thinkhard.SI

Prompt repository Image & video

Color keyframes for each shot of a short video, to use as starting frames

You are planning a short video and want one finished-looking still per shot, with matching characters, to approve and then animate with an image-to-video model.

Registry no.
MED-040
Collection
Image & video
Laid out for
Image model, plain language
Score
10 out of 10
Blanks to fill
12

Image & videoImage, plain

The prompt

Score 10 of 10, Excellent

Use it when You are planning a short video and want one finished-looking still per shot, with matching characters, to approve and then animate with an image-to-video model.

  • {{shot_number}}
  • {{video}}
  • {{use_case}}
  • {{approver}}
  • {{shot_description}}
  • {{continuity}}
  • {{shot_type}}
  • {{camera_angle}}
  • {{lens}}
  • {{look}}
  • {{light_direction}}
  • {{aspect_ratio}}
Full prompt and why it works
Prompt
A single keyframe image for shot {{shot_number}} of {{video}}, for {{use_case}}. It will be approved by {{approver}} and then used as the first frame of the shot in an image-to-video model, so it must look like a frame of finished footage and be easy to animate.

The shot:
"""
{{shot_description}}
"""

Characters and setting, the same in every keyframe of this video: {{continuity}}. If I attach an earlier approved keyframe, it is the reference: match its characters' faces, hair, clothing and proportions, and its setting, light, color grade and lens, so that the shots cut together.

Camera: {{shot_type}}, at {{camera_angle}}, with a {{lens}} lens look and the depth of field that lens would give.

Light and style: {{look}}, with light from {{light_direction}} that stays consistent across the shots of the same scene.

Composition for animation: frame the moment just before the action starts, with the subject fully inside the frame and space in the direction they will move. Keep hands and faces clear of each other and of the frame edges, with no motion blur, so the video model has clean shapes to animate.

Format: aspect ratio {{aspect_ratio}}, matching the video, at the highest resolution available.

A good keyframe is one a viewer would mistake for a paused frame of the real video. Leave out text, captions, subtitles, logos, watermarks and recognizable real people, and keep any screens or signs free of readable words.

If the shot description is unclear or leaves something out, such as where a character stands, choose the simplest arrangement that fits the scene and tell me what you chose, rather than adding new characters or events.

Variables

{{shot_number}}
The shot's number in your shot list, for example 3. Make one keyframe per shot, in order.
{{video}}
What the video is, for example "a 30-second ad for our bakery".
{{use_case}}
Where the video will be shown, for example "our Instagram account".
{{approver}}
Who signs off, for example "the client" or "my manager".
{{shot_description}}
What happens in this shot, for example "Medium shot: the baker slides a tray of loaves out of the oven and smiles at the bread."
{{continuity}}
Fixed traits of each character and the setting, for example "a baker in her fifties with short red hair and a white apron, in a small bakery with blue tiles and wooden shelves". Describe fictional people.
{{shot_type}}
The framing, for example "medium shot" or "close-up".
{{camera_angle}}
The angle, for example "eye level" or "slightly from above".
{{lens}}
The lens look, for example "35mm" or "85mm".
{{look}}
The style, for example "warm, natural and photorealistic, with gentle film grain".
{{light_direction}}
Where the light comes from, for example "the shop window on the left".
{{aspect_ratio}}
16:9 for most video, 9:16 for vertical social video.

Why it works

  • An example. Each new keyframe is made with the last approved one attached, which keeps faces, clothing and light the same from shot to shot, the hardest thing to hold across a video.
  • Format. The frame shows the moment just before the action, with space to move into and no motion blur, which gives an image-to-video model clean shapes to animate.
  • If unsure. Gaps in the shot description get the simplest arrangement and a note, instead of new characters or events the script never had.

How to use it

  1. Press Fill in blanks and type your details into each blank. What you type stays in this browser.
  2. Copy the filled prompt into the AI tool you already use, such as ChatGPT, Claude or Gemini, or open it in the Grader or the Prompt Tester here.
  3. Before you rely on it, try it on three or four real examples. The Prompt Tester runs them all at once and shows which ones go wrong.
Use it from code

Get this prompt as JSON from the free REST API, with your own text in the blanks, or let an AI assistant fetch it through the MCP server. No key.

curl -s https://thinkhard.si/api/v1/prompts/video-keyframes-for-image-to-video

With the MCP server added to Claude, ChatGPT, Cursor or VS Code, call get_prompt with {"prompt": "video-keyframes-for-image-to-video"}, or just ask for it by name.