Upload a photo, describe the movement you want, and download a video clip a few minutes later. That is the entire workflow. Lumineer's image-to-video workspace runs on the same engine as our full AI video generator, which hosts 17 models — including Seedance 2.0, Kling 2.5, Veo 3, and LTX — so you can animate the same image with different engines until the motion looks right.
A single clip starts at 15 credits, and every new account gets 100 free credits with no card required. That is enough to test the pipeline on your own images before you pay anything.
How image-to-video AI works on Lumineer
An image-to-video (i2v) model uses your still image as the literal first frame of the video, then predicts how the scene should move over the following seconds based on your text prompt. Unlike text to video, where the model invents the scene from scratch, i2v locks the composition, subject, and style to your upload — which is exactly what you want when the image is a real product, a real face, or a finished artwork.
The process in the playground has four steps:
- 1Upload your image — JPG, PNG, or WebP.
- 2Write a motion prompt describing what should move and how the camera behaves.
- 3Pick a model based on the quality level and credit budget you need.
- 4Generate. Most clips render in one to four minutes; Creator and Studio plans skip ahead of the free queue.
The output ratio matches your image automatically
You do not need to crop or letterbox anything before uploading. Lumineer reads the aspect ratio of your image and sets the output format to match: a 9:16 phone photo becomes a 9:16 vertical video, a 4:5 product shot stays 4:5, a 16:9 landscape stays widescreen. Tools that force a fixed frame either crop your subject or stretch it — both ruin product shots and portraits, so this default saves real cleanup time.
First frame and last frame
Every i2v generation uses your upload as the first frame. Several models — including Seedance 2.0 I2V and Kling — also accept an optional last frame: a second image the clip must end on. That turns a single generation into a controlled transition: a product rotating from front view to side view, a portrait turning from profile to camera, a pencil sketch resolving into the finished painting. Provide only a first frame and the model improvises the ending; provide both and you control it.
Image-to-video models compared
These are the most-used i2v models on Lumineer, with real credit costs per clip:
| Model | Credits per clip | Max duration | Resolution | Strengths |
|---|---|---|---|---|
| Seedance 2.0 I2V | ~250 | 10s | 1080p | Best motion realism and facial fidelity; supports last-frame control |
| Seedance 2.0 Fast I2V | ~100 | 10s | 720p | Same motion engine at 60% less cost — the iteration workhorse |
| Kling 2.5 I2V | ~120 | 10s | 1080p | Strong prompt adherence, expressive character and body motion |
| Grok Imagine I2V | ~60 | 6s | 720p | Quick stylized motion, well suited to social clips |
| LTX I2V | ~15 | 5s | 768p | Cheapest per clip; ideal for drafts, motion tests, and high volume |
How to choose
The practical workflow is two-tiered. Draft on LTX at 15 credits: run three or four motion variations of the same image for under 60 credits and pick the movement that works. Then re-run the winning prompt on Seedance 2.0 I2V at ~250 credits for the final render. You spend roughly 310 credits total instead of 750–1,000 credits brute-forcing variations on the premium model.
Seedance 2.0 Fast sits in between: the same motion behavior as full Seedance at 720p, useful when the clip will be viewed on a phone anyway. Kling 2.5 is the pick when your prompt involves a person doing something specific — gestures, turns, expressions — and Grok Imagine covers fast, stylized social content at 60 credits.
What people animate with i2v
Old family photos
This is the most personal use case: a scanned photo of a grandparent, animated with a blink, a slow smile, a slight head turn. The rules here are strict — keep motion minimal, because faces are where artifacts show first. Use Seedance 2.0 I2V for its facial fidelity, keep clips to around 5 seconds, and prompt for one gesture only. A working prompt: "Subtle animation: the woman blinks naturally and her smile deepens slightly, hair moves gently as if in a light breeze, camera completely static, photorealistic." Scans work fine as input, though results improve if you crop out damaged borders first.
E-commerce product shots
A static packshot becomes a hero loop: steam rising from a coffee bag's cup, fabric swaying on a hanger, a slow 180° rotation of a sneaker, liquid pouring into a glass. Because output matches your input ratio, a 4:5 catalog image comes back as a 4:5 clip that drops straight into a product page or Instagram feed. If you sell the results or use them in ads, note that the commercial license is included from the Creator plan ($69/month) upward — see pricing for the exact terms.
