Skip to main content
Adobe Employee
March 23, 2023
Answered

P: Text and symbol generation in images

  • March 23, 2023
  • 131 replies
  • 94719 views

I am trying to generate a logo with a desired text. It is not accepting it.

Correct answer Kartika Rawat

Hi e339816051nxw!

Sorry about that. Currently, Firefly has limitations in creating realistic body features in generated images. The quality of our models will improve with time, and generated content will continue to better follow prompts. Select the report icon to provide feedback about distorted facial or body features.

 

Let us know.

Thanks,

KR

131 replies

Participant
July 6, 2025

I attempted to generate an image for a personal article I am writing on X. I was excited to see the Firefly feature and burned one of my credits to generate an image. I have had an adobe account with monthly billing of $21.73 since October 27, 2022 and of course this is first time using the 25/25 of credits I had (and they don't roll over). I gave a very specific prompt, and Firefly was unable to generate an image with the correct spelling of words from the prompt.

My Prompt was - Create Image, with the following words; "Democrat Party", "Republican Party", "Reform Party", "Libertarian Party", "Green Party", "America Party", "Constitution Party", "Alliance Party", "Forward Party", "No Labels Party" in random locations and a black background

Please see the attached image of the junk I was provided.

Community Expert
July 7, 2025

Hi @sagoad73 

Firefly has never been great at inserting text into an image. It is getting better but is still has problems. If you ask for a few words in an image, it may get it right in some generations. If you ask for a large number of words, it is unlikley that all the words will appear correctly. Maybe in future versions of Firefly, it will manage text better. For now, inserting text in images is not reliable.

Participant
July 4, 2025

I tell it how to spell something and it continues to misspell the word and I continue to use credits to generate videos or images attempting to have the word corrected

Kartika Rawat
Community Manager
Community Manager
July 4, 2025

Hi @otto_5475!

Sorry to hear that. I'm merging your thread here since it’s part of a similar discussion. Please see the correct answer above.

 

Let us know if this helps.

Thanks,

^KR

Participant
July 3, 2025

In all images I am generating, the font is in symbols and therefor looks like jibberish.

Kartika Rawat
Community Manager
Community Manager
July 3, 2025

Hi Chris37335260351r!

Thanks for writing in. I'm merging your thread here since it’s part of a similar discussion. Please see the correct answer above.

 

Let us know if this helps.


^KR

Participant
June 28, 2025

Firefly is terrible at handling words on screen.  Which only explains why it doesn't understand most the prompts I give it.  Even simpler prompts like  "a semi exiting the highway, signs that say 'no parking' or symbols with 'no semis'  always come out looking like alien language.   

 

In my mind, this is a major bug. 

Because you are Adobe, masters of font, creators of icons, leaders in design + innovation, you aren't supposed to be the company that forces long-time customers into lame credits that are used up after wasting my time. 

 

As another point, I took all the exact prompts I used and pluged them into Sora and I got what I wanted and Kling AI was better than Firefly.

 

Time to level up...

Participant
June 26, 2025

Also why can't we click on the misspelled word and correct it.I have to remove/erase and add Text which is difficult!

Monika Gause
Community Expert
Community Expert
June 26, 2025

You can upload the generated image into the Remove Background modul and have it erased.

 

If you want that amount of editing options, probably use Photoshop for it.

Participant
June 27, 2025
How do you do that in photoshop?

Remy Chappell


Get Outlook for iOS<>
Participant
June 20, 2025

I am asking for it to create a word in all capital letters and it creates the word but it cuts it off on both edges when you can't even read the whole thing. 

Monika Gause
Community Expert
Community Expert
June 20, 2025

@adam_0813  schrieb:

I am asking for it to create a word in all capital letters and it creates the word but it cuts it off on both edges when you can't even read the whole thing. 


 

Which module are you using?

Participant
June 6, 2025

Firefly is incredibly disappointing. Inputs were specific. Results were misspelled mishmash

Participant
May 23, 2025

Wenn ich ein Bild mit einem Monitor darstellen möchte, auf dem ein Text steht, wird der Text nicht richtig dargestellt (Buchstaben doppelt, falsch...). Woran liegt das? Kann das verbessert werden?

droopydog500
Community Manager
Community Manager
May 23, 2025

Hi @wolfram_2801,

 

Getting text right in AI image generation can be challenging. Generally, the best approach is to generate the item and add the text in with another app such as Photoshop or Adobe Express.

 

    droopy

Adobe Community Expert (not an Adobe employee)
Participant
May 23, 2025

It's bizarre to me that Adobe "doesn't do text" ... fonts are one of the core competencies, and graphics the second.  They must be concerned with 'offending' - taking too much artistic license or some sort of weird copyright thing - as there are sites popping up that DO THIS... never mind that putting in text that a user actually specifies shouldn't be any big deal in the big picture. (pun intended). 

Participant
May 17, 2025

When I enter text to be added on a throne the words engraved be in a different language.

Participant
May 11, 2025

Getting overlaid, incorrect text and at times text in foreign languaes when not prompted to do so

jane-e
Community Expert
Community Expert
May 12, 2025

@Vince337174220wyx 

 

See the answer marked "correct" from Droopy on why generative AI is not able to do text very well, and use Adobe express or another application for your text.

https://community.adobe.com/t5/adobe-firefly-discussions/p-text-and-symbol-generation-in-images/m-p/14585035#M10885

 

Jane