By Jim Shimabukuro (assisted by Gemini)
Editor
Introduction: These are instructions on how to prompt a chatbot for a photorealistic image. The source is Gemini, which I recently discovered creates very good images. However, for the illustration that accompanies this article, I wasn’t able to generate a useful image, so I turned to ChatBox. Since I got some excellent images from Gemini earlier, I’m assuming the problem was with the prompt I created. -js
Gemini: To get a highly detailed and photorealistic image from an AI chatbot, you need to think like a professional photographer and include specific, descriptive details in your prompt. Don’t just describe the subject; describe how it’s being captured.
Here’s a breakdown of the key elements to include, often called “prompt engineering”:
1. The Subject
Start with a clear, specific description of the main subject.
- Instead of: “a dog”
- Try: “a golden retriever puppy playing in a field”
- For people, be even more detailed: “a close-up portrait of a mid-30s woman with freckles and a kind smile, looking directly at the camera”
2. The Environment and Scene
Describe the setting in detail. This adds context and realism.
- Location: “in a bustling city street at night,” “on a sun-dappled forest path”
- Time of day: “at golden hour,” “at dawn,” “under a full moon”
- Weather/Atmosphere: “a light mist in the air,” “during a gentle rain,” “with a dramatic, cloudy sky”
3. Technical Photography Details
This is where you push the realism to the next level. Think about what a photographer would control.
- Camera and Lens: Specify the equipment. The AI is trained on images taken with specific cameras and lenses, and this helps it mimic their unique visual properties.
- Examples: “shot on a Canon EOS R5,” “with a 50mm f/1.4 lens,” “using a wide-angle lens”
- Lighting: Lighting is one of the most crucial elements for realism.
- Types: “cinematic lighting,” “dramatic side lighting,” “soft, natural light,” “harsh overhead light”
- Specifics: “volumetric lighting,” “rim lighting,” “sunlight filtering through a window”
- Resolution and Quality: Use keywords that tell the AI to produce a high-quality, detailed output.
- Keywords: “hyper-realistic,” “photorealistic,” “ultra-detailed,” “8K resolution,” “4K,” “HDR”
- Composition and Framing: Tell the AI how to frame the shot.
- Framing: “close-up portrait,” “wide-angle shot,” “from a low angle,” “rule of thirds composition”
- Depth of Field: This creates a realistic sense of focus.
- Keywords: “shallow depth of field,” “bokeh effect,” “blurred background,” “everything in sharp focus”
4. Realistic Imperfections and Specific Textures
Avoid making your subject look “too perfect” or synthetic. Real-world photos have imperfections.
- Keywords: “authentic,” “real-world imperfections,” “natural grain/noise,” “subtle lens distortions”
- For people: “realistic skin texture,” “sub-surface scattering,” “flyaway hair strands,” “uneven skin tone”
- For objects/environments: “water droplets on petals,” “weathered stone,” “dust particles in the air”
5. Negative Prompts
If the AI model you’re using supports it, you can tell it what to avoid. This is a powerful tool to remove unwanted artifacts or styles.
- Negative Prompt Examples: “not an illustration,” “no cartoon,” “no smooth skin,” “not blurry”
Example Prompts to Try:
Portrait Prompt:
A hyper-realistic, high-resolution close-up portrait of a woman in her late 20s. She has natural, soft facial features with realistic skin texture and a few subtle pores. Her hair is styled with a few loose strands escaping. The lighting is soft and cinematic, with natural window light casting gentle shadows. Shot on a Canon EOS R5 with an 85mm f/1.8 lens. Photorealistic, ultra-detailed, 8K, HDR.
Landscape Prompt:
A cinematic, photorealistic image of a lone hiker standing on a mountain peak at dawn. The sky is a gradient of pink, orange, and purple, with a light morning mist filling the valley below. The foreground includes detailed, weathered rock textures. Shot with a wide-angle lens. High detail, 8K resolution, realistic lighting and reflections.
Filed under: Uncategorized |
Educational Technology and Change Journal 
Leave a comment