How to Make an AI Video from Text or Photos (Beginner Guide)
You can make a complete AI video in under five minutes: write a prompt or upload a photo, pick a model like Kling or Seedance, and generate. This beginner guide walks through every step, with real costs and prompt examples.
How to Make an AI Video from Text or Photos
Making an AI video takes three decisions: what the video shows (your prompt or photo), which model renders it, and how long it runs. Everything else is a button. This guide walks a complete beginner from empty page to finished clip in about five minutes.
The two ways to make an AI video
1. Text to video — describe a scene and the model films it: "a chef flambéing a pan in a busy kitchen, handheld camera, steam and sparks". Best for scenes that never existed.
2. Image to video — upload a picture and describe the motion. Your photo becomes the first frame and comes to life. Best when you already have the subject — yourself, a product, an artwork. (Full image-to-video guide here.)
Both run in Reelina's AI Studio in the browser.
Step by step
- Open the studio and choose the video tab.
- Write the prompt (or upload a photo + motion prompt). Structure it as subject → action → camera → mood: "an old sailor mending a net · he looks up and smiles · slow dolly-in · warm evening light".
- Pick a model. Kling 3.0 Pro for expressive people, Seedance 2.0 for cinematic physics, WAN 2.7 for speed and value, Reelina v1 as the balanced default (12 coins / 5s). The exact coin cost is shown before you generate.
- Generate. A 5-second clip renders in roughly one to three minutes.
- Finish it. Extend to make it longer, upscale to sharpen, add a voice with the Voice Generator or sync lips with Lip Sync, and burn in subtitles.
What does an AI video cost?
Reelina prices per generation in coins, always shown up front. Short clips start around 12 coins. On the $9.99 Starter plan (1,000 coins) that is about 60–80 short videos monthly; Creator ($19.99) and Pro ($29.99) scale that up, and unused coins never expire.
Prompt patterns that produce good videos
- Action first: models follow verbs better than adjectives — "she spins, her dress flares" beats "a beautiful spinning scene".
- One scene per clip: don't script a story arc into 5 seconds; chain clips instead.
- Name the camera: "static tripod shot", "drone orbit", "handheld follow" dramatically change the feel.
- Steal from templates: the template gallery is full of proven scenes — run one with your photo, then study its motion.
Common beginner mistakes
- Writing a paragraph when a sentence renders better.
- Judging a model on one attempt — regenerate once or twice; diffusion has variance.
- Rendering 15 seconds before nailing 5.
- Skipping the upscaler before posting — it is the cheapest quality upgrade there is.
FAQ
Can I make an AI video for free? Video generation runs on serious GPUs, so free tools are heavily limited. Reelina starts at $9.99/month for ~60–80 short clips.
How long can an AI video be? Single generations run 5–15 seconds; the extend tool chains them into longer cuts.
Can AI make a video of me? Yes — upload your photo and use image-to-video, or swap yourself into an existing video.
Which AI video model is best? Depends on the shot: people → Kling, cinematic scenes → Seedance, value → WAN. Compare them on one prompt inside the studio — see also our best AI video generator comparison.