Skip to main content
Known Participant
May 23, 2026
Answered

How come chatGPT makes better job

  • May 23, 2026
  • 1 reply
  • 8 views

Respecting  guide image boundaries and general geometry and edges , understanding the target style and requirements while Firefly just ignores it all altogether and makes something while maybe nice looking but totally outside of what I requested.    Looks rather like nice toy than an instrument . 

In fact I gave up with fierfly for some time already but  still waiting if it might be getting more intelligent?           It took chatGPT a few tries maybe but it really understand what I am asking and does closer and closer with each new iteration  while with Firefly it stay pure  dice throwing all the time.

Am I missing something?   Some special wording  technique , special commands  to stay on track? 

I especially need it to follow my  initial guide image boundaries and specific scale,   an ortho or map  projection  rather than perspective image  recognizing and  following vector shapes.  Chat can do it for example. 

    Correct answer Oh.N8

    Hey ​@kirill_7256 

    You may want to use the BETA version. It is a mix of different models and gives you more options to refine an image.
    https://firefly.adobe.com/ai-assistant


    According to ChatGPT

    1. Different priorities

    Adobe Firefly is designed first around:

    • commercial safety
    • copyright cleanliness
    • integration into Adobe workflows
    • predictable editing tools

    OpenAI image generation is optimized more aggressively for:

    • realism
    • prompt understanding
    • composition
    • aesthetic quality
    • conversational iteration

    Those goals lead to different outputs.

    Firefly often feels “safe,” polished, and stock-photo-like because Adobe intentionally constrains the model.

    ChatGPT image generation tends to take more creative risks and follow nuanced instructions more closely.

    Cheers

    Nate

    1 reply

    Oh.N8Community ExpertCorrect answer
    May 24, 2026

    Hey ​@kirill_7256 

    You may want to use the BETA version. It is a mix of different models and gives you more options to refine an image.
    https://firefly.adobe.com/ai-assistant


    According to ChatGPT

    1. Different priorities

    Adobe Firefly is designed first around:

    • commercial safety
    • copyright cleanliness
    • integration into Adobe workflows
    • predictable editing tools

    OpenAI image generation is optimized more aggressively for:

    • realism
    • prompt understanding
    • composition
    • aesthetic quality
    • conversational iteration

    Those goals lead to different outputs.

    Firefly often feels “safe,” polished, and stock-photo-like because Adobe intentionally constrains the model.

    ChatGPT image generation tends to take more creative risks and follow nuanced instructions more closely.

    Cheers

    Nate