Back to All Events
AI Individual
AI Prompting
Based on a theme given on-spot, contestants use natural language prompting to generate an image that matches the style, composition, and subject as closely as possible. Open to Group B to E (Individual).
Rules & Guidelines
Task Description
Based on an on-the-spot secret theme, contestants must craft text prompts to generate an image matching the style, composition, and subject of the theme as closely as possible.
Grade Eligibility
- •Group B to Group E (Grade 5 to University 4th Year)
- •Individual participation only
Participant Requirements
- •Own Device (BYOD): Any hardware from: laptop, tablet, smartphone, or mini PC.
- •Internet Connection: Must bring your own mobile data/hotspot connection.
- •Pre-Login: Must be logged into preferred AI image generation platform (e.g. Midjourney, DALL-E, Stable Diffusion, Firefly, Bing Image Creator) prior to the event start.
Rules & Regulations
- 1New Session: Must start a fresh chat or terminal of the AI model specifically for the competition.
- 2Safety Compliance: Must not violate the host platform’s major safety guidelines (NSFW, extreme violence, etc.).
- 3Reproducibility Check: The prompt history must match the generated image. Prompts will be test-run by judges if any suspicion arises to validate prompt authenticity.
- 4Text-Only Rule: The prompt history must NOT include any uploaded image or file other than text typed by the participant. Each instruction MUST be typed by the participant completely (autocorrect and suggestions are allowed; image-to-image cloning is prohibited).
- 5Submission: Completed output and prompt history must be submitted via the official Google Form link provided at competition time.
Event FAQs
Which AI platforms are permitted?
Any major text-to-image platform including Midjourney, DALL-E 3, Stable Diffusion, Adobe Firefly, or Microsoft Copilot / Designer.
Can I upload reference pictures into the AI?
No. Image-to-image inputs or uploaded files are strictly forbidden. All generation must be 100% prompt-driven from typed text.