🚀 MrXin LTX 2.3 I2V EROS V6 This workflow has continuously evolved based on user feedback to deliver one of the strongest LTX 2.3 I2V experiences available. Thx everyone for the feedbacks. V6 is a major leap forward, focusing on stronger motion preservation, better prompt adherence, and more user control while keeping excellent stability and VRAM efficiency.<h3 id="what's-new-in-v6-(compared-to-v5)">🔥 What’s New in V6 (Compared to V5)</h3><ul><li>Added LTX2 NAG (Negative Attention Guidance) on both First Pass and Final Pass for significantly improved prompt following and motion quality.</li><li>Switched to VAELoader KJ nodes for Video VAE and Audio VAE (better compatibility and stability).</li><li>Improved Final Pass by repositioning the LTXVCropGuides node for cleaner latent processing.</li><li>Added flexible sigma system: Manual Sigmas (8 steps) or automatically generated sigmas via LTXVScheduler + Sigmas Sigmoid (12 steps).</li><li>Final Pass is Manual Sigmas only for stability.</li><li>Audio volume default set to 0 (neutral) with easy slider control.</li><li>Enhanced image conditioning using LTXVImgToVideoInplaceKJ in both passes.</li><li>Cleaner workflow layout, better group organization, and more helpful in-workflow tips.</li></ul>✅ Core Features<ul><li>Extremely strong, natural and consistent motion from First to Final Pass</li><li>Excellent anatomy, lighting, physics and NSFW detail</li><li>High-quality 24 FPS video with perfectly synced audio</li><li>Easy model switching between 10Eros V1 FP8 and Distilled 22B</li><li>Built-in Video Editor (RTX Super Resolution + nmkdSiaxCX upscaler + RIFE 48/60 FPS)</li><li>Optimized for 16GB VRAM setups and 32GB RAM</li></ul>📥 Required Models + Direct Download Links<ul><li>10Eros V1 FP8 (Main Checkpoint) → <a target="_blank" rel="ugc" href="https://civarchive.com/api/download/models/2892069?type=Model&amp;format=SafeTensor&amp;size=full&amp;fp=fp8">ltx2310eros_v1_FP8.safetensors</a></li><li>Distilled Model → <a target="_blank" rel="ugc" href="https://huggingface.co/Kijai/LTX2.3_comfy/resolve/main/diffusion_models/ltx-2.3-22b-distilled_transformer_only_fp8_input_scaled_v3.safetensors?download=true">ltx-2.3-22b-distilled_transformer_only_fp8_input_scaled_v3.safetensors</a></li><li>Text Encoder → <a target="_blank" rel="ugc" href="https://huggingface.co/GitMylo/LTX-2-comfy_gemma_fp8_e4m3fn/resolve/main/gemma_3_12B_it_fp8_e4m3fn.safetensors?download=true">gemma_3_12B_it_fp8_e4m3fn.safetensors</a></li><li>Text Projection → <a target="_blank" rel="ugc" href="https://huggingface.co/Kijai/LTX2.3_comfy/resolve/main/text_encoders/ltx-2.3_text_projection_bf16.safetensors?download=true">ltx-2.3_text_projection_bf16.safetensors</a></li><li>Video VAE → <a target="_blank" rel="ugc" href="https://huggingface.co/Kijai/LTX2.3_comfy/resolve/main/vae/LTX23_video_vae_bf16.safetensors?download=true">LTX23_video_vae_bf16.safetensors</a></li><li>Audio VAE → <a target="_blank" rel="ugc" href="https://huggingface.co/Kijai/LTX2.3_comfy/resolve/main/vae/LTX23_audio_vae_bf16.safetensors?download=true">LTX23_audio_vae_bf16.safetensors</a></li><li>Preview VAE → <a target="_blank" rel="ugc" href="https://github.com/madebyollin/taehv/raw/main/safetensors/taeltx2_3.safetensors">taeltx2_3.safetensors</a></li><li>Spatial Upscaler → <a