Awenfold field manualAI video
How to Turn a Novel Scene into AI Video Prompts and a Storyboard
Contents
- 1. Choose one visual turn, not a whole chapter
- 2. Decide the delivery before writing prompts
- 3. Lock continuity references
- 4. Build a timed storyboard
- 5. Write the start frame before the motion
- 6. Generate one shot at a time
- 7. Assemble, add sound and review the sequence
- What the free tool gives you
Section 1 of 8
To turn a novel scene into AI video, do not ask one prompt to perform the whole adaptation. First choose a short visual turn, then separate the work into continuity references, timed shots, start frames, motion prompts, sound and editing. Each generated clip should contain one readable action.
Awenfold's free novel-to-AI-video prompt generator creates that editable production pack without signup. It does not render the final video; you take the prompts and references into your chosen video generator.
1. Choose one visual turn, not a whole chapter#
Select a passage with a visible change: someone arrives, a secret is revealed through an object, a door opens onto an unexpected place. Keep only the prose needed to understand that turn. Long internal history and several location changes may work in a novel, but they compete for time in a short clip.
Write the scene's miniature arc in one sentence: starting image → change → final image. For example: “Mara waits alone → her missing brother steps off the train unchanged → he warns her not to let him board again.”
2. Decide the delivery before writing prompts#
Choose the total duration and aspect ratio first. A 15-second 16:9 scene and a 60-second 9:16 sequence need different shot counts and compositions. The tool offers 15, 30 and 60 seconds, with landscape or vertical framing.
Also choose the target platform. The underlying story and shot order should remain stable; platform-specific controls affect how you phrase camera movement, start frames and sound. Check the current interface of the platform you use rather than assuming every control is available forever.
3. Lock continuity references#
Before motion, describe the recurring elements that viewers must recognize:
- character face, apparent age, hair and silhouette;
- wardrobe and a distinctive prop or color;
- location geography and screen direction;
- time of day, light, palette and image texture.
Create and approve reference stills for these elements. Reuse them when the platform supports image references. A red scarf, for example, can help connect three shots, but it cannot compensate for changing the character's face, coat and platform layout at the same time.
4. Build a timed storyboard#
Give every shot a start time, end time and story beat. A beat says what changes in the story; it is not yet a visual prompt. Keep one continuous action per shot and leave enough time for the viewer to read it.
For the train example, a 15-second storyboard could be:
- 00:00–00:05: Mara waits as the last train enters.
- 00:05–00:10: the doors open and Elias steps down, unchanged.
- 00:10–00:15: he catches her wrist and gives the warning.
If a shot requires entering, turning, finding an object, speaking and leaving in five seconds, split it or remove an action.
5. Write the start frame before the motion#
Treat each shot first as a still image. Specify subject, composition, location, light, lens feel and the continuity details that matter. Describe what is visible now, not a sequence of future actions.
Copy this model-neutral template:
START FRAME
[shot size and composition] of [subject and appearance] in [location].
[position, eyeline and important prop]. [time, light, palette and texture].
Preserve [continuity anchors]. [aspect ratio and visual style].
MOTION
Over [duration], [subject] performs [one continuous action].
The camera [one clear movement or remains locked].
Preserve [identity, wardrobe, geography and screen direction].
Avoid [unwanted cuts, added people, object changes or extra gestures].
SOUND AND EDIT
[ambience or dialogue]. Start on [cue]; cut or transition on [cue].
6. Generate one shot at a time#
Create the accepted first frame, then animate it when identity and composition matter. Text-to-video can still be useful for exploratory establishing shots. Generate each shot separately so a failure does not force you to rebuild the entire sequence.
Compare takes for story clarity, continuity and usable beginning and ending frames—not only visual spectacle. Save the selected take and note which reference image and prompt produced it.
7. Assemble, add sound and review the sequence#
Place the clips on a timeline in storyboard order. Add dialogue, ambience and music after the picture works. Check that gaze direction, movement and sound carry naturally across cuts. Trim pauses before requesting a more dramatic generation; editing often solves timing more reliably.
Watch once without sound. Can you understand the change? Listen once without looking. Does the sound explain too much or contradict the image? Then review the complete sequence against the original scene's emotional turn.
What the free tool gives you#
The generator produces an editable visual bible, reference-image prompts, a timed storyboard, voiceover, start-frame prompts, platform-ready motion prompts, negative constraints, sound cues and edit notes. You can copy the result or download it as Markdown.
Start with the complete 15-second worked example, then adapt your own passage. Use text you own or have permission to adapt, and review the policies of the video platform where you generate the clips.