Creative video

AI Fantas editorial team · Reviewed

AI Filmmaking Workflow: From Story and Shot List to Final Edit

AI filmmaking puts generation inside a filmmaking process: decide what the story needs, plan the shots, create and inspect individual assets, then shape the film through editing and sound. Start with an original five-shot exercise, Paper Star at Night.

The main entry opens a planning prompt draft. A separate shot prompt opens video mode. Review inputs and the current quote before submitting; opening a draft generates no footage.

Make your first version

Prepare your references

  • A one-sentence story, one principal subject and a change that can be shown on screen. An object or unoccupied setting makes a manageable first exercise.
  • Your own or authorized references if a shot needs a particular character, composition or color direction. Keep each reference role explicit.
  • A project folder or canvas for the shot plan, reference images, raw clips and editing versions, plus a limit on attempts and credit spending.

Steps in AI Fantas

  1. Use Studio Chat to plan the story with the planning prompt below. Read and revise the shot list before submitting any generation task.
  2. Give each shot a subject, action, framing, camera instruction and purpose in the edit. Prepare and inspect reference images where identity or composition matters.
  3. Create shots individually in video mode using a model that accepts the intended inputs. Check the aspect ratio, duration and quote for each attempt. Test one shot first.
  4. Review the clips, assemble accepted footage in your editing tool, and add authorized sound and subtitles. Watch the export in full and confirm that the project and original assets are saved.

Open Studio Chat · Open Canvas · Read the help center

Adapt the prompt to your scene

These prompts are starting points rather than generated examples or guaranteed results. Upload your own reference for image-to-video. Text-only examples attach no image. Check the current model inputs, duration, aspect ratio and quote.

Plan the five-shot film first

Try this prompt →

Then make one independent shot

Try this prompt →

Check the result before using it

  • Ask what new story information each shot contributes. Remove repeated images and camera moves that do not help the sequence.
  • Check subject color, material, count, orientation, props and light between shots. References constrain generation; they do not guarantee continuity.
  • Watch the complete edit for bad action matches, flicker, deformation, subtitle mistakes, abrupt sound changes and unwanted frames at the ends.
  • Retain the model, settings, prompt versions, attempts, actual credit records and manual fixes. Do not invent unmeasured production times or costs.

What is AI filmmaking?

Generating a video answers what one shot might look like. Filmmaking also decides what the audience sees first, why the next shot follows, when a sound enters and what feeling remains at the end. AI can help plan and make assets, while the filmmaker still selects, revises and arranges them. A finished short film is a sequence of decisions rather than a single prompt.

1. Write a story that can be shown

The exercise is simple: a paper star in an empty room moves from night toward morning light. One subject, one main location and one change make it possible to inspect the film without hiding continuity problems behind spectacle. Decide the ending and atmosphere before expanding the shot count.

  • Story goal: show a change from quiet darkness to warmth without relying on explanatory subtitles.
  • Subject rule: one pale-blue folded paper star with visible creases. Keep the material and color description consistent.
  • Visual rule: the same room and deep-blue desk, with one chosen aspect ratio. Select an output format actually supported by the current model.

2. Give five shots different jobs

This is an editable shot list, not a film that has already been generated. Suggested screen time is an editing decision; a generation model may require a different clip duration that you trim later.

  • Shot 1, establish: a wide view of the room and window, with a locked camera so the audience understands the desk and light source.
  • Shot 2, introduce: a close view of the paper star. Side light reveals folds against a simple background before movement begins.
  • Shot 3, change: a side view as the breeze slides the star slightly. Use one principal action and preserve screen direction.
  • Shot 4, connect: an insert of the window edge or moving light gives the edit room to bridge the action. A separate detail shot can be more useful than forcing one continuous generated motion.
  • Shot 5, resolve: return to the star resting in warmer light. Reduce motion and hold the ending long enough for the change to register.

3. Choose text exploration or approved image references

Text-to-video can explore an unsettled setting or atmosphere. Once a composition or subject design matters, prepare and inspect a reference image before choosing a video model that accepts it. Keep character, composition and style references distinct so you can identify what caused a change.

  • Approve one key image before expanding into additional shots.
  • First-frame, last-frame, multi-image and character reference are different capabilities. Check the selected model rather than assuming that all reference types are interchangeable.
  • Reuse a clear subject description and approved assets, then inspect every new output. A reference image is not an identity guarantee.

4. Generate one shot before scaling the workflow

Choose a simple shot with a readable subject and restrained action. Check the current inputs, parameters and credit quote, then submit one attempt. Review the entire result before making the next shot. The tutorial entries load drafts; they do not execute a sequence or spend generation credits automatically.

  • Keep a shot record: identifier, input reference, prompt, model and settings, raw output and the reason for accepting or rejecting it.
  • Change the most visible problem first. Avoid changing the model, subject, light and movement at the same time.
  • Use the conversation workspace for planning and individual tasks, and Canvas to organize references and results. Editing, voice and music remain separate work unless the current product explicitly supports the needed step.

5. Shape the film with editing and sound

Bring accepted clips into your editing tool and make a silent rough cut first. Add authorized ambience, music or dialogue after the sequence reads clearly. Cut around actions and changes in attention; you do not have to play each generated clip from its first frame to its last.

  • Arrange the original framing before cropping. A crop can remove the subject or change the spatial relationship between shots.
  • Keep raw video, music, ambience, dialogue and subtitle tracks separate so the edit stays revisable.
  • Watch the final export for abrupt level changes, missing sounds, subtitle errors and unwanted black frames. Check music and sound-effect usage rights independently.

6. Review continuity, costs and saved assets

Separate the planned number of shots from the number of attempts you are willing to make. Use live quotes for decisions and actual credit records for accounting. Unmeasured costs and completion times should stay unknown. A small test helps you find unsuitable inputs before expanding the sequence.

  • Preserve the story, shot list, references and prompt versions, plus raw footage and the final edit.
  • Confirm the save state and reopen the project from the list when needed.
  • After finishing, record which shots started from text, which depended on references and which needed manual fixes. That becomes the practical starting point for the next film.

If the first result needs work

If clips feel like different films, return to the shot plan and reference images. Align the subject description, lighting and framing, then repair one clip at a time. For a difficult transition, try another edit point or a detail insert before regenerating every shot.

Keep learning