Example inputhover

fps: 25
prompt: "The young farmer gently brushes the grey horse's mane with the wooden brush in slow, caring strokes, smiling warmly. Then he lowers the brush and STOPS brushing completely — the brush stays down and does not come back up. With his free hand he affectionately strokes the horse's head and cheek, and keeps petting until the end of the video. He never brushes again after this point. The horse stays calm, ears twitching slightly, gently leaning into his touch. Soft natural farm lighting, warm peaceful atmosphere, camera stays still."
duration: "auto"
image_url
resolution: "1080p"
aspect_ratio: "16:9"
generate_audio: false

Ltx v2.5 | Image to Video | Pro

Video·ltx-v2.5·by LTX

LTX 2.5 Image-to-Video Pro animates a starting image into a high-quality video with prompt, resolution, duration, audio, and camera-motion controls.

Try it now →

API reference

Runtime (p50): 1m
Estimated price: From $0.12

Call the API

prediction.sh

curl -X POST \
  -H "Authorization: Bearer $EACHLABS_API_KEY" \
  -H "Content-Type: application/json" \
  --data '{
    "model": "ltx-2-5-image-to-video-pro",
    "version": "0.0.1",
    "input": {
        "fps": 25,
        "prompt": "The young farmer gently brushes the grey horse's mane with the wooden brush in slow, caring strokes, smiling warmly. Then he lowers the brush and STOPS brushing completely — the brush stays down and does not come back up. With his free hand he affectionately strokes the horse's head and cheek, and keeps petting until the end of the video. He never brushes again after this point. The horse stays calm, ears twitching slightly, gently leaning into his touch. Soft natural farm lighting, warm peaceful atmosphere, camera stays still.",
        "duration": "auto",
        "image_url": "https://cdn-us.eachlabs.ai/defaults/f33c6431173941bb982b2dc8b5d4b483.png",
        "resolution": "1080p",
        "aspect_ratio": "16:9",
        "generate_audio": false
    },
    "webhook_url": ""
}' \
  https://api.eachlabs.ai/v1/prediction/

Documentation8 sections

Overview
Ltx v2.5 | Image to Video | Pro Overview

Ltx v2.5 | Image to Video | Pro turns a single starting image into a guided video clip with prompt control, camera motion, duration settings, and high-fidelity output. It belongs to the LTX family from Lightricks, and the current API naming appears to use the official model id ltx-2-5-pro for the Pro variant. The main differentiator of Ltx v2.5 | Image to Video | Pro is that it is optimized for fidelity, while still supporting image-to-video generation with camera-motion control and automatic duration behavior in the LTX-2.5 line. For each::labs users, that makes it a strong fit when the source image must preserve visual intent while still feeling animated and cinematic.

LTX-2.5 also adds newer generation features such as native multi-shot support, HDR-oriented workflows, and synchronized audiovisual generation across the family, although the Pro API variant is positioned as the quality-focused option rather than the highest-resolution option.
Capabilities
Capabilities
- Converts a single image into a video clip with guided motion.
- Accepts prompt-based direction for scene behavior and visual style.
- Supports camera motion control for more cinematic framing.
- Uses last_frame_uri in image-to-video workflows for more controlled sequence setup.
- Supports portrait and landscape outputs in the LTX-2.5 API family.
- Offers automatic duration behavior in supported workflows.
- Belongs to a family designed around higher-fidelity generation, with the Pro variant positioned as the quality-oriented option.
- Fits API-driven production workflows on each::labs through the Ltx v2.5 | Image to Video | Pro API integration pattern.
Use cases
Use Cases for Ltx v2.5 | Image to Video | Pro

Creators can animate still artwork into a polished social clip by combining a strong key image with gentle camera motion. A prompt like "slow push-in, subtle light flicker, preserve the composition" fits the model’s image-to-video workflow and helps keep the scene coherent.

Marketers can turn product renders into short motion assets for landing pages or ads. The ability to direct motion and choose portrait or landscape framing makes Ltx v2.5 | Image to Video | Pro useful for campaign variants such as "rotate the bottle slightly, soft reflections, premium studio lighting."

Designers can preview poster concepts, key art, or character illustrations as living scenes. A prompt like "gentle wind movement, cinematic parallax, keep facial structure stable" uses the model’s prompt plus camera-motion controls effectively.

Developers building on the Ltx v2.5 | Image to Video | Pro API can automate asset animation inside content pipelines, especially where repeatable image-to-video generation is needed with a fixed source frame and controlled output duration.
Tips & tricks
Tips and Tricks

Write prompts that describe what should move, how the camera should move, and what must stay stable. LTX-2.5 supports camera motion controls, so prompt and motion settings should reinforce each other instead of competing. Keep the first image compositionally strong, because image-to-video quality depends heavily on the starting frame. When using Ltx v2.5 | Image to Video | Pro, avoid overloading the prompt with too many scene changes; this model is strongest when the scene evolves in a single clear action.

Example prompts:
- "Slow dolly-in on the subject, subtle hair movement, natural daylight, keep the background stable."
- "Gentle handheld camera motion, fabric fluttering in the wind, realistic motion, preserve face identity."
- "Cinematic push-in from the original image, soft parallax, smooth motion, no scene cut."
For best results, pair a concise prompt with the correct duration and aspect ratio, then iterate on motion strength before changing the source image.
Technical spec
Technical Specifications
- Model family: LTX-2.5, from Lightricks, with the Pro API variant named ltx-2-5-pro.
- Category: image-to-video.
- Resolution support: documented for the Pro variant at 720p and 1080p.
- Aspect ratios: 16:9 and 9:16 are supported in the LTX-2.5 API family.
- Duration: the Pro variant is documented as reaching up to 10 seconds; the family also supports automatic duration selection in some API workflows.
- Inputs: starting image, prompt, and optional camera motion; the API also supports last_frame_uri for image-to-video workflows.
- Output: generated video, with the broader LTX-2.5 line also supporting HDR-oriented production workflows.
- Processing time: public LTX timing reports a 10-second clip from an image in 23.7 seconds through the managed API, while self-hosted runs can be faster on specialized hardware.
Things to be aware of
Things to Be Aware Of

The biggest quality issues usually come from weak source images or prompts that ask for too many changes at once. Because this model animates an existing image, scene edits that fundamentally alter layout can produce unstable motion or visual drift. The Pro variant is also not the highest-resolution option in the family, so teams that need 4K output should not expect that from this API path. Keep aspect ratio, duration, and motion settings aligned with the original composition, and avoid treating image-to-video like a full scene re-generation tool.
Key considerations
Key Considerations

Ltx v2.5 | Image to Video | Pro is best used when quality and control matter more than maximum resolution. The Pro variant is documented as a fidelity-focused model that tops out at 1080p, while the broader family includes faster paths for higher resolutions. It works best with a clear source image and a prompt that describes motion, camera behavior, and scene intent in simple language. For each::labs workflows, this model is a good choice when you want a reliable image-to-video API rather than a raw experimental output. If you need longer clips or 4K output, the family’s other variant is the better fit, but the Pro model can still be the safer choice for cleaner motion and tighter visual consistency.
Limitations
Limitations

Ltx v2.5 | Image to Video | Pro is documented as a fidelity-focused variant, but it caps out at 1080p rather than the higher resolutions available in other LTX-2.5 paths. Public documentation also places the clip length at up to 10 seconds for the Pro model. It does not replace a full editing suite, and it is not designed to reliably invent large new scenes, complex multi-shot storyboards, or major structural changes from a single image.