The best AI prompts for film and video, 20 to copy (2026)

A good cinematic prompt starts with a directing decision: what happens in the shot, where the camera looks from, which lens and which light. This library gives you 20 film prompts to copy into ChatGPT, Gemini and Copilot, and for Midjourney, Veo and Kling. To learn the method in depth, see the AI innovation program at Founderz.

Add Founderz as a preferred source

What you’ll take from this

These are the five blocks you will find, built to take you from a loose idea to a shot you can review and edit.

• 20 film prompts ready to copy: loglines, scenes, consistent characters, storyboards, still images and video clips.
• Prompts to ask ChatGPT, Gemini and Copilot for the logline, the scene breakdown and the shot list.
• Templates for image and video generators such as Midjourney, Veo or Kling, with shot type, lens, light and camera movement.
• A seven-field template for writing your own prompts, and an example of a weak prompt made strong.
• A workflow from logline to storyboard to clip, with the checks a person must do.

ChatGPT prompts for film and video

ChatGPT is good for writing, organizing and correcting: the logline, the scene breakdown, the character sheet and the final prompt. It also accepts images you attach as a reference. What it can generate in image or video depends on your plan and version, so check your account. What is in brackets is an example: swap it for your project.

From an idea to a logline with protagonist, want and obstacle

Act as a short film script doctor. I give you an idea and I want you to turn it into a logline.
Idea: [a detective who discovers the case she is investigating is her own].
Genre and tone: [sober thriller].
Write three loglines of one or two sentences. Each must name the protagonist, what they want, the obstacle and what is at stake. No clichés and no titles of existing films.
Then say which one is easiest to put on screen and why.

Scene breakdown with the visible action of each scene

Act as a pre-production director. Start from this logline: [approved logline].
Make a breakdown of [8] scenes in a table with these columns: number, place and time of day, visible action (what a camera can record), function in the story and the emotion the viewer should feel.
Avoid actions that only work with dialogue. One sentence per cell at most.
End with the two scenes that would work best as a poster image.

Consistent character sheet to repeat in every prompt

Act as a character designer for film. Create a visual sheet for [detective Marta Ríos] that I can paste at the start of every image or video prompt.
Include approximate age, visible features, hair, wardrobe, objects she carries, proportions and color palette, in one sentence of about 50 words.
Use concrete, observable words, no subjective adjectives such as "attractive" and no resemblance to real people.
Also give me a 15-word version for when the prompt is short.

Turn a scene into an image or video prompt

Act as a director of photography. Turn this scene into a prompt for an [image / video] generator.
Scene: [a detective waits alone in an empty police station before dawn].
Write a single paragraph in this order: purpose of the shot, action and subtext, shot type and camera height, lens and depth of field, light and palette, one camera movement if video, format [2.39:1] and constraints.
No legible text or logos. Make it ready to copy and add a shorter version below.

Gemini prompts for film and video

Gemini works with images and Google documents, so it fits when you start from your own material: a script in Google Docs, location photos or a moodboard. Depending on your plan and country, it can also generate clips with Google video models such as Veo. If not, copy the prompt to another tool.

Script in Google Docs to shot list

I am sharing the script of a [five-minute] short film from Google Docs. Read it and give me a table with one shot per scene: what we see, shot type, camera movement and approximate duration.
Flag the passages that only work with dialogue and cannot be shown with an image.
Then write the video prompt for shot number [1], with a single action and the elements that must stay the same.

Visual moodboard from location photos

I am attaching [four] photos of possible locations for [a night scene at a harbor]. Analyze each one: type of light, palette, texture and what dramatic function it could serve.
Choose the best one for [a scene of contained tension] and write an image prompt that respects that light and palette.
Say which elements of the photos I should not reproduce if recognizable people or visible brands appear.

Video prompt with camera, light and sound

Write a video prompt to paste into a generator, with this structure: scene, action, camera, light, style, sound and duration.
Idea: [a scientist enters an abandoned greenhouse and stops in front of a plant that gives off light].
Slow lateral tracking shot, 35mm lens, blue dawn light with light fog, [eight seconds], 16:9.
Sound in a separate sentence in case the tool supports it: [soft ambience, no music or dialogue].
No cuts or generated text. If you can generate the clip with my version, do it.

Copilot prompts for film and video

Copilot works inside Word, PowerPoint and Teams with your organization’s files. It fits the preparation of a brand or corporate piece: brief, script, storyboard and meeting notes. The clip is generated afterward in an image or video tool.

Storyboard table from a Word brief

Use the Word document [Runner spot brief] as the only source. Create a storyboard in a table with one row per shot: number, what we see, shot type, camera movement, duration and continuity notes (wardrobe, product, time of day).
[6] shots at most for a vertical 9:16 piece of [30 seconds].
If the document lacks information, add it as a list of questions at the end instead of inventing it.

