This is the exact process we run in the AI Basic course on day three. You start with a pencil sketch on paper and end with a 10-second cinematic clip you can post the same evening.
Step 1: Sketch the idea, not the pixels
On a sheet of A4, draw the three key frames of your story. Not pretty, just clear. Composition, character pose, camera angle. Everything that follows is built on this.
Step 2: Turn the sketch into a still
Photograph the sketch and feed it into ChatGPT image generation or Midjourney with this kind of prompt:
"Cinematic still based on this composition, [describe scene], shot on 35mm anamorphic, soft natural light, shallow depth of field, hyperrealistic, no text"
Iterate four or five times until the still feels right. Save the strongest one.
Step 3: Animate the still
Drop the still into Sora, Kling or Runway. The trick most beginners miss is to keep the motion prompt simple. Long prompts confuse the model.
"Slow camera push-in. Subject turns head slightly. Wind moves hair. Subtle background motion."
Generate two or three variations. Pick the best one.
Step 4: Sound and edit
Open the clip in After Effects or CapCut. Add a sound bed. Cut the dead frames. Colour-grade if you need to. Done.
The mistakes you will make
- Over-prompting. The longer your prompt, the further the model drifts from your sketch. Keep it tight.
- Skipping the sketch. If you cannot draw the idea on paper, no amount of prompting will save you.
- Generating too long. Ten seconds is enough. Twenty is rarely better.
By the end of day three in our AI Basic course every student has a finished 10-second clip in their portfolio. By the end of the course they have eight.