How to Generate AI Videos from Text or Images: Full Guide (2026)

AI video generation creates short video clips from a text prompt or a still image, complete with camera motion and animated subjects. NamiFusion's AI Generator gathers the leading text-to-video and image-to-video models into one interface, so you pick the right engine per shot instead of juggling separate subscriptions. New accounts get free credits, and every model card shows its approximate price per video before you spend anything ($1 = 100 credits).

TL;DR

Generate videos from plain text (text-to-video) or animate a still image (image-to-video).

Top video models aggregated in one place — compare them on the same prompt and keep the best.

Chain your workflow: generate an image, click 'use as reference,' and turn it into a video without leaving the page.

Seeds, batch runs, and on-page history make iteration systematic instead of random.

Free signup credits; pay-as-you-go at $1 = 100 credits with per-video prices shown up front.

How do you generate an AI video from text?

1. Open the AI Generator, sign in, and switch to a text-to-video model — free signup credits cover your first runs.

2. Check the model card for its approximate per-video price and supported options.

3. Describe the shot: subject, action, setting, camera movement, and mood.

4. Hit generate and let the clip land in your on-page history.

5. Compare a few takes and reuse the seed of the best one for controlled variations.

6. If the motion is not right, rerun the same prompt on a different model — switching models is one click, not a new subscription.

How do you turn an image into a video?

Image-to-video starts from a picture — upload a photo, or use an image you just generated — and the model animates it, using your image as the visual anchor. When you care about composition, a character's look, or a product's appearance, this route is far more controllable than text alone.

The workflow I recommend on NamiFusion is the chained one: polish the look cheaply as a still image first — images cost far less than videos — then click 'use as reference' and hand it to an image-to-video model. You lock the look in static form, and only pay video prices once you are happy. For anything where the subject's appearance matters, this is the way.

What are people actually making with AI video?

Short-form content is the volume king: hooks, B-roll, and visual gags that would be unfilmable on a real budget. Marketers render products into impossible settings, musicians storyboard an entire music video before booking a single location, and studios lean on text-to-video for previsualization — cheap moving drafts of expensive shots.

My own most common use is filler footage: when an edit needs three seconds of 'drone shot over a coastline at dusk,' generating it costs less than licensing stock and matches the exact grade I need. Combined with the image-to-video chain above, you can keep the same character or product consistent across every clip in a series.

How do you write prompts for AI video?

Describe motion explicitly — that is the biggest difference from image prompts: what moves, how fast, and what the camera does (dolly, pan, locked off). A video prompt is a shot description, not an image caption.

Keep each prompt to one scene and one continuous action. Cramming a multi-scene storyline into a single clip almost always fails — generate the shots separately and cut them together in an editor.

A slow dolly-in on an astronaut walking through a sunflower field
at golden hour, petals drifting in the wind, warm backlight and
lens flare, cinematic color grading, smooth stabilized camera motion

How much does AI video generation cost?

It starts free: your signup credits work on video generation too. Beyond that, credits cost $1 = 100. Video is inherently pricier than images and rates vary noticeably between models, but every model card shows an approximate '$X per video' figure, so you know the cost before you run.

The budget-friendly rhythm: iterate on composition and look as still images first (images are far cheaper), animate only when settled; draft with a faster, cheaper video model, and save the flagship model for the final render.

Also, every model in the API marketplace has a free Playground, so you can gauge a specific video model's quality without topping up first.

Can you generate videos through the API?

Yes. Video models sit behind the same unified REST API as the rest of NamiFusion's 98+ marketplace models: create a key at /api-keys, POST a job, then poll the task endpoint. Video generation takes a while, so be generous with your polling interval.

For pipelines that render many shots — say, storyboarding an entire short episode — this API plus per-call credit pricing beats clicking through a web UI shot by shot.

curl -X POST "https://www.namifusion.com/api/v1/marketplace/run/<model_id>" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "input": {
      "prompt": "a slow dolly-in on an astronaut walking through a sunflower field at golden hour"
    }
  }'

# Video jobs take longer — poll the task until it completes
curl "https://www.namifusion.com/api/v1/marketplace/run/tasks/<task_uuid>" \
  -H "Authorization: Bearer YOUR_API_KEY"