target="_blank" rel="ugc" href="https://huggingface.co/Lightricks/LTX-2.3/resolve/main/ltx-2.3-spatial-upscaler-x2-1.1.safetensors?download=true">ltx-2.3-spatial-upscaler-x2-1.0.safetensors</a></li><li>Model Upscaler → <a target="_blank" rel="ugc" href="https://civarchive.com/api/download/models/164677?type=Model&amp;format=SafeTensor">nmkdSiaxCX_200k.safetensors</a></li><li>Distilled LoRAs (<a target="_blank" rel="ugc" href="https://huggingface.co/TenStrip/LTX2.3_Distilled_Lora_1.1_Experiments/resolve/main/ltx-2.3-22b-distilled-lora-1.1_fro90_ceil72_condsafe.safetensors?download=true">First </a>&amp; <a target="_blank" rel="ugc" href="https://huggingface.co/Lightricks/LTX-2.3/resolve/main/ltx-2.3-22b-distilled-lora-384-1.1.safetensors?download=true">Second Pass</a>)</li></ul>📁 Folder Structure<pre><code>ComfyUI/
├───📂 models/
│ ├───📂 diffusion_models/
│ │ └─── ltx2310eros_beta.safetensors
│ │ └─── ltx-2.3-22b-distilled_transformer_only_fp8_input_scaled_v3.safetensors
│ │
│ ├───📂 text_encoders/
│ │ └─── gemma_3_12B_it_fp8_e4m3fn.safetensors
│ │ 
│ ├───📂 clip/
│ │ └─── ltx-2.3_text_projection_bf16.safetensors
│ │ 
│ ├───📂 VAE/
│ │ └─── LTX23_video_vae_bf16.safetensors
│ │ └─── LTX23_audio_vae_bf16.safetensors
│ │ └─── taeltx2_3.safetensors
│ │
│ ├───📂 latent_upscale_models/
│ │ └─── ltx-2.3-spatial-upscaler-x2-1.1.safetensors
│ │
│ ├───📂 upscale_models/
│ │ └─── nmkdSiaxCX_200k.safetensors
│ │
│ └───📂 Lora/
│ └─── ltx-2.3-22b-distilled-lora-1.1_fro90_ceil72_condsafe.safetensors
│ └─── ltx-2.3-22b-distilled-lora-384-1.1.safetensors
</code></pre> Custom Nodes Overview – MrXin LTX 2.3 I2V EROS V6<ul><li>ComfyUI-KJNodes</li><li>rgthree-comfy</li><li>ComfyUI-easy-use</li><li>ComfyUI-mxToolkit</li><li>ComfyUI-VideoHelperSuite</li><li>ComfyUI-LTXVideo</li><li>controlaltai-nodes</li><li>comfyui_nvidia_rtx_nodes</li><li>GACLove/ComfyUI-VFI (voor RIFE)</li><li>comfyui_memory_cleanup + comfyui-impact-pack </li></ul>Pro Tip: Use the new 10Eros V1 FP8 model with its recommended trigger words for the best results. ✅ Quick OOM Fix GuideIf you’re getting Out of Memory errors or weird faces in the final pass, follow these steps in order:1. Update or Fresh Install ComfyUI + Install Missing Custom Nodes<ul><li>Download and run the Easy Installer: 👉 <a target="_blank" rel="ugc" href="https://github.com/Tavris1/ComfyUI-Easy-Install">https://github.com/Tavris1/ComfyUI-Easy-Install</a></li><li>Or watch this short video: 👉 <a target="_blank" rel="ugc" href="https://www.youtube.com/watch?v=CgLL5aoEX-s&amp;t=1041s">https://www.youtube.com/watch?v=CgLL5aoEX-s&amp;t=1041s</a></li></ul>After updating/installing:<ul><li>Open ComfyUI → go to Manager → click Install Missing Custom Nodes</li><li>Restart ComfyUI completely.</li></ul>2. Add the Low VRAM Flags<ol><li>Go to your ComfyUI folder.</li><li>Find the file ComfyUI.bat.</li><li>Right-click → Edit (open with Notepad).</li><li>Find the line that has ComfyUI\<a target="_blank" rel="ugc" href="http://main.py">main.py</a> .</li><li>Add the flags at the end so it looks like this:</li></ol>bat<pre><code> ComfyUI\main.py --lowvram --reserve-vram 6 --preview-method none --disable-xformers --disable-smart-memory</code></pre><ol><li>Save and launch ComfyUI using this .bat file.</li></ol>3. Increase Windows Swap File (Virtual Memory) This helps when VRAM still spikes:<ol><li>Right-click This PC → Properties → Advanced system settings.