<h3 id="overview">📌 Overview</h3>This is a ComfyUI workflow built to run the LTX-2.3 22B model alongside the IC-LoRA Outpaint module created by oumoumad. It allows you to expand and outpaint both still images and existing video clips.To make this accessible for mid-range hardware (12GB–16GB+ VRAM), the workflow is pre-configured to use the lightweight GGUF versions of the LTX-2.3 models rather than the massive standard FP8/BF16 checkpoints. It handles both Text-to-Video (T2V) and Image-to-Video (I2V) generation within a single setup.<hr /><h3 id="how-to-use-and-configurations">⚙️ How to Use &amp; Configurations</h3>1. Switching Between Text-to-Video (T2V) and Image-to-Video (I2V)<ul><li>For T2V: Set the bypass_i2v node to True. (Note: The Load Image node still requires an image input to run the workflow, but it will be bypassed and will not affect the final video).</li><li>For I2V (Outpainting): Set the bypass_i2v node to False and upload your reference image or video frame into the Load Image node.</li></ul>2. Controlling the Outpaint Area Locate the ImagePad node in the workflow. Use the left, right, top, and bottom values to set your padding amount. This dictates how much extra area the model generates in each direction. If you do not want the model to outpaint a specific side, set that value to 0.3. General Settings<ul><li>Prompts: Use detailed, descriptive positive prompts to guide the outpainted areas.</li><li>Resolution: Adjust your width and height carefully. Higher resolutions directly increase VRAM usage.</li></ul><hr /><h3 id="required-models-and-directory-structure">🧩 Required Models &amp; Directory Structure</h3>You will need the following models placed in their respective ComfyUI directories.<ul><li>Credit to oumoumad for the IC-LoRA Outpaint model, and Kijai/unsloth for the GGUF quantization and model splits.</li></ul><pre><code>📂 ComfyUI/
├── 📂 models/
│ ├── 📂 unet/
│ │ └── ltx-2.3-22b-dev-Q3_K_S.gguf
│ ├── 📂 text_encoders/
│ │ └── gemma_3_12B_it_fp8_scaled.safetensors
│ │ └── ltx-2.3_text_projection_bf16.safetensors
│ ├── 📂 vae/
│ │ └── LTX23_audio_vae_bf16.safetensors
│ │ └── LTX23_video_vae_bf16.safetensors
│ └── 📂 loras/
│ └── ltx-2.3-22b-ic-lora-outpaint.safetensors &lt;-- (By oumoumad)
│ └── ltx-2.3-22b-distilled-lora-384.safetensors
│ └── ltx-2.3-22b-distilled-lora-dynamic_fro09_avg_rank_105_bf16.safetensors</code></pre><h3></h3><hr /><h3 id="links-and-resources">🔗 Links &amp; Resources</h3><ul><li>1-Click Installer: If you want to skip downloading the models and nodes manually, I have created a portable Windows 1-Click Installer that sets up this exact environment. You can find it on my Patreon: <a target="_blank" rel="ugc" href="https://www.patreon.com/posts/155654936">Link to Patreon Post</a></li><li>Tutorial Video: Coming Soon</li></ul>

Files with SHA256: 63083f5920aa150af35c73e3f34586ef9612219bce529469064ed72013cbf128