Reusable Seeds, Fixed Palettes, and Visual Modularity in DALL·E

ChatGPT
1

Author: Luiz Alves
Plan: ChatGPT Plus

:writing_hand: Context

I’m currently developing a fully illustrated children’s book using DALL·E and ChatGPT. During this process, I encountered several technical challenges related to visual consistency across scenes, particularly when reusing characters, environments, and objects.

:warning: Issues Identified

  • Non-persistent seeds: When opening a new chat or session, the visual seed from a previously saved image is lost, making it impossible to reproduce or reuse exact characters or objects.

  • Lack of reusable scene layouts: There’s no way to fix the geometry and object placement in an environment (e.g., bedroom, park) for use across multiple illustrations.

  • Prompt history contamination: Ongoing edits and generations within a single chat affect the consistency of future image outputs, even with clean prompts.

  • Limited modularity: It’s currently not possible to isolate and reuse visual elements (characters, props, backgrounds) as modular components.

  • Color palette inconsistency: Although I use Pantone references in prompts, DALL·E does not follow strict color systems (Pantone, RGB, CMYK), which limits visual coherence across scenes.

:white_check_mark: Feature Suggestions

  1. Nameable and reusable seeds for characters, objects, and backgrounds, accessible through the user’s gallery.

  2. Fixed scene seeds, maintaining layout, geometry, and spatial relationships across multiple images.

  3. Modular gallery of visual elements, allowing reuse and combination of characters, furniture, props, etc., based on previously generated assets.

  4. Scene composition tool, with a visual interface or layered structure for combining reusable blocks:
    fixed background + fixed character + action/pose variation

  5. Support for real-world color palettes such as Pantone, RGB, or CMYK for consistent color application.

:chart_increasing: Expected Positive Impact

  • Enables professional creative workflows for illustrators, educators, writers, and designers.
  • Reduces rework and improves production speed.
  • Expands DALL·E’s usability in serious visual storytelling and editorial-grade publishing.

:speech_balloon: Personal Note

Even though this is my first children’s book, I’ve manually developed ways to work around these limitations — such as creating a new chat for each scene and regenerating all characters using the same seed.
It works, but it’s time-consuming and not scalable.

I believe the features proposed here would greatly benefit all visual creators using DALL·E to tell consistent and engaging stories.

:camera: Optional: Visual Examples

Consider adding 1–2 comparison images (e.g., how the same character changes across generations) to reinforce the value of the suggestions above.

Thanks in advance for your consideration.
I’d be happy to collaborate with the product team if needed.

— Luiz Alves

2

This topic was automatically closed after 22 hours. New replies are no longer allowed.