40Retake Mode: Setting up frames for selective video editing 10:28Advanced settings: Saving timelines, Prompt Relay, and Ollama integration 13:21Use Case 1: Text-to-Video generation using motion LoRAs 14:13Use Case 2: Image-to-Video and utilizing Prompt Relay 15:40Creating multi-shot videos with segment prompts 17:21-- Using Image with prompt relay 19:38Setting up exact frame control with multiple images and prompts 23:12Using Start and End frames to control shot extensions 24:20-- Extending videos using relay 26:18Reference-to-Video: Using character sheets and backgrounds 30:41Ghost Mask Mode 32:39Video-to-Video Editing: Changing specific details (e.g., car color) 34:31Using Retake Mode for selective frame modification

AI Jigyasa Workflow-

Workflow downloadbuymeacoffee.com

Model links

Models (dev)

BF16huggingface.co FP8huggingface.co GGUFhuggingface.co

Models (distilled)

BF16huggingface.co FP8huggingface.co GGUFhuggingface.co

models/text_encoders

gemma_3_12B_it.safetensorshuggingface.co ltx-2.3_text_projection_bf16.safetensorshuggingface.co

models/vae

LTX23_video_vae_bf16.safetensorshuggingface.co LTX23_audio_vae_bf16.safetensorshuggingface.co

models/loras

• 🔹 Distilled LoRA (Lightricks) — required for distilled pipeline

ltx-2.3-22b-distilled-lora-384-1.1.safetensorshuggingface.co

• 🔹 VBVR — Video-Based Video Reasoning (LiconStudio) — motion control I2V

More infohuggingface.co Officialhuggingface.co R32 390Khuggingface.co

• 🔹 OmniNFT RL LoRA (Kijai/zghhui) — lip sync & audio-video alignment

More infohuggingface.co LTX-2.3-OmniNFT-RL-Lora_bf16.safetensorshuggingface.co

• 🔹 Edit Anything (Alissonerdx) — video editing / modification

More infohuggingface.co

• (Standard + Module .safetensors — see repo for latest filenames)

• 🔹 IC-LoRA Ingredients (Lightricks) — reference sheet consistency

ltx-2.3-22b-ic-lora-ingredients-0.9.safetensorshuggingface.co

models/latent_upscale_models

ltx-2.3-spatial-upscaler-x2-1.1.safetensorshuggingface.co

Vocals Extraction

MelBandRoformer_fp32.safetensorshuggingface.co

• 👍 Support the Channel If you found this helpful:

• 👍 Like the video
• 🔔 Subscribe for more AI workflows
• 💬 Comment what you want to see next

#LTX23#ComfyUI#AIVideo#LTXDirector#CGlide#WhatDreamsCost#GenerativeAI#AIFilmmaking