- Some strategies to try to get better results with two characters:
- Generate each character separately first, then composite them together manually with image editing software. This gives more control.
- Break the prompt into very small, detailed sections describing each character individually – name, gender, physical attributes, clothing, accessories, emotions/actions, etc. Be as specific as possible.
- Add prompts like “two characters”, “a pair”, “together” to indicate you want two. Also add negatives like “not one character”, “not solo”.
- Experiment with image sizes – moving further away from 512px can reduce duplicate characters.
- Use AI extensions like ControlNet or Latent Diffusion that are designed to give more control over image generation.
- When creating prompts, focus on writing rich, unique details about each character and scene to guide the AI, rather than just general descriptions. Prompts should tell a story.
- Test out prompts and keep notes on settings/prompts that work to consistently generate your desired characters. Getting quality outputs takes experimentation.
In summary, creating two distinct characters in a single AI-generated image requires very careful, detailed prompt engineering and often additional tools to control image synthesis. Let me know if you have any other questions!