Uploaded January 2026 | Updated November 2026, 3 days ago
LTX-2 ComfyUI is impossible to run without massive enterprise hardware. That is not true. I found a way to get full Cinematic AI Video Production working on a standard GPU using the new GGUF format.
The official model crashed my computer instantly. So, I swapped it for the Q4 GGUF version. It is half the size but keeps the same Commercial-Grade quality. The best part? It renders smoothly without freezing your system.
Important: Update your ComfyUI nodes before loading this, or it will fail.
I also added a specific Tiled VAE step to the workflow. This stops the memory from spiking at the end of the render. If you want to create Professional 1-Minute Videos without upgrading your PC, this pipeline is the answer.
The Workflow All About:
Creating 1 minutes video with LTX2
No Crashes: Using Tiled VAE to manage VRAM.
File You Need To Download
Essential Models:
Kijai's Q4 GGUF Model: huggingface.co/Kijai/LTXV2_comfy/tree/main/diffusion_models
VAE Encoder: huggingface.co/Kijai/LTXV2_comfy/tree/main/VAE
Text Encoders: huggingface.co/Kijai/LTXV2_comfy/tree/main/text_encoders
Upscaling Nodes: huggingface.co/Lightricks/LTX-2/tree/main
gemma_3_12B_it.safetensors: huggingface.co/Comfy-Org/ltx-2/blob/main/split_files/text_encoders/gemma_3_12B_it.safetensors
gemma_3_12B_it_fp8_e4m3fn.safetensors: huggingface.co/GitMylo/LTX-2-comfy_gemma_fp8_e4m3fn/blob/main/gemma_3_12B_it_fp8_e4m3fn.safetensors
Camera LoRA: github.com/Lightricks/LTX-2?tab=readme-ov-file
Get The Workflow JSON: aistudynow.com/how-to-run-ltx-2-on-low-vram-the-complete-gguf-comfyui-workflow
LTX-2 ComfyUI is impossible to run without massive enterprise hardware. That is not true. I found a way to get full Cinematic AI Video Production working on a standard GPU using the new GGUF format.
The official model crashed my computer instantly. So, I swapped it for the Q4 GGUF version. It is half the size but keeps the same Commercial-Grade quality. The best part? It renders smoothly without freezing your system.
Important: Update your ComfyUI nodes before loading this, or it will fail.
I also added a specific Tiled VAE step to the workflow. This stops the memory from spiking at the end of the render. If you want to create Professional 1-Minute Videos without upgrading your PC, this pipeline is the answer.
The Workflow All About:
Creating 1 minutes video with LTX2
No Crashes: Using Tiled VAE to manage VRAM.
File You Need To Download
Essential Models:
Kijai's Q4 GGUF Model: huggingface.co/Kijai/LTXV2_comfy/tree/main/diffusion_models
VAE Encoder: huggingface.co/Kijai/LTXV2_comfy/tree/main/VAE
Text Encoders: huggingface.co/Kijai/LTXV2_comfy/tree/main/text_encoders
Upscaling Nodes: huggingface.co/Lightricks/LTX-2/tree/main
gemma_3_12B_it.safetensors: huggingface.co/Comfy-Org/ltx-2/blob/main/split_files/text_encoders/gemma_3_12B_it.safetensors
gemma_3_12B_it_fp8_e4m3fn.safetensors: huggingface.co/GitMylo/LTX-2-comfy_gemma_fp8_e4m3fn/blob/main/gemma_3_12B_it_fp8_e4m3fn.safetensors
Camera LoRA: github.com/Lightricks/LTX-2?tab=readme-ov-file
Get The Workflow JSON: aistudynow.com/how-to-run-ltx-2-on-low-vram-the-complete-gguf-comfyui-workflow
![USO (New IP-Adapter Flux) ComfyUI Workflow: Image Style Transfer, Same Face, Full Demos
Quick demo of the USO Lora (New IP-Adapter Flux) by bytedance research ComfyUI Workflow with FLUX where I copy style from a reference image and keep the same face across looks]
In this video, youll learn:
• how I set up the files and the LoRA so style transfer from image just works
• how I lock the subject face while swapping styles in ComfyUI
• what guidance settings and sizes gave me fast, clean results
Here’s what actually happened. I wired the USO ComfyUI Workflow, dropped in a style image, and the model picked up the look right away. Same prompt, new vibe. Then I added a subject image so the face stays the same when I move from manga lines to a wool texture or a Ghibli look. Let’s be real, I didn’t expect this part to be so quick. With FLUX and the USO LoRA, it worked better than I thought. If you’re on smaller VRAM, keep the sizes modest and it still feels snappy. I’ll show the exact nodes I used, the guidance band that behaved for me, and a couple of easy fixes when the style looks weak. Anyway, if you’re searching for a solid FLUX ComfyUI workflow, this should save you some testing time.
Resources mentioned:
• Free USO ComfyUI workflow: https://aistudynow.com/uso-comfyui-workflow-for-flux-style-from-image-same-face/
#USOComfyUI #FLUX #StyleTransfer USO (New IP-Adapter Flux) ComfyUI Workflow: Image Style Transfer, Same Face, Full Demos](https://i.ytimg.com/vi/_FrdkX0ReR8/mqdefault.jpg)



![WAN 2.1 MAGREF ComfyUI Workflow: Turn 3 Image into Video in 5 Steps [2025]
The WAN 2.1 MAGREF ComfyUI workflow which i have created, worked way better than I expected — and honestly, I was able to generate high-quality videos in just 5 steps using this LoRA model name Wan21_T2V_14B_lightx2v_cfg_step_distill_lora_rank32.safetensors.
In this video, youll learn:
• How to create AI-generated videos from a single image
• Why this workflow works great with multiple characters
• A quick trick to manually add background support even though it’s not built-in
The setup is pretty straightforward once you know where everything goes. I used the WanX 21 - LightXV LoRA because it generates videos faster and only takes 5 steps.
For backgrounds, there’s no direct third input, but I found a way using the Image Concatenate (Multi) node. It’s not perfect, but it opens up creative options if you’re willing to play around.
Resources mentioned:
https://aistudynow.com/how-to-use-wan-2-1-magref-in-comfyui-5-step-ai-video-workflow/
#wan21 #magref #comfyuiworkflow #aivideogeneration WAN 2.1 MAGREF ComfyUI Workflow: Turn 3 Image into Video in 5 Steps [2025]](https://i.ytimg.com/vi/avrdKIiyq2Y/mqdefault.jpg)





