Best prompt structure I've found for maintaining face consistency across generations
Hi,
After a lot of testing with different approaches in both my SFW and NSFW prompts, I found a structure that has worked incredibly well for maintaining character consistency and preserving the face across generations.
The key is organizing the prompt into clearly defined sections:
Subject: Describe the woman's exact body position, posture, and what she is doing. Minimum 2 sentences.
People: Describe the woman's clothing in detail, including fabrics, colors, and accessories. Minimum 2 sentences.
Environment: Describe the setting, background elements, and surroundings. Minimum 2 sentences.
Lighting: Describe the light sources, shadows, and atmosphere. Minimum 2 sentences.
Camera/Composition: Describe the camera angle, framing, and shot type. Minimum 1 sentence.
Color/Texture: Describe the dominant colors, textures, and materials visible in the scene. Minimum 2 sentences.
SFW Prompt Example:
photograph. the woman from image1. Keep the same face as image1. slender curvy body, hourglass figure.
Subject: woman standing confidently facing camera, one hand on hip, slight body lean showing curves.
People: wear red micro bikini two-piece triangle top with high-cut bottoms, gold hoop earrings, minimal jewelry.
Environment: outdoor poolside setting, tropical resort background with palm trees, blue pool water visible, white lounge chairs in background.
Lighting: natural daylight golden hour, soft warm sunlight from side creating gentle shadows, bright outdoor atmosphere.
Camera/Composition: medium shot full body, centered composition.
Color/Texture: warm tones golden and coral sunset colors, smooth skin texture, shiny bikini fabric reflecting light.
This separated structure by categories gives the model much more clarity about which elements to keep consistent and which can vary. The results have been notably better than when I had everything mixed in a single text block.
Hope this helps you as much as it helped me!
Breaking prompts into clear sections like Subject, People, Environment, Lighting, Camera, and Color really helps maintain consistency and control. I tried a similar approach recently and noticed much better results compared to mixing everything in a single block.
I really like this breakdown method! Separating prompts into sections like Subject, People, and Environment makes it so much easier to maintain character consistency and control over details.
This structure is very helpful. I’ve noticed that when I keep prompts organized by lighting, camera angle, and textures, the outputs are much closer to what I imagined.
Clear categories for prompts definitely improve results. Mixing everything together often leads to inconsistent character features, so this approach makes a lot of sense.
I’ve started using a similar sectioned approach, and it’s amazing how much more predictable and consistent the generated images are compared to unstructured prompts.
Breaking down prompts by elements like Subject, People, Environment, and Color really makes a difference. Outputs are cleaner, and it’s easier to tweak specific parts without affecting the whole scene.