Video script from a PowerPoint presentation

Use the presentation [Product launch] to write a [45-second] script for [the intranet].
One voiceover sentence per key slide, [5] at most, with the shot that would go with it (shot type and action) and its duration.
Tone: [friendly and professional]. Do not add figures or claims that are not in the presentation.

Clips from a Teams meeting for an internal piece

Using the transcript of the Teams meeting [Campaign review, October 14], suggest [3] clips of [20 seconds] to summarize the decisions for [the sales team].
For each clip give the message, the shots to record or generate and one voiceover sentence.
Flag which passages contain confidential data or client names and should not appear.

More film prompts for Midjourney, Veo, Kling and other generators

These templates go straight to the generator. Each separates purpose, scene, shot, lens, light and limits, and what is in brackets is what you swap for your project. For a sequence, write one prompt per shot and decide the order in the edit.

Still image: the frame that sets the look

A still image solves a single frame. Use it to fix composition, light and palette before animating anything.

Final image for a restrained thriller

Medium shot of a detective alone in the empty waiting room of a police station before dawn. She is seated, holding a closed folder and looking toward a door just outside the frame.
Eye-level camera, 50mm lens, medium depth of field to keep the space visible. Dim fluorescent light mixed with cold blue from the street, muted tones, sober thriller atmosphere.
Aspect ratio 2.39:1. No legible text, no logos, no graphic violence.

Portrait in golden hour light

Close-up of [a veteran fisherman] mending a net on the dock, eyes down and focused.
85mm lens, shallow depth of field, soft bokeh of the harbor behind him. Side golden hour light, slight backlight in the hair, desaturated warm palette, fine grain.
Format 4:5. No text, no logos and no resemblance to any real person.

Night scene with neon light

Low-angle wide shot of [a wet city street] at night. A figure in a dark coat crosses under a red neon sign with no legible text.
24mm lens, deep depth of field, reflections on the asphalt, magenta and blue neon light, deep shadows, modern noir look.
Format 2.39:1. No brands or generated letters.

Video: one shot, one action, one movement

In video, describe one visible action, a single camera movement and a short duration. If visual identity matters, start from the approved image.

Reveal a place with a tracking shot

Establishing shot of an abandoned greenhouse at dawn. A scientist enters through the main door, stops when she sees a plant giving off a soft light and decides not to approach yet.
Slow lateral tracking shot from left to right, 35mm lens, early blue light with light fog, muted greens and warm reflections from the plant.
Duration [eight seconds], 16:9. Keep the gray coat, the greenhouse and the direction of the light consistent between frames. No cuts, no generated text, believable physics.

Dolly in to a close-up of a decision

Medium shot of [a man] sitting in front of a sealed letter in a kitchen at night. Single action: [he looks at it, takes a deep breath and tears it without opening it].
Slow dolly in to a close-up, warm practical light from a table lamp, background in shadow, restrained pace, [six seconds], 16:9.
Keep face, clothes and objects identical throughout the clip. No legible text.

Animate the approved frame (image to video)

Use the reference image as the visual source of truth and keep exactly [character, wardrobe, composition, palette and background].
Add only a very slow camera movement, [a lateral pan to the right], and believable physical movement in [the smoke and the coat fabric].
Duration [four seconds], calm pace. Do not introduce people, objects or changes of expression.

Point-of-view shot with a handheld camera

Point-of-view shot of [a runner entering a tunnel at dawn]. Handheld camera with slight instability, 24mm lens, light that goes from warm to cold as she enters.
Single action: [she speeds up toward the exit]. [Five seconds], 9:16, breathing and footsteps as the only sound if the tool supports audio.
No cuts or on-screen text.

Product shot for a brand spot

Close-up of [a reusable water bottle] on [a wet rock] at dawn. Single action: [a drop slides down the surface].
Slow dolly in, macro lens, warm side golden hour light, blurred background, natural palette.
[Six seconds], 9:16, free space in the top third for text added in editing. Keep product and label identical. No generated logos.

Storyboard and continuity between shots

In a storyboard, name each shot and protect what connects the sequence: face, wardrobe, objects, time of day and direction of movement.

Six-shot storyboard for a vertical spot

Create a storyboard of six still images for a vertical 9:16 piece. A runner finishes training at dawn and drinks water before heading home.
Keep the same face, navy tracksuit, clear bottle and golden hour in every frame.
Shots: 1) wide arrival shot, 2) detail shot of the shoes, 3) medium shot catching her breath, 4) insert of the bottle, 5) close-up of relief, 6) wide exit shot.
Warm side light, natural skin, clean style, free space at the top for copy added in editing.

