23 models · start frame, end frame or references

AI video generator
from image.

Leaxor's AI video generator turns an image into a video clip. Attach a photo, describe how it should move, and pick one of 23 models, Kling v3, Veo 3.1, and Seedance 2.5 among them, to animate it from 1 credit a clip. Give it two images and it fills in the motion between them. Give it up to 4 and it keeps those faces and props the same through the shot.

Animate a photo See the 23 models

Same price as text to video13 models add sound9 video labs, one balance

A dragon on a volcanic ridge at dusk, jaws closed, smoke curling from its nostrils — Drag the line. Left is the photo, right is the clip that started from it.
Still · Flux 2 Pro · 1 creditClip · Kling v3 Pro · 5s · 7 creditsRendered in 1 min 58 s
Both prompts, word for word
Still
A photorealistic dragon perched on a jagged volcanic ridge at dusk, seen in profile facing right, jaws closed, thin smoke curling from its nostrils and a faint orange glow between its teeth. Charcoal-grey scales with a wet sheen, torn leathery wings half folded, claws gripping black rock. Behind it a stormy sky in deep blue and ember orange, ash drifting through the air. The right third of the frame is open sky. Wildlife photograph on a 400mm telephoto lens, shallow depth of field, crisp detail on every scale, low sun breaking through the clouds behind it. No text, no watermark, no logo, no border.
Clip
The dragon lifts its head, opens its jaws and breathes a long, roaring jet of fire to the right across the open sky. The flames light its scales and the drifting ash from the front, heat haze ripples the air, then the fire dies back to smoke as its jaws close. The camera holds still.

Three ways to start a video from images

How many pictures you bring decides the job. One photo is the usual start. Two pin down the beginning and the end. Several keep a character looking like themselves.

One photo

Text/Image to Video · 23 models

The photo becomes the first frame and your prompt says what happens next. Every clip model on Leaxor takes one, and attaching it doesn't change the price.

Use it for: product shots that need to move, a portrait that turns to camera, a landscape with weather rolling in.

Two photos

Frames to Video · 12 models

An opening frame and a closing one. The model invents everything in between, so the clip lands exactly where you want it. On 7 of them the end frame is optional.

Use it for: before-and-after reveals, transitions between two scenes, a day-to-night change.

Several photos

Reference to Video · 5 models

Up to 4images the shot has to honour: the same face, the same jacket, the same room. They aren't frames. The model builds a new shot around them.

Use it for: a recurring character across a series, a product shown in a new setting.

Frames to Video runs on Veo 3.1 Lite, Flux 3 Draft, Kling 2.6 Pro, MiniMax H3, Flux 3, Kling v3, Kling v3 Pro, Kling O3 Pro, Veo 3.1 Fast, Veo 3.1, Kling O1, and Kling O1 Pro. Reference to Video runs on Grok Imagine 1.5, Wan 3.0, Gemini Omni Flash 1.1, Wan 3.0 Prime, and Kling 1.6 Elements.

Which models animate a photo

All 23 clip models, from ByteDance, Google, PixVerse, Alibaba, MiniMax, Black Forest Labs, Kuaishou, xAI, and Lightricks. Two things change with a photo attached, and the table shows both.

Framing. On 12 models the photo decides the shape of the clip, so a portrait photo gives a portrait clip. The rest let you pick a shape and fit the photo to it.
Sound. 13 of them can generate audio with the clip. The others are silent.

Every image to video model on Leaxor, its lab, how it handles the photo's framing, whether it adds sound, its longest clip, and its starting price in credits
Model	Lab	Framing	Sound	Longest	From
Seedance 1.0 Pro	ByteDance	9:16 · 16:9 · 1:1	—	12s	1 credit · $0.10
Veo 3.1 Lite	Google	9:16 · 16:9	Yes	8s	2 credits · $0.20
PixVerse V6	PixVerse	Your photo's	Yes	15s	2 credits · $0.20
Wan 2.5	Alibaba	Your photo's	—	10s	3 credits · $0.30
MiniMax H3 Max	MiniMax	Your photo's	—	15s	3 credits · $0.30
Hailuo 2.3	MiniMax	Your photo's	—	10s	4 credits · $0.40
Flux 3 Draft	Black Forest Labs	9:16 · 16:9 · 1:1	Yes	20s	4 credits · $0.40
Gemini Omni Flash 1.1	Google	9:16 · 16:9	—	10s	1 credit · $0.10
Wan 3.0	Alibaba	9:16 · 16:9 · 1:1	Yes	15s	3 credits · $0.30
MiniMax H3	MiniMax	Your photo's	—	15s	3 credits · $0.30
Kling 2.6 Pro	Kuaishou	Your photo's	Yes	10s	4 credits · $0.40
Wan 3.0 Prime	Alibaba	9:16 · 16:9 · 1:1	Yes	15s	4 credits · $0.40
Hailuo 2.3 Pro	MiniMax	Your photo's	—	6s	6 credits · $0.60
Grok Imagine 1.5	xAI	Your photo's	—	6s	6 credits · $0.60
LTX-2.5 Fast	Lightricks	9:16 · 16:9	—	10s	7 credits · $0.70
LTX-2.5 Pro	Lightricks	9:16 · 16:9	—	10s	9 credits · $0.90
Flux 3	Black Forest Labs	9:16 · 16:9 · 1:1	Yes	20s	10 credits · $1.00
Kling v3	Kuaishou	Your photo's	Yes	15s	3 credits · $0.30
Kling v3 Pro	Kuaishou	Your photo's	Yes	15s	4 credits · $0.40
Kling O3 Pro	Kuaishou	Your photo's	Yes	15s	4 credits · $0.40
Veo 3.1 Fast	Google	9:16 · 16:9	Yes	8s	5 credits · $0.50
Veo 3.1	Google	9:16 · 16:9	Yes	8s	10 credits · $1.00
Seedance 2.5	ByteDance	Your photo's	Yes	30s	13 credits · $1.30

“From” is the shortest, lowest-resolution render without sound. Longer clips, higher resolution and sound cost more, and the dashboard shows the exact figure before you generate.

