Dr. Samia shares an AI prompt with reference photo: adventure in the jungle with Donkey Kong

On July 2, Dr. Samia (@oye_samia) published on X an image prompt that starts from a facial reference photo —marked as top priority— to insert the protagonist in an adventure scene: hanging from a vine next to Donkey Kong as they cross a tropical jungle full of ancient ruins, waterfalls and giant banana trees, with volumetric lighting passing through the canopy and aesthetics of Nintendo promotional art.

AI Prompt · How to do it

Second result of the same prompt maintaining the features of the reference photo. @oye_samia

Steps

1. Open an image model that supports reference photo (editing with input image).
2. Upload a clear face photo as your highest priority facial reference.
3. Paste the entire prompt; ask for 1080×1350 vertical format if it does not respect it.
4. The thread includes two variants of the same scene.

Prompt to copy

Use the provided female reference photo as the highest-priority facial reference. Vertical 1080x1350. A stunning young woman with exact facial features preserved from the reference and Donkey Kong swing together through an enormous tropical jungle filled with ancient ruins, waterfalls, and giant banana trees. The woman hangs gracefully from a vine beside Donkey Kong, laughing joyfully and naturally with an expression of pure fun and adventure. Sunlight penetrates the dense jungle canopy creating beautiful volumetric lighting. Wooden bridges cross deep canyons. Tropical birds fly overhead. Rich cinematic colors, lush vegetation, highly detailed jungle environment, premium action composition, official Nintendo promotional artwork quality, Unreal Engine 5 realism.

Text in English according to the original publication. Replace the reference with your own face photo.

What makes this prompt special

The key is in the first sentence: «use the provided female reference photo as the highest-priority facial reference». By declaring the reference photo as top priority, the model attempts to preserve the exact features of the real person while the rest of the prompt reconstructs the environment. Without this instruction, the models usually "reinterpret" the face and lose the resemblance when integrating it into a scene so loaded with elements.

The technical closure —“official Nintendo promotional artwork quality, Unreal Engine 5 realism”— pushes the result towards a cinematic rendering finish rather than a photo: saturated colors, leafy vegetation, volumetric lighting and action composition. It is the same resource that many viral prompts use to give the appearance of video game key art.

How to adapt it

You can change the accompanying character (another icon, an animal, a pet) and the biome (jungle, desert, floating city) maintaining the structure: priority facial reference + scene + style + ratio. Keep the mention of the 1080×1350 format for vertical framing of networks and adjust the expression (laughing joyfully) according to the emotion you are looking for. Keep Nintendo's character rights in mind if you are going to disseminate the result for commercial purposes.

Before publishing: Faces generated from real people require consent, and images with protected characters are for personal use only without permission. Review the policies of the model you use.