DALL-E 3 prompt engineering: Getting consistent brand imagery

Mia Johnson
Mia JohnsonAug 30, 2024

I'm using DALL-E 3 to generate marketing images and struggling to maintain consistent brand style across generations. Each image looks completely different even with similar prompts.

Things I've tried:

  • Including detailed style descriptions ("flat illustration, minimal, pastel colors")
  • Referencing specific art styles ("in the style of corporate Memphis")
  • Using seed parameter (doesn't seem to work for consistency)
  • My prompt template:

    Create a [subject] illustration in flat, minimal corporate style with soft pastel 
    colors (light blue #E3F2FD, mint green #E8F5E9, warm gray #F5F5F5). Clean lines, 
    no text, 16:9 aspect ratio, professional marketing style.
    

    Results vary wildly. Anyone found reliable techniques for visual consistency?

    4.5k views22 replies56 likes
    1 Reply
    Yuki Tanaka

    For speaker diarization accuracy, pre-processing the audio to enhance voice separation helps a lot. I use a simple bandpass filter + noise reduction before running pyannote.

    Log in to reply to this topic.