Step-by-step guide: how to create a full AI video

Before you start: what to have on hand

If you're looking for how to make an AI video, here's the short answer: in Reelora it's an eight-step path — from topic to finished file. Nothing to install: the studio runs in your browser and generation happens on servers, so any laptop will do.

More good news: to create a video from text you don't need a ready-made script. One or two sentences about your topic are enough — the AI will build the rest of the outline itself. But a few things are worth deciding before you start:

• A topic or ready text — one sentence or a full script, both work.
• Format: vertical short or horizontal long-form video.
• Language of the video and voiceover.
• Your own photos or videos, if you plan to mix them with generated ones.

How to make an AI video: eight steps

1. Pick format and settings

   Set the approximate length, format — long or short — and language. Everything else depends on it: script structure, number of scenes and frame proportions.

2. Describe your topic

   Write your topic in one or two sentences — and the AI will build a ready outline for your video. No need to invent a script: just clearly say what the video is about and who it's for.

3. Review the outline

   Everything sits right in the outline: voiceover prompts, reference prompts, photo and video prompts — each in its place. Scroll through it and check whether the narrative logic works. Something off? Edit the text right here.

4. Generate references and shots

   Start generating references, then video shots based on the scene prompts. Don't like a particular shot? Regenerate just that one — the rest of the video stays untouched.

5. Add your own footage if needed

   Got footage or photos you've already shot? Add them instead of generated shots at any moment. Lots of material? Just swap the shots for yours — the studio will handle the editing itself.

6. Voice it over

   Pick an ElevenLabs voice: there are many languages, including Ukrainian. The voiceover syncs with scenes and subtitles, so lines land precisely on the shots.

7. Turn on subtitles

   Automatic speech recognition, line-by-line splitting, appearance settings — and subtitles get burned right into the video. This matters especially for vertical shorts: they're often watched without sound.

8. Assemble the edit and grab your video

   Automatic assembly places shots against the voiceover, keeps the rhythm and creates transitions between scenes. Then export — and the finished video is yours: time to publish.

One click or step-by-step control

Once the outline is ready, Reelora offers two paths. The first is one button: generation starts for the whole script, and you get an assembled video without extra moves. This is the option when the topic is simple and you liked the outline right away.

The second path is step by step. You go through the stages manually: generate references, review shots, tweak prompts, replace failed scenes. Slower, but you control every shot — that's how videos get made where every second counts.

• One button — speed: the whole script generates at once.
• Step by step — control: you review and adjust every stage.
• You can copy the outline and feed it to your own trained AI: it already knows the structure, will return a ready script that you simply paste back into Reelora.

Your own footage instead of generated clips

Reelora isn't a factory for purely artificial shots. Any generated photo or video can be swapped for your own file — and vice versa. Editors mix live footage with generated inserts this way: part of the scene is real, part gets filled in.

There's also a special case: a photo the AI refuses to animate due to moderation or an overly complex frame. You can throw effects and overlays onto it that turn a static image into living video.

The key thing: edits are surgical. Regenerate one shot, swap one scene — the rest of the video stays untouched. For more on working with your own files, see the section on your own footage.

What affects the quality of your final video

Set the frame proportions at the very start: reformatting a finished video later is wasted work. Vertical for shorts, horizontal for YouTube. The guide to formats and aspect ratios will help you figure it out.

Second — prompts. The outline gives you ready wording, but you can edit it: the more specific the shot description, the fewer regenerations. More examples and techniques in our collection of video prompts.

And being honest about AI weak spots: hands, small text in frame, long continuous scenes and precise physics are still tricky. Plan scenes so these don't end up center-frame, and just regenerate failed takes separately.

See also

• AI Builds Your Video Blueprint: Long Videos and Shorts in Reelora
• Video Generation Prompts: Ready-Made Examples for Different Genres
• Video aspect ratios: which format to choose for your platform