How to make a video from a photo

Four steps, and the first frame is already done.

Pick Text/Image to Video and a model
In video mode, choose Text/Image to Video and one of the 23 models. Each one shows its price for the length you pick before anything runs.
Attach the photo
Drag the image onto the prompt box or tap attach. It becomes the clip's opening frame, so the subject, the light and the composition are already decided.
Describe the motion
Write what happens next and how the camera moves: “she turns toward the window as the camera slowly pushes in.” The photo has already said what everything looks like.
Set the length and generate
Choose the length, and sound if the model offers it, then generate. To keep going, the finished clip's Last frame button starts a new clip from where this one ended.

Getting a good clip from a still

Working rules, drawn from Google's documentation for Veo rather than from side-by-side tests on this page. They hold up on the other models too, because every one of them starts from the same place: your first frame.

Start big, in the right shape

Google recommends a starting image of 720p or larger, in 16:9 or 9:16, and warns that other shapes may be resized or centrally cropped. Crop the photo yourself first, so you choose what gets cut rather than the model.

Write the verb, not the picture

Google's prompt guide calls the action “the verb of your video”. With a photo attached the subject and setting are already on screen, so spend the words on what moves and what the camera does.

One movement per clip

A clip of a few seconds has room for one clear action. Ask for three and you tend to get a blur of all of them. Chain clips with Last frame instead.

Need a set ending? Give it one

If the clip has to land on a specific image, don't describe it and hope. Google documents first-and-last-frame generation for exactly this, and Frames to Video runs it on 12 models here.

No photo yet? Make the still first

The cheapest way to a specific shot is often an image first. A still costs 1–3 credits on Leaxor's image generator, across 10 models, so you can settle the composition for a few cents and only then spend video credits on the motion. The cheapest still plus the cheapest clip comes to 2 credits ($0.20). The prompt library has full image prompts to start from.

Both live in the same account and the same balance. There's no one-click hand-off yet: save the image, switch to video mode and attach it.

Animating an old family photo? Fix it before you animate it, because scratches and fading in the first frame get animated too. The free photo restoration and colorizer tools clean the scan first. Only animate photos you have the right to use, of people who'd agree to it.

What it costs

Pay as you go. Credits cost $0.10 each, you buy at least 50 at a time, and they never expire. There is no subscription and no free tier.

What each image-to-video job costs in credits
A clip from one photo	From 1 credit, by model and length. The same as from text.
A clip between two frames	From 2 credits, by model and length.
A still to animate	1–3 credits, by image model.

Failed generations are refunded automatically. Full details on the pricing page.

AI video from an image — frequently asked questions

Is there a free AI video generator from image?

Leaxor isn't one: there's no free tier and no free credits. You pay per clip instead, and a clip from a photo starts at 1 credit ($0.10) on the cheapest models. Credits cost $0.10 each, you buy at least 50, there's no subscription, and unspent credits never expire.

How does AI turn a photo into a video?

An image-to-video model uses your photo as the first frame and generates the frames that follow it, guided by your prompt. That's why the subject, lighting and composition hold: the model starts from them instead of inventing them. Your words decide what moves and how the camera behaves.

Can I make a video from an image and text together?

Yes, and that's the normal way to do it. The image sets the opening frame and the text says what happens next. A prompt that describes motion and camera work gets more out of the photo than one that describes the photo again.

Can image to video AI add sound?

13 of the 23 models can generate sound along with the clip, including Kling v3, Veo 3.1, and Seedance 2.5. The table on this page marks which. Sound can raise the price of a render, and the dashboard shows the figure with it on before you generate.

What is the best image to video AI?

It depends on the shot. Kling v3, Veo 3.1, and Seedance 2.5 are the models most people compare, each with its own look. Seedance 1.0 Pro and Gemini Omni Flash 1.1 start at 1 credit, which makes them good for testing a photo before you spend more. Run the same image on two models and keep the better clip.

Is there an image to video AI with no restrictions?

Not here. Each model applies its maker's content rules, so some images and prompts are refused, most often explicit ones and pictures of real, identifiable people. Only animate photos you have the right to use, and people who have agreed to it.

Can AI animate old photos?

Yes, and it works better if you fix the photo first. Scratches and fading become part of the first frame, so the model animates them too. Restore and colorize the scan, then attach the clean version and describe a small, natural movement.

How long can a video from an image be?

It depends on the model. Seedance 2.5 goes up to 30 seconds; the others stop between 6 and 20. For something longer, the Last frame button on a finished clip starts the next one from where it ended, and Extend Video continues a clip you already have.

Your photo is frame one.
Write what happens next.

23 video models, one prompt box, and a balance that never expires.

Animate a photo