Skip to main content
Ambrosial_Lionheart3804
Participant
August 17, 2026
Answered

How to place the AI-model of myself into an UNCHANGED (!) background ?

  • August 17, 2026
  • 5 replies
  • 37 views

My beginner prompts are very simple and clear:

Set myself into the image ! (which works easily and well with Nano Banana, ChatGPT for example)

It’s doesn’t work for me with Adobe FireFly and the Adobe-KI-model of myself.

When I try to set myself into a picture of my hometown Dresden then the background gets changed into an irrelevant fantasy world, a weird mix of Prague and Paris with some Dresden buildings.

I don’t want any building changed, absolutely please no, stop !

Is there an easy way to do so ? I successfully tried the fill-feature, I can set a reference photo to fill me in, but I look for a way to use the KI-model of myself as well ? 

    Correct answer N8.

    Hey ​@Ambrosial_Lionheart3804 

    When I use a reference image and want to keep the reference I tend to use one these prompts at the beginning of the Prompt.

    Prompt suggestion 1 (for your case this one could work best)
    Keep the reference image…

    Prompt suggestion 2
    Use the reference image…

    Cheers

    Nate

     

    5 replies

    N8.
    Community Expert
    N8.Community ExpertCorrect answer
    August 19, 2026

    Hey ​@Ambrosial_Lionheart3804 

    When I use a reference image and want to keep the reference I tend to use one these prompts at the beginning of the Prompt.

    Prompt suggestion 1 (for your case this one could work best)
    Keep the reference image…

    Prompt suggestion 2
    Use the reference image…

    Cheers

    Nate

     

    Ambrosial_Lionheart3804Author
    Participant
    August 19, 2026

    Thanks, this actually works best. However, I found that this trained KI of myself is not usable for my (possibly naive) purpose. The quality of the outputs is so incredibly low quality that I will stick to much better Nano Banana and the generative fill. (12 baroque sculptures get changed into 12x the same toy etc). 

    Tarun Saini
    Community Manager
    Tarun SainiCommunity Manager
    Community Manager
    August 18, 2026

    Hi ​@Ambrosial_Lionheart3804,

     

    To better understand what's happening, could you please share:

    1. Which Firefly feature are you using (Generate Image, Scene to Image, Generative Fill, Boards, etc.)?
    2. Are you using a custom Firefly model trained on yourself, or are you uploading a reference photo each time?
    3. Does the background change every time, or only with certain images of Dresden?

     

    Regards,

    Tarun

    Ambrosial_Lionheart3804Author
    Participant
    August 18, 2026

    Thanks you for your reply. (My questions might be weird as I’m all-new to imaging.)

    1. I try to combine Generative Fill with the custom Firefly model trained on myself.
    2. I upload a background as only reference, not a reference to fill, of course I’m open to try that.
    3. The entire background gets changed every time, this is what I don’t like.

    I want to insert myself (a trained AI version of me) into the central foreground of the image, in front of the railing. The output I get is a all-new image featuring a completely different bridge and new buildings in the background—a different city I don't recognize. Is there a way to mark a specific area that is the only part allowed to change? Essentially, I want to use Generative Fill (only a selected area) to add the trained AI ​​version of myself.

    I want to stand here. It works with generative fill. ? Can I also use the trained AI-model in selected parts, without changing the whole image ?

    regards, Steffen (Steve)

    Ambrosial_Lionheart3804Author
    Participant
    August 18, 2026
    This is what I get. Wrong buildings, wrong ships, wrong kind of tram, looks like a fantasy game ?

     

    Terms & ConditionsAccessibility statement
    Using the community Terms of use Privacy Cookie preferences Do not sell or share my personal information ad choicesAdChoices