Shot and reverse shot without crossing the line

Generate two still images of a dialogue between [a lawyer] and [her client] in [a meeting room]. Shot 1: medium shot of her, looking right. Shot 2: reverse shot of him, looking left.
Same camera axis, same 50mm lens, same warm side light and same blurred background in both shots.
Keep wardrobe and objects on the table identical. No text or logos.

How I built these prompts: a seven-field template

A film prompt works like a short cinematography sheet. Each field answers a different question, so when the result fails you know what to adjust without rewriting everything. Not every field is required: drop the ones that do not change the result.

Five tips for writing your own prompts

1. Write the action before the look. Instead of “a sad woman”, write “she folds a letter without opening it and puts it in her pocket”. The emotion becomes visible through the action.
2. Choose the shot type by the information it gives. A wide shot sets the world, a medium shot holds an interaction, a close-up reveals a decision and a detail shot turns an object into story information.
3. Use the lens with intent. A 35mm gives context, a 50mm brings you closer to the subject and an 85mm compresses the background. Ask for selective focus or deep focus, not both.
4. Define light as an observable condition: where it comes from, what quality it has and what effect you want. “Cold dawn light through a window” gives more direction than “dramatic and premium”.
5. Ask for a single camera movement per clip and change one variable per attempt. Keep the prompt that worked as a template.

Example: from a weak prompt to a strong one

From logline to storyboard to clip: a workflow

The most controllable way to make film with AI is to work in phases. Each approved output becomes the reference for the next one, which cuts contradictions and unusable material.

1. Logline: sum up protagonist, want, obstacle and tone in one or two sentences.
2. Scene breakdown: identify the visible action each moment needs.
3. Visual bible: fix character, wardrobe, location, palette, optics and your own references.
4. Storyboard: create still images to test order, framing and continuity.
5. Video shot: animate only the approved frames, with one camera movement per clip.
6. Edit and review: decide pace, sound, music and accessibility outside the generator.

If you work on the script before the shot, the guide on AI prompts for creative storytelling covers the earlier step.

How the prompt changes in ChatGPT, Gemini, Copilot, Claude and video generators

• ChatGPT: good for writing the logline, the character sheet and the prompt, and it accepts reference images. Ask for one variable per change.
• Gemini: fits scripts in Google Docs and location photos. Depending on your plan, it can generate clips with Google video models.
• Copilot: works with your Word, PowerPoint and Teams files. Ask it to use only the information in the document.
• Claude: does not generate images or video. It is useful for carefully drafting a long script or treatment and for writing the prompt you then paste into a generator.
• Midjourney, Veo and Kling: each tool has its own controls for format, duration and reference. Adapt the prompt to what you have available instead of copying it unchanged.

Frequently asked questions about AI film prompts

These are the most common questions when you go from an idea to a first generated shot, with the answer first.

What is a cinematic prompt?

It is an instruction that turns script, directing and cinematography decisions into something an AI model can interpret. It defines a shot or a micro-scene with action, shot type, lens, light and format, not just a style.

How do I write a storyboard prompt?

Ask for a limited set of shots, name each one (wide, detail, close-up) and protect what connects the sequence: face, wardrobe, objects, time of day and palette. There is a six-shot prompt in the library.

What should a cinematic video prompt include?

Subject, visible action, setting, shot type, lens if it matters, light, format, duration and a single camera movement. Add continuity constraints and do not ask for several major events in one clip.

Can I make film with AI using ChatGPT?

ChatGPT helps you write the logline, the scene breakdown, the character sheet and the prompt. What it can generate in image or video depends on your plan and version; if it does not include it, paste the prompt into Midjourney, Veo or Kling.

Can Gemini generate video from a prompt?

Depending on your plan and country, Gemini can generate clips with Google video models such as Veo. Check your account. If not, use it to prepare the prompt from your scripts and photos and paste it into another generator.

Is Copilot useful for preparing a video?

Yes, for the preparation. Copilot turns a Word brief, a presentation or a Teams meeting into a script and a storyboard. The clip is generated afterward with another tool.

Can I use Claude to make film with AI?

Claude does not generate images or video. It helps you carefully draft a long script or treatment and write the final prompt, which you then paste into an image or video generator.

Do I need to know film to write prompts?

You do not need to have shot a film, but knowing the basic vocabulary of framing, lens, lighting and editing helps. The seven-field template helps you learn to justify each adjustment.

Is there a free film prompt generator?

Yes, the Founderz film prompt generator is free and structures your brief. The tools you use afterward may have their own plans and limits.

Your next step: generate, review and keep the method

Pick a scene you need to solve, copy one of the prompts, change what is in brackets and generate a version. Review one criterion at a time and keep the decisions that preserve continuity.

If you want to bring AI into script, pre-production, image, video, sound and post-production with a method, take a look at the AI innovation program at Founderz.