Image & video prompts
Ready-made templates for images and video
The patterns I use to make my avatar videos and covers. Swap the [brackets] for your own scene and you’re set.
Image
Characters, covers, mockups.
Consistent character
ImageAttach a reference image and keep the same character in every scene.
Use the attached image as the exact character reference. Keep face shape, glasses, beard, outfit and colors identical. Scene: [what the character is doing], [location]. Style: clean cartoon illustration, soft studio light. Vertical 9:16, character in the upper two thirds, empty space at the bottom for captions. No text.
Post cover
ImageLeaves room to add your headline later.
Cover image for a tech post about [topic]. Dark background with a subtle grid and one orange glow. A single bold visual metaphor for [topic] on the right side. Leave the left half empty for a headline. Aspect ratio 4:5, high contrast, no text, no logos.
App mockup
ImageShows your app screen in a realistic setting.
Realistic photo of a hand holding a modern smartphone, screen showing [describe your app screen]. Blurred cozy desk background, natural window light, shallow depth of field, 4:5 vertical. Keep the screen sharp and front-facing.
3D toy scene
ImageFor miniature scenes with a soft plastic look.
3D toy-style miniature scene, soft plastic materials, rounded shapes, tilt-shift look. Main subject: [character or object] doing [action] on a [desk / tiny city / game board]. Pastel lighting, orange and dark accents, 16:9.
Video
Talking shots and b-roll.
Selfie talk
VideoUse the character image as the first frame; add the voice later.
Animate the character from the first frame talking directly to the camera, selfie style, handheld feel. Natural head movement, blinking, small hand gestures. Background: [bookshelf / office], softly blurred. 8 seconds, vertical 9:16. No on-screen text, no music.
B-roll
VideoShort cinematic shots to cut between talking parts.
Slow dolly-in on [subject], [time of day] light. Cinematic, shallow depth of field, subtle camera shake. One continuous shot, 6 seconds, 9:16. Keep the subject centered; no people, no text.
Tips that actually help
Why one scene, one motion?
Asking for too much in one prompt breaks consistency. Give each prompt a single scene and a single motion.
How do I keep a character consistent in video?
Generate the image first and start the video from it. Reuse the same reference image for every new scene.
Can the model add text to the image?
Don’t let the model draw text; add titles and captions in your editor. It looks cleaner and stays editable.
What should I avoid?
Real people, brands and other creators’ characters. Build your own character instead.
For text work: Prompt Library →