LTX-2 Image to Video in ComfyUI: Turn Any Still Image into a Realistic AI Video

Prompt used

Animate the car accelerating directly toward the camera at high speed. As it passes close by, the camera pans left smoothly to follow the car's aggressive drive down the road.

Add realistic motion blur on the background and wheels, with light tire smoke and dust trailing behind the car. Introduce a subtle handheld camera feel for intensity, as if the camera operator is struggling to keep up.

The car then initiates a fast drift, sliding sideways and turning around in a controlled arc. Tire smoke increases briefly as the car completes the turn.

After the drift, the car drives back toward the camera, accelerating into an extreme close-up of the front end, headlights and grille filling the frame. End on a dramatic, high-speed stop or pass-by.

Motion should feel fast, cinematic, and realistic, with smooth vehicle physics and dynamic camera movement.

Prompt used

A bright, playful stage scene with a colorful puppet character standing center frame, holding a microphone. Medium, front-facing shot. The puppet sings an upbeat, cheerful song with clear lip-sync, its fabric mouth opening and closing rhythmically.

Lyrics sung clearly:
🎵 "Lights are on, the beat is right, Sing it loud, feel the night!" 🎵

The puppet bounces lightly, nods its head to the beat, blinks expressively, and waves one hand in a playful, comedic way. Movements are smooth and well-timed with the music.

The camera stays steady with a gentle push-in. Bright, even lighting highlights vibrant colors and soft puppet textures. The performance ends with the puppet raising the microphone and striking a silly, triumphant pose.

Prompt used

Cinematic, high-tension night scene on a busy city street. The woman stands still under glowing streetlights, neon reflections shimmering on wet asphalt behind her. She slowly turns her head toward the camera, her expression focused and controlled, breathing steady but tense.

The camera gently pushes in toward her face, capturing subtle eye movement, a soft blink, and a slight tightening of her jaw. She speaks calmly but firmly: "This just got serious."

She shifts her weight slightly, adjusting her stance, her hair moving gently in the night air. Traffic lights and headlights blur softly in the background, creating depth and motion without chaos.

The camera holds on her face in a steady medium close-up, city lights flickering behind her as the tension lingers, ending on her intense, unwavering gaze.

Prompt used

A bright daytime outdoor scene. The camera starts in a medium shot, tracking a confident young woman as she continues walking forward along the pathway, her movement smooth and natural with proper weight shift and relaxed arm swing. She looks slightly toward the camera, blinking naturally, her facial muscles moving subtly as her lips begin to move as if she is speaking calmly.

As she walks, she makes a small, natural hand gesture and gently turns her head, maintaining balanced posture and realistic body motion. The camera maintains a steady forward follow with slight cinematic micro-movement, keeping her face and upper body in clear focus while the background remains softly blurred.

Her hair and clothing respond subtly to her movement, and the lighting stays consistent and even throughout, emphasizing realism and lifelike motion.

Update via the update folder

Go to your ComfyUI_windows_portable\update folder and double-click update_comfyui to update to the latest version.

Update all nodes via Manager

Launch ComfyUI, open the Manager, and click 'Update all' to update every installed custom node.

Switch to v0.8.1 or higher

Click Switch Version in ComfyUI. Ensure the version is v0.8.1 or higher.

Set channel to dev / nightly

Confirm your channel is set to dev and the update version is ComfyUI nightly for the latest compatibility.

Open the workflow

Open ComfyUI and drag-and-drop the downloaded JSON workflow file onto the canvas.

Select the checkpoint

Click on any node or subgraph containing model selection. Confirm CKPT_name is set to ltx-2-19b-dev-fp8.safetensors.

Select the text encoder

In the LTXV audio text encoder loader node, select gemma_3_12B_it_fp8_e4m3fn.safetensors.

Add your input image

Load your source image in the image input node. This is the still image that will be animated.

Write your animation prompt

Enter a detailed description of the motion you want — camera direction, character movement, lighting changes, and scene dynamics all help guide the model.

Queue Prompt

Click 'Queue Prompt' to run the workflow. The output will be a smooth, high-quality AI video saved to your ComfyUI output folder.

LTX-2 I2V is a generative AI model that animates still images into smooth, realistic videos inside ComfyUI. It supports camera movement, facial animation, and dynamic backgrounds from a single input image.

You need: ltx-2-19b-dev-fp8.safetensors (checkpoint), gemma_3_12B_it_fp8_e4m3fn.safetensors (text encoder), ltx-2-19b-ic-lora-detailer.safetensors and ltx-2-19b-distilled-lora-384.safetensors (LoRAs), and ltx-2-spatial-upscaler-x2-1.0.safetensors (latent upscaler).

LTX-2 I2V requires ComfyUI v0.8.1 or higher. Use the Switch Version option and set the channel to dev with ComfyUI nightly for the latest compatibility.

Open run_nvidia_gpu.bat in Notepad and add --reserve-vram 4 to the launch command. Lower this value to 3, 2, or 1 on lower-end GPUs. Also consider increasing your system page file size.

Place the checkpoint in ComfyUI/models/checkpoints/, the text encoder in ComfyUI/models/text_encoders/, LoRA files in ComfyUI/models/loras/, and the upscaler in ComfyUI/models/latent_upscale_models/.