Portraits and headshots
Ambient-motion portraits — breathing, blinking, slight parallax — turn a static headshot into a living element for a personal site, a team page, or a video intro. Kling 2.5 and Seedance 2.0 both handle this well; the differentiator is that Kling follows explicit gesture instructions ("she looks up and smiles") more literally.
Artwork and illustrations
Painters and digital artists animate finished pieces for reels and shop promos: drifting clouds in a landscape painting, flickering candlelight in a still life, a slow push-in on an illustration's focal point. Because i2v preserves your exact style and linework, the result reads as your art in motion, not an AI reinterpretation of it.
Prompting tips for photo animation
Motion prompts are not image prompts. The model already sees your image — spend every word on movement. Our guide on how to animate a photo goes deeper, but these four rules cover most failures:
Describe motion, not the scene
Skip "a woman in a red dress in a garden" — the model knows. Write "her dress and hair move in a gentle breeze, she turns her head slowly toward the camera."
Keep movement subtle
Large motions force the model to invent pixels it cannot see, which is where warped hands and melted faces come from. Words like gentle, slow, slight, and subtle consistently produce cleaner clips than dramatic or fast, especially on faces.
Direct the camera explicitly
If you say nothing about the camera, the model picks for you. Say it: "slow push-in," "camera static," "gentle orbit left," "handheld drift." On product shots, "camera static" plus subject motion usually beats a moving camera.
One subject, one motion
A prompt listing five actions gets you five half-rendered actions. Pick the single movement that sells the clip; if you need more, chain clips or use a last frame as the bridge.
What it costs in practice
Lumineer runs on credits, and the i2v math is straightforward:
- Free start — 100 credits, no card: three LTX clips to test your images.
- Starter, $19/month — 1,800 credits: about 80 LTX drafts, or 12 Seedance Fast clips, or 4 full Seedance 2.0 renders.
- Creator, $69/month — 6,800 credits: roughly 18 Seedance 2.0 finals or 45 Fast clips, plus the commercial license and priority queue.
- Studio, $199/month — 20,000 credits: around 56 Seedance 2.0 renders, with 4K output and API access for automated pipelines.
- Credit packs from $9 that never expire, for one-off projects without a subscription.
Full details are on the pricing page. If you are deciding between a monthly plan and a pack: packs suit occasional projects, plans suit anyone rendering weekly.
Frequently asked questions
What is image to video AI?
Image to video AI takes a still image and generates a short video from it, using the image as the exact first frame. A text prompt controls what moves and how the camera behaves, while the composition and style of the original image are preserved. On ai-lumineer.com this works with models such as Seedance 2.0, Kling 2.5, and LTX.
How much does it cost to animate a photo with AI?
On Lumineer, one animated clip costs between roughly 15 credits (LTX) and 250 credits (Seedance 2.0 I2V). Plans start at $19/month for 1,800 credits, one-time credit packs start at $9 and never expire, and new accounts get 100 free credits with no card required.
Can I animate old family photos?
Yes — scanned photos are a common input, and the key is prompting for subtle motion only: a blink, a slight smile, gentle hair movement. Seedance 2.0 I2V handles faces with the highest fidelity among Lumineer's models. Cropping out damaged borders before upload noticeably improves results.
Will the video keep my image's aspect ratio?
Yes, automatically. Lumineer detects the ratio of the uploaded image and renders the video in the same format — 9:16 stays vertical, 4:5 stays 4:5, 16:9 stays widescreen. No pre-cropping or letterboxing is needed.
Can I use the generated videos commercially?
Videos made on the Creator plan ($69/month) and Studio plan ($199/month) include a commercial license, covering client work, ads, and product listings. Free credits and the Starter plan are intended for personal and evaluation use. The current terms are listed on the Lumineer pricing page.
What image should I upload for the best result?
A sharp, well-lit image of at least 720 pixels on its short side, in JPG, PNG, or WebP. Higher-resolution inputs give the model more detail to move, and a single clear subject animates more reliably than a busy scene. Heavy compression artifacts in the source tend to get amplified in motion.
