LTXV0.9.5: The Fastest Video Generation T2V&I2V

Looking for a lightning-fast video generation tool? Meet LTXV0.9.5, the speed king of video creation models. Powered by high-end hardware like the Ultra-pro (L40S), it churns out a 5-second, 24 FPS video in just 50 seconds. Whether you’re turning text into animations or transforming static images into dynamic clips, LTXV0.9.5 makes it happen seamlessly in ComfyUI. In this guide, we’ll break down three killer workflows—Text-to-Video, Simple Image-to-Video, and Multiple Images-to-Video—complete with workflow breakdowns and pro tips to get you started!

Why LTXV0.9.5 Stands Out

LTXV0.9.5 isn’t just fast—it’s a game-changer. With optimized algorithms and ComfyUI’s node-based design, it offers unmatched flexibility. Here’s what makes it shine:

• Blazing Speed: Generates a 5-second video in 50 seconds on Ultra-pro (L40S)—faster than real-time playback.
• Versatile Workflows: Handles text, single images, or multiple images with ease.
• User-Friendly: Perfect for beginners and pros alike in ComfyUI.

Let’s dive into the three workflows and see how LTXV0.9.5 can supercharge your video projects.

1. Text-to-Video Workflow: Turn Words into Video Magic

Workflow at a Glance

• Key Nodes: Text Input → LTXV Model Loader (ltx-video-2b-v0.9.5.safetensors) → Text Encoder (t5xxl_fp16.safetensors) → LTXV Preprocessing → Sampling → Video Output.
• Output: A 5-second, 24 FPS video at 768x512 resolution.

How to Do It

1. Add a Text Input node in ComfyUI and type a prompt like “A serene lake at sunrise, rippling water, morning mist rising, birds flying overhead.”
2. Load the LTXV0.9.5 model (from ComfyUI/models/checkpoints) and the text encoder (from ComfyUI/models/text_encoders).
3. Use the LTXV Preprocessing node to encode your text, set 20 diffusion steps, and hit run.
4. In about 50 seconds, you’ll have a smooth, high-quality video ready to roll.

Best For

This workflow is ideal for whipping up creative content fast—think short ads, concept demos, or artistic animations. LTXV0.9.5’s speed and detail are a total win.

2. Simple Image-to-Video Workflow: One Image, Endless Motion

Workflow at a Glance

• Key Nodes: Image Loader → LTXV Preprocessing → LTXV Guide (LTXVAddGuide) → LTXV Model Loader → Sampling → Video Output.
• Output: A 5-second, 24 FPS video based on a single image.

How to Do It

1. Upload a 768x512 image (like a landscape shot) to the Image Loader node.
2. Run it through the LTXV Preprocessing node to prep it for the model.
3. Set the image as a guide frame (usually the first frame) with the LTXVAddGuide node.
4. Load LTXV0.9.5, run the sampling, and boom—50 seconds later, you’ve got a dynamic video.

Best For

Super simple and effective, this workflow turns static photos into motion masterpieces. Perfect for product showcases or animating nature shots—great for beginners too!

3. Multiple Images-to-Video Workflow: Control the Story with Keyframes

Workflow at a Glance

• Key Nodes: Multiple Image Loaders → Multiple LTXV Preprocessing → Multiple LTXV Guides → LTXV Model Loader → Sampling → Video Output.
• Output: A 5-second, 24 FPS video with smooth keyframe transitions.

How to Do It

1. Upload multiple images (e.g., a start frame and end frame) into Image Loader nodes.
2. Process each with LTXV Preprocessing and assign them as keyframes in LTXVAddGuide nodes (e.g., 0 for start, -1 for end).
3. Load the LTXV0.9.5 model, run it, and watch it stitch everything together in 50 seconds.

Best For

This workflow gives you serious control—ideal for storytelling or projects needing smooth transitions, like evolving scenes or narrative clips.