How to Create AI Videos: 12 Ready-to-Use Prompts and Guide (2026)

Creating video with AI means generating a short clip, a few seconds long, with motion and sound, from a text description you write or a photo you upload. You need no camera, actors or editing software. You describe the scene, the camera move and the sound in words, and the model builds the picture and the audio together. In 2026, the most accessible way to do this from Turkey is the Gemini Omni model in the Gemini app.

According to Google Keyword Planner, the Turkish query for 'AI video creation' is searched about 22,200 times a month in Turkey. The first page of results is mostly tool sites and YouTube videos, and most pages that offer ready-made prompts skip Omni's chat editing and extension features. This guide gives you 12 copy-paste prompts and explains how each one is built around the limits in Google's official documentation.

What Is AI Video Generation and How Does It Work?

AI video models read the text, plus any image you add, and generate consecutive frames and sound at the same time. Google's Veo 3.1 documentation lists seven parts of a good video prompt: subject, action, style, camera position and motion, composition, focus and lens effects, and ambiance. The same documentation recommends putting dialogue in quotation marks and describing sound effects explicitly. Every prompt below follows this skeleton.

Google introduced Gemini Omni at Google I/O on May 19, 2026, and said it would replace Veo in the Gemini app. Omni 1.1 Flash, released in August 2026, added scene extension, transitions between a first and last frame, and 4K upscaling. ChatGPT is a different story: according to the OpenAI Help Center, the Sora web and app experience was shut down on April 26, 2026. That is why this guide uses ChatGPT to write the prompt and Gemini to make the video.

How to Create a Video with AI: 6 Steps

Making a video with Gemini takes six steps: check your plan, open the video tool, write the prompt, add photos if needed, send it and wait, then review and download the result. The steps are the same on desktop and mobile; only the menu location changes.

1. Check your plan: To make videos with a personal account you need a Google AI Plus, Pro or Ultra plan; the free plan has no video generation.
2. Open the video tool: In Gemini on the web, open the sidebar and click Create video; pick a ready-made template if you like.
3. Write the prompt: Copy one of the prompts below and fill in the square brackets with your own scene.
4. Add images or a video: Upload up to 5 images for photo-to-video, or 1 video for editing; the video's aspect ratio follows the image you upload.
5. Send it and wait: Generation can take a few minutes. You can't use the same chat in the meantime, but you can open a new one.
6. Review and download: If needed, ask for a single change in the same chat; when it's done, download the video from the Share menu or share it on YouTube.

12 Ready-to-Use AI Video Prompts (Copy and Paste)

The 12 prompts below were written for the Gemini app (Gemini Omni); the same skeleton also works in Veo 3.1 and Google Flow. Replace the fields in square brackets with your own scene and delete the brackets. Subject, action, camera, style and sound each sit on their own line, so when you don't like a result, you can see right away which line to change.

Text-to-video prompts

Create a 10-second, single-scene product promo video.
Product: [a white ceramic coffee mug].
Scene: The product sits [on a wooden kitchen counter]; morning sunlight comes in from the left, and a thin wisp of steam rises from the mug.
Camera: Starts in close-up and slowly circles halfway around the product (slow orbit); the product stays centered and in sharp focus the whole time.
Style: Realistic, commercial quality, warm tones, shallow depth of field.
Sound: Soft acoustic guitar music and the sound of coffee pouring into the mug; no dialogue.
Do not change the product's shape, color or proportions; do not add any text or logo to the frame.

Create a cinematic landscape video made of one continuous shot.
Location: [An old seaside harbor town with steep streets], at [sunset].
Action: Seagulls fly across from the right of the frame, a few people walk along the coastal road in the distance, fishing boats rock gently.
Camera: Starts high in the air (aerial shot), then slowly descends and glides forward toward the shore.
Style: Documentary cinematography, natural colors, golden hour light, 16:9 landscape.
Sound: Waves, distant seagulls and a light wind; no music and no dialogue.
Do not cut between scenes.

Create a 10-second, single-scene video of two people talking.
Characters: [a woman in her 30s wearing a gray sweater] and [an older coffee maker with a mustache and glasses]. They must not resemble any real person.
Setting: [A small, dim neighborhood coffee shop; a cezve and coffee cups on the counter].
Action and dialogue: The woman walks up to the counter and asks with a smile: “[The usual today?]” The coffee maker nods and replies: “[Of course, light on the sugar.]”
Camera: Medium shot at shoulder height, keeping both people in the same frame; the camera does not move.
Sound: Clear, easy-to-understand dialogue; cups clinking and a low murmur of conversation in the background; no music.

Create a vertical (9:16), 10-second video for Instagram Reels and YouTube Shorts.
Topic: [Three-ingredient homemade chocolate cookies].
First 2 seconds: An eye-catching moment that stops the viewer: [melted chocolate flowing over a cookie in slow motion].
Next 8 seconds: [A warm cookie is lifted off the tray and broken in half, and the chocolate inside stretches].
Camera: From above and very close, smooth and steady movements; the subject stays centered at all times.
Style: Bright, appetizing, natural daylight.
Sound: An energetic, light beat and crunching sounds; no dialogue. Do not put any text on screen.

Create a 10-second video in one continuous shot that shows the events in the time order below.
[0-3 sec] [An empty cafe in morning light; a waiter flips the sign on the door glass.]
[3-6 sec] [The camera slowly moves inside; steam rises from the espresso machine on the counter.]
[6-10 sec] [The first customer walks in and the waiter waves; the camera shows both in a wide shot and stops.]
Style: Warm, friendly, realistic; soft daylight.
Sound: Door chime, the coffee machine and light jazz; no dialogue.

Photo-to-video prompts

Photo-to-video is one of the most searched branches of this topic: the Turkish query for making a video from a photo with AI gets about 1,600 searches a month. In Gemini you can attach up to 5 photos per prompt, and the video's aspect ratio follows the image you upload. Prompt 8 was written for the first-and-last-frame transition announced with Omni 1.1; it gives the most consistent results in Google Flow or the API.

Turn the attached photo into a 10-second video; use the photo as the first frame.
Keep: The people's faces, clothes, the location and the colors must stay the same as in the photo.
Motion: [The wind gently moves their hair; the tree leaves in the background sway; the person slowly turns to the camera and smiles].
Camera: A very slow move in (slow push in); no shake.
Sound: [Birdsong and a light wind]; no dialogue and no music.
Do not add any new person or object that is not in the photo.

Create a 10-second video with a single story flow from the attached [3-5] photos.
Topic: [Our Cappadocia trip: hot air balloons, a hike in the valley, dinner].
Order: Use the photos in the order I uploaded them; move from each photo to the next with a soft, natural camera move, and do not use hard cuts.
Keep: Places and people must stay as they are in the photos; do not add people who are not in the photos.
Style: Warm, nostalgic, light film grain.
Sound: [A calm, emotional piano melody]; no dialogue.

Use the two attached photos: the first is the video's first frame, the second is its last frame.
Show the change between them in one continuous shot, in a physically plausible way: [an empty room turning into a furnished, lit room].
Camera: Move very slowly while keeping the viewpoint of the two photos; do not zoom suddenly.
Pacing: The change speeds up in the middle of the video, then slows down at the end and settles on the last frame.
Sound: [Light, modern electronic music]; no dialogue.

Prompts for editing and extending video through chat

Gemini Omni's standout difference is that it can fix a video through chat without generating it again from scratch: Google's examples include swapping a character, adjusting the lighting, stabilizing the video and replacing the background. According to Omni's developer documentation, simple prompts work best for editing, too much description leads to unwanted changes, and audio editing is not supported.

In the video you just made, change only this: [turn the lighting into a night scene with the street lamps on / replace the background with a snowy mountain landscape / stabilize the video and remove the shake].
Keep everything else the same: do not change the people, clothes, camera movement, length or sound.

Extend this video, continuing the same story.
What happens next: [The woman takes her coffee and walks to the door; when she opens it, she sees it has started raining outside].
The characters' appearance, the setting, the lighting and the camera style must stay exactly the same as in the previous part; no cut at the transition.
Sound: Keep the previous ambient sound; add the sound of rain when the door opens. Do not add new dialogue.

Writing video prompts with ChatGPT, plus a quick template

In this workflow ChatGPT is your assistant director: you describe your idea, and ChatGPT turns it into prompts split into subject, camera, lighting and sound lines that you can paste into Gemini. Google's API documentation lists English as the fully supported language, so Prompt 11 also asks for an English version of each prompt. If a prompt in another language gives inconsistent results, try the English one.

You are an experienced commercial director and AI video prompt writer.
My idea: [A short promo video for our newly opened bakery that I will post on Instagram].
Write me 3 different video prompts I can use with Gemini Omni or Veo 3.1. Each prompt must:
- Be a single scene lasting no more than 10 seconds,
- Include subject, action, setting, camera angle and movement, lighting, style and sound (music, effects, and if any, dialogue in quotation marks of at most two short sentences) on separate lines,
- State the aspect ratio: [9:16 vertical / 16:9 landscape],
- Be written in English, with a one-line summary in my language after each prompt.
Do not use real people, celebrity names, brand logos or copyrighted music.

[Subject] [action] in [setting].
Camera: [close-up / wide shot], [static / slow push in / tracking shot].
Lighting and style: [golden hour / studio lighting], [realistic / cinematic / animation].
Sound: [music genre], [ambient sound]; no dialogue. One scene, no cuts.

The 8 Parts of a Good Video Prompt

A video prompt gives consistent results when it answers the seven parts in Google's Veo documentation plus sound. The table shows what each part does and what happens if you leave it empty; all 12 prompts above are made of these lines.

The row people skip most often is sound. Gemini Omni and Veo 3.1 generate sound together with the picture on their own, so when the sound line is empty the model may add music or dialogue of its own choosing. Google's Omni documentation explicitly recommends short negative phrases such as 'no dialogue' for this.

Gemini App or Veo 3.1 API? Which Route Fits Whom

There are three official Google routes for creating AI video from Turkey: the Gemini app, Google Flow and the Gemini API. For individual users and small businesses, the Gemini app is the shortest path. Flow is built for assembling a story scene by scene with the same Omni Flash model, and it runs on credits that come with your plan: AI Plus 200, AI Pro 1,000, AI Ultra 10,000 or 25,000 credits. The table compares the app with the two API models.

With Omni, you can combine images, audio, video and text as input and create high-quality videos grounded in Gemini's real-world knowledge.

Here is the practical advice: if you are making a single clip for social media, start with the Gemini app. If you need to control the dialogue, the camera angle and the timing down to the second, or you plan to connect video generation to your own app, use Veo 3.1 through the API. In the API, generation can take up to 6 minutes at peak times, and videos are kept on the server for 2 days; any video you don't download in that window is deleted.

Camera and Sound Language in Video Prompts: An 8-Term Glossary

You can write a video prompt in any language. If you write it in a language other than English, adding the English camera and lighting terms in parentheses helps the model understand exactly which shot you mean. The rule is simple: describe what you want in your own words, then add the technical term in English in parentheses.

For dialogue, Google recommends writing the lines in quotation marks and describing sound effects separately and explicitly, for example 'a door creaks' or 'a train passes in the distance'. In a single clip, dialogue longer than two short sentences won't fit the length and makes lip sync harder.

7 Techniques for Stronger Video Prompts

• Ask for one scene: The phrase 'one continuous shot' or 'no scene cuts' stops the model from squeezing several scenes into 10 seconds; the Omni documentation explicitly recommends it.
• Write the timing: If the order of events matters, use timecodes in the form '[0-3 sec] ... [3-6 sec] ...' (Prompt 5).
• Describe sound on its own line: Write the music genre, the ambient sound and any dialogue in quotation marks; if you want no talking, say 'no dialogue'.
• Keep negatives short: One-sentence bans such as 'do not put text on screen' or 'do not add new people' work more clearly than long explanations.
• Ask for one change when editing: Change one thing per message and add the sentence 'keep everything else the same' (Prompt 9).
• Set the ratio with the photo: The aspect ratio of the image you upload sets the video's ratio; if you want a vertical video, upload a vertical photo.
• Test short first: In the API, test the prompt on a 4-second 720p clip, and when you like it, move to 8 seconds and 1080p or 4K.

All seven techniques share one idea: leave the model nothing to guess. When a video comes out badly, the usual cause is a line the prompt left empty. If you don't write the camera, the light or the sound, the model picks them itself.

6 Common Mistakes When Creating Videos with AI

1. Several scenes in one prompt: A description like 'first on the beach, then at a cafe, then at home' leads to broken transitions in 10 seconds; generate each scene separately and join them with extensions.
2. Not describing the camera: If you don't write the camera angle and movement, the model picks its own and the subject may leave the frame.
3. Forgetting the sound: Omni and Veo 3.1 generate sound on their own, so if you don't describe it, you may get music or dialogue that doesn't fit the scene.
4. Long non-English dialogue: Google's API documentation lists only English as fully supported; keep dialogue in other languages, Turkish included, to two short sentences and check the pronunciation.
5. Asking for real people and brands: A video with a celebrity name, a well-known brand logo or someone else's photo used without permission may be refused and carries legal risk.
6. Asking for everything at once when editing: Changing the lighting, background and clothes in a single message makes the model reinterpret faces and motion.

The last two mistakes are about responsibility. As realistic video gets easier to make, the question of who and what a video shows, and with whose permission, matters more; in ads and promo videos, that line directly affects how much people trust the brand.

Using It from Turkey: Pricing, Free Options, KVKK and Watermarks

As of September 2026, Gemini's subscription page for Turkey lists Google AI Plus from 199.99 TL a month, Google AI Pro from 869.99 TL and Google AI Ultra from 1,479.99 TL. Google says Plus gives 2 times and Pro 4 times the usage limits of the free plan, and it positions Ultra as the widest video access with Veo 3.1. If you're looking for a free route, Google announced that Omni is free in the Remix feature of YouTube Shorts and in the YouTube Create app; the rollout is gradual, so it doesn't show up on every account at the same time.

Under Turkey's Personal Data Protection Law (KVKK), a person's photo is personal data because it can identify them. Get explicit permission before animating someone else's photo, and keep that permission in writing for promo videos. Google says every video made in the Gemini app carries an invisible SynthID watermark that can be verified through Gemini, Chrome and Google Search.

In AI video creation, the prompt has the biggest say in the result: when you write the subject, action, camera, lighting and sound separately, Gemini Omni produces a consistent 10-second clip with sound. With the 12 prompts above you can start from text, a photo or an existing video, and use ChatGPT to turn your idea into a prompt. On your first try, start with one scene and a short sound description, then grow the story with the extend and edit prompts.