An open index of curated prompts for image & video generation models.
/15-real-estate Create a cinematic property tour for this luxury home [attach image]
Generate a liquid glass motion graphics promo for https://linear.app
templates: - **Liquid Glass** — translucent UI panels with reflections (the default) - **Floating Desktop** — MacBook/iPad product shots in black void - **Glass UI Ad** — close-ups on glass interactions - **Abstract Motion Identity** — timestamped sequences with geometric transformations ## Sweet Spot | Refs | Duration | Use Case | Cost (Pro 720p) | |------|----------|----------|-----------------| | 3 refs | 4s | Quick test | $1.21 | | **5 refs** | **10s** | **Production promo** | **$3.03** | | 1 ref | 4s 480p | Cheapest test | ~$0.81 | **Golden rule: 1 reference image per 2 seconds of video.** ## Key Rules 1. **Never describe logos in prompts** — pass the actual logo image as a reference
Create a 10s sci-fi mech battle video prompt
Create a 6-panel storyboard for a 15-second brand promotional video. Label each panel with a shot description. Style: cinematic, cool color tone, widescreen 16:9. Content: the journey of a product from factory to the customer's hands.
Cinematic brand advertisement, slow camera push-in, product centered in frame, warm side lighting, soft background blur, no people, 3 seconds.
controls></video></td> </tr></table> **Steps:** 1. Describe your scene to GPT Image 2 and generate a storyboard image 2. Import the storyboard into Seedance 2.0 using Image-to-Video mode 3. Export each clip and assemble in your editing software **GPT Image 2 Prompt:** ``` Create a 6-panel storyboard for a 15-second brand promotional video. Label each panel with a shot description. Style: cinematic, cool color tone, widescreen 16:9. Content: the journey of a product from factory to the customer's hands. ``` **Seedance 2.0 Prompt:** ``` Cinematic brand advertisement, slow camera push-in, product centered in frame, warm side lighting, soft background blur, no people, 3 seconds. ``` > [!NOTE] > Output storyboard images in 16:9 to avoid Seedance auto-cropping. Set frame rate to 24fps to match film standards. Keep each storyboard panel simple — the simpler the content, the more accurate the motion output. <!-- Case 2: 3×3 Grid Storyboard Method (by @servasyy_ai) --> ### Case 2: [3×3 Grid Storyboard Method](https://x.com/servasyy_ai/status/2047198012750143999) (by [@servasyy_ai](https://x.com/servasyy_ai)) A key technique discovered by the community: composing all storyboard panels into a single 3×3 grid image before importing to Seedance significantly reduces failure rate compared to importing frames one by one. <table><tr> <td align=
** ``` Create a 6-panel storyboard for a 15-second brand promotional video. Label each panel with a shot description. Style: cinematic
** ``` Cinematic brand advertisement, slow camera push-in
I need a horror prompt using VHS style, Dutch angle camera, and the Horror Face preset
I have a Soul ID character. Write 3 different scene prompts with her — office, party, rooftop
Use $ultimate-ai-media-generator-skill to generate one image: - model: nano-banana-pro - scene: text-to-image - prompt: A cute kitten dancing, 3D cartoon style, dynamic full body, clean stage background - options: aspect_ratio=16:9, resolution=1k Return task id, final status, and output image URL.
Use $ultimate-ai-media-generator-skill to generate one video: - model: seedance-2.0-pro - scene: text-to-video - prompt: Cinematic wide shot of a futuristic city at sunrise, smooth drone motion - options: duration=10, resolution=standard If seedance-2.0-pro is unavailable, fallback to sora-2. Return task id and final video URL.
{ "prompt": "The character in @image1 runs through a forest", "images_list": ["__file_1__"], "aspect_ratio": "16:9", "quality": "basic", "duration": 5 }
The character in @image1 runs through a forest", "images_list
run-story-pipeline \ --character ./character.png \ --theme "樱花树下的邂逅" \ --style anime_2d \ --scenes 4 \ --output-dir ./story_output
产品在晨光中缓慢旋转展示" --ratio 16:9 --duration 5 --output ./output.mp4 manga-text-video --prompt
或直接修改 `seedance_video.py` 中的 `ARK_API_KEY` 变量。 #### 3. 编写镜头脚本 在 `seedance_video.py` 中编辑 `SHOTS` 列表,定义每个镜头的参数:
Music Accompanying Scene 1: "Sun Wukong stands on the roof of the city, brandishing his golden cudgel, his robe fluttering in the wind. The city lights illuminate the night, creating a cinematic and epic feel. The camera slowly zooms in on the image, 4 seconds." Music Accompanying Scene 2: "Homelander descends slowly from the sky, his eyes glowing with red laser light. Dark clouds churn, superhero movie style, slow motion, 4 seconds." Music Accompanying Scene 3: "Sun Wukong attacks Homelander at high speed, swinging his golden cudgel. The air shakes, action movie effect, motion blur effect, slow motion, 4 seconds." Music Accompanying Scene 4: "Homelander fires a red laser, colliding with the golden cudgel's energy. The sky explodes, the clouds tear apart, science fiction film style, intense light effects, 4 seconds." Scene 5: "The two float in the air, facing each other, energy ripples spreading. Epic closing scene, wide angle, cinematic composition, 4 seconds."
Sun Wukong stands on the roof of the city, brandishing his golden cudgel, his robe fluttering in the wind. The city lights illuminate the night, creating a cinematic and epic feel. The camera slowly zooms in on the image, 4 seconds.
Homelander descends slowly from the sky, his eyes glowing with red laser light. Dark clouds churn, superhero movie style, slow motion, 4 seconds.
Sun Wukong attacks Homelander at high speed, swinging his golden cudgel. The air shakes, action movie effect, motion blur effect, slow motion, 4 seconds.
Homelander fires a red laser, colliding with the golden cudgel's energy. The sky explodes, the clouds tear apart, science fiction film style, intense light effects, 4 seconds.
Generate a video not an image, a cyberpunk cat walking in the rain.