Try LTX 2.3 Now

Examples of LTX 2.3

Why You’ll Love It

Sharper Detail

Preserves fine textures, hair, text, and edges with enhanced clarity and improved visual fidelity.

Prompt Precision

Accurately follows complex prompts, handling multiple subjects, layouts, and detailed stylistic instructions.

Stronger Motion

Delivers smoother, more natural motion with improved consistency and fewer visual distortions.

Cleaner Audio

Produces synchronized audio with reduced noise, fewer gaps, and clearer overall sound quality.

Native Portrait

Generates true vertical 1080 × 1920 videos optimized for portrait viewing, not cropped from landscape.

Better Image-to-Video

Ensures stable motion from images with less freezing, reduced drift, and improved frame consistency.

What Creators Are Saying

Nadia Mercer
LTX 2.3 feels much sharper than older video models. Fine textures and motion hold up beautifully.
Tobias Lindgren
The prompt accuracy is a huge step up. It handles more detailed creative instructions without falling apart.
Arjun Mehta
Native portrait video is a big win for short-form content. The output feels ready for production.

How LTX 2.3 Works

Step 1

Choose LTX 2.3

Select the LTX 2.3 model from MagicShot’s video generation options.

Step 2

Add Your Prompt

Describe your scene, motion, style, and subject details for accurate video output.

Step 3

Generate Video

Create high-quality videos with sharper visuals, cleaner audio, and smoother motion.

Frequently Asked Questions

LTX 2.3 is the latest release in the LTX-2 model family, a diffusion transformer video model that generates high-fidelity video and synchronized audio from one model. It supports text-to-video, image-to-video, and audio-to-video generation up to 1080p, including native portrait video.

LTX 2.3 improves four major areas over LTX 2: sharper detail, tighter prompt adherence, native portrait video support, and cleaner audio quality.

Yes. LTX 2.3 supports image-to-video generation and improves motion consistency, reducing freezing and unwanted artificial movement.

Yes. It supports native 9:16 portrait generation up to 1080 × 1920, trained specifically for portrait orientation instead of cropping from landscape.

Yes. LTX 2.3 generates synchronized audio and improves cleanliness by reducing silence gaps and training noise artifacts.

Yes. The model weights are available on HuggingFace under an open license, including base, fp8, and distilled variants.

Yes. Multiple checkpoints are available for local use, including full, quantized, distilled, and latent upscaler variants.

Yes. LTX 2.3 supports ComfyUI with updated custom nodes and reference workflows for text-to-video, image-to-video, and multi-stage generation.