Midjourney troubleshooting: inconsistent character features
Designer sees drifting facial details between generations despite using same seed and wants targeted prompt strategies or pipeline fixes to lock attributes.
Best tools for this use case
Based on the workflow in this discussion, these tools are useful starting points to review.
Midjourney
Premium image model with standout visual quality and strong artistic range.
Leonardo AI
Flexible image generation platform with strong controls and good creator value.
DALL·E
Accessible image generation option especially useful inside broader assistant workflows.
Answers
Approved replies, operator insight, and tactical follow-up from the community.
Practical fixes to lock facial features in Midjourney:
- Use the same model version and --seed plus --chaos 0.
- Give a high-quality reference image and set image weight (e.g., imgURL::1.0).
- Use multi-prompts with weights (face::2, eyes::1.5) and strict descriptors (shape, expression, hairstyle).
- For consistent edits, run img2img with low strength (0.3–0.5) so features stay intact.
- If absolute consistency is required, generate a base face, upscale it, and composite in post.
CTA: See Midjourney vs Leonardo AI for alternatives.