</li><li>Click Settings under Performance → Advanced tab → Change (under Virtual memory).</li><li>Uncheck “Automatically manage paging file size for all drives”.</li><li>Select your main drive (usually C:) → choose Custom size.</li><li>Set:<ul><li>Initial size (MB): 49152</li><li>Maximum size (MB): 98304</li></ul></li><li>Click Set → OK → restart your PC.</li></ol>After doing all three steps, reload the workflow and test with Longer Side 1024 and Video Length 20 seconds first. This combination fixes OOM for almost everyone on 12GB cards. <hr /><h3 id="version-history-and-previous-updates">Version History &amp; Previous Updates</h3>V5 — Major base model upgrade to 10Eros V1 FP8. Significantly better realism, anatomy, lighting and motion coherence. Updated model loading and separate LoRA handling for each pass.V4 — Introduced separate Manual Sigmas for First and Final Pass, much stronger Final Pass motion, full TwoWaySwitch system for model selection, expanded LoRA stack, and refined Video Editor.V3 — Solved the biggest V2 complaint by preserving much more motion in the Final Pass. Added Model Upscaler (nmkdSiaxCX), option to disable Final Pass, distilled model support, and custom resolution.V2 — Added full built-in Video Editor (RTX Super Resolution + RIFE to 48 FPS), improved LoRA suite, better audio sync, and major stability enhancements.V1 — The original production-ready workflow. Dual-pass system, strong audio integration, excellent low VRAM optimization (12GB cards), and live previews.<hr /><h3 id="disclaimer:">Disclaimer:</h3><h3 id="this-workflow-is-provided-for-entertainment-artistic-and-creative-purposes-only.-it-may-not-be-used-for-any-illegal-harmful-non-consensual-or-malicious-activities.-please-use-it-responsibly-and-respect-all-applicable-laws-and-ethical-guidelines.">This workflow is provided for entertainment, artistic, and creative purposes only. It may not be used for any illegal, harmful, non-consensual, or malicious activities. Please use it responsibly and respect all applicable laws and ethical guidelines.</h3>— MrXin (May 2026) 🔥

MrXin LTX 2.3 I2V - Eros - 12GB VRAM & 32GB RAM - Workflow

🚀 MrXin LTX 2.3 I2V EROS V5This is the latest evolution of the popular MrXin LTX 2.3 I2V EROS workflow.V5 introduces a major upgrade to the base model: the new 10Eros V1 FP8 checkpoint. This updated model delivers noticeably better anatomy, lighting, motion coherence, and overall visual quality compared to previous versions.🔥 What’s New in V5 (Compared to V4)<ul><li>New 10Eros V1 FP8 base model — significantly improved realism, detail and prompt adherence (see full details <a target="_blank" rel="ugc" href="https://civitai.red/models/2447875?modelVersionId=2892069">here</a>)</li><li>Updated model loading with DiffusionModelLoaderKJ for better compatibility and performance with the new FP8 checkpoint</li><li>Separate LoRA loading for First Pass and Final Pass (optimized distilled LoRAs for each stage)</li><li>Improved RTX Super Resolution (now default 2x upscale)</li><li>Refined node layout and connections for better stability</li><li>Default video length remains 20 seconds with strong motion preservation</li></ul>

Files with SHA256: 1d0d93ec8e7794561008e6b8faba0b802b6d3127d03579e45cb0a90b84a7387e