🚀 Feature Request: Enhancements for DALL·E Image Generation

:hammer_and_wrench: Issue Description

DALL·E is an incredible tool, and we truly appreciate all the innovation that has gone into it! Major shoutout to the OpenAI team for their amazing work in pushing AI-generated imagery to new heights. :blue_heart::clap:

That said, there are a few key improvements that would make DALL·E even more user-friendly, especially for social media, branding, and professional design workflows. Here are four essential areas that could enhance its usability:


:small_blue_diamond: 1. Option to Select PNG/JPEG Format at Image Generation

Current Challenge:

  • DALL·E only generates WEBP format, requiring manual conversion.
  • Many platforms (social media, design tools, CMS) prefer PNG or JPEG for better compatibility.

Proposed Solution:
:white_check_mark: Allow users to choose PNG or JPEG before generating an image.
:white_check_mark: Provide a default format selection in settings for convenience.


:small_blue_diamond: 2. Custom Aspect Ratios for Social Media & Banners

Current Challenge:

  • DALL·E only provides square (1024x1024) and wide (1792x1024) images.

  • However, most professional use cases require specific aspect ratios:

    :pushpin: YouTube Banner2560x1440
    :pushpin: Twitter/X Header1500x500
    :pushpin: Instagram Story/Reels1080x1920
    :pushpin: LinkedIn Cover Photo1584x396

Proposed Solution:
:white_check_mark: Let users input custom aspect ratios (e.g., 16:9, 4:5, 9:16).
:white_check_mark: Offer preset templates for common social media sizes.


:small_blue_diamond: 3. “Remove Text” Functionality Without Changing the Image

Current Challenge:

  • When users ask DALL·E to remove text, it completely regenerates the image.
  • This makes it hard to iterate on a design without starting over.

Proposed Solution:
:white_check_mark: Enable text-layer removal while keeping the original design intact.
:white_check_mark: Provide a way to make small iterative edits instead of full redesigns.


:small_blue_diamond: 4. Manual Text Overlay & Basic Editing (Canva-Style Controls)

Current Challenge:

  • DALL·E generates text in images, but there is no manual control over font, size, position, or style.
  • Users must download the image, edit it externally, and then re-upload it for further processing.

Proposed Solution:
:white_check_mark: Provide a basic text overlay tool within DALL·E.
:white_check_mark: Allow users to manually add/edit text (select font, size, color, alignment, position).
:white_check_mark: Offer pre-made text styles for common design needs (e.g., social media, branding).


:rocket: Why This Matters

These enhancements will significantly improve DALL·E’s usability for content creators, marketers, and designers who rely on AI-generated visuals for branding and professional workflows.

By allowing custom formats, aspect ratios, manual text controls, and better text removal, DALL·E can become a true end-to-end solution for social media marketing, website design, and automation—eliminating the need for external tools like Photoshop or Canva.

Would love to hear thoughts from the OpenAI team and the community on these improvements! :rocket::fire: