P: Text and symbol generation in images

[Angelic_child5CF3]

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

[Tracy.H]

Text generation in images is not supported yet.

[Tracy.H]

Text and symbol generation in images is not supported yet in the Text to Image module. The quality of our models will improve with time, and generated content will better respect the prompt. If you are trying to create just text, please try the Text Effects module. (Merging all similar threads here so we can track the number of reports for this)

[Tracy.H]

Text and symbol generation in images is not supported yet in the Text to Image module. The quality of our models will improve with time, and generated content will better respect the prompt. If you are trying to create just text, please try the Text Effects module. (Merging all similar threads here so we can track the number of reports for this)

[mark22680575xh8s]

Trying to add text within an image does not work correctly. For example, creating a sign with a message on it will does not render the message correctly.

[Brad29123175cgtr]

I prompted with "Band of Certainty". The 4 resulting images had text in them that were close to "Certainty", but were misspelled. Images looked cool though!

[Tracy.H]

Text generation in images is not supported yet. Try expanding your prompt with more words to get better results

[Gary36964750unj4]

This is bizarre that it generates semi- random hieroglyphics rather than the text specified.

[Kevin Stohlmeyer]

@Gary36964750unj4 Most Image Generative AI models do not understand language enough to generate anything comprehensive. It's trying and ends up producing gibberish.

[Gary36964750unj4]

I think they're trying to do more than needed - to understand the text - rather than just insert the wording (lettering).

A more sinister interpretation is that Adobe doesn't add this, so that it necessitates the use of other Adobe Suite products.... Illustrator/etc.



Gary

[Kevin Stohlmeyer]

@Gary36964750unj4 that is 100% false. Try creating text using other AI solutions if you think Adobe is being "sinister" and see how the results turn out.

[Gary36964750unj4]

OK, thanks - good to know. Still seems odd, at the same time my knowledge of the ins and outs of AI is severely limited.


Gary

[jane-e]

@Gary36964750unj4 

The free version of Adobe Express works for text.

Jane

[Kevin Stohlmeyer]

@jane-e that is a different solution - the Generate Text Effects is generating text styles on existing text using AI not adding text to an AI generated image.

[droopydog500]

I think what @jane-e was suggesting was adding the text to an already created image using Adobe Express, which for many cases is sufficient.

Please see this article on why generating text in AI image generators is hard:

If AI image generators are so smart, why do they struggle to write and count?

    droopy

[TheRealThaddeus]

19 out of 20 submissions result in Firefly not being able to simply spell the name "Thaddeus" correctly. It adds a random letter, or drops a needed letter.

[krishk67117260]

for example i have tested a name of a company and then tried to create a cover for that company flyer. but unfortunately the company name and other english words are not spelled correctly 

[Saiva Veshrajsena]

It would really be great if you could add text to map feature where we could get maps of various cities or countries. This would help professers like me or teachers who could educate indegenous people regarding their ancient terriotaries. This would help to show professers and teachers to educate indegenous children regarding their ancestral history and community.

[carlb12392193]

I've been trying a few very basic applications of Firefly (text to image) to create quick personalised greeting 'posters' (which is in itself incredible, especially when I compare to how I have done these until now!).

I find that the interpretation of the graphic (ie, image) request (eg, multi-color text with a background of fireworks bursting) is spot on but the text request (eg, type 'Wow! Caron' as part of the 'text to image' instruction) gives some very unpredictable images - including letters missing, different letters, added letters, a change in spelling, etc - including even with the same instruction and the 4 different versions that pop up as suggestions. 

Maybe I'm doing something wrong in how I'm asking Firefly for this but thought it useful to share this feedback as a constructive contribution to the growth of this amazing tool. 

[Henrik Heigl]

Hi,

First of all, let's not forget that this is still a beta version. So all we here discuss can drastically change over the next time.

That Problem with Text in the Image is the same with other KI Tools in this area. I also suspect that it is related to the fact that you could prevent unwanted violations of the guidelines (see also FAQ https://firefly.adobe.com/faq ) but all that maybe change in the next Updates. So, be Patience and let us have a lot of fun with Firefly.

[John293759942c45]

A lot of image generative AI creates some very cool images. However, the one issue I see most often, is that sometimes I want to create an image that also includes text.  When trying to add this text to the description, the AI engine doesn't know how to interpret it and it ends up adding the characters randomly to the image.

It would be great to have a way to descibre the image I want to create AND the text I would like included on the image.

[Tracy.H]

@John293759942c45 Merging all similar threads here so we can track the number of reports for this

[Matthew26536919jh19]

Text to logo 

Prompt: "Create logo for a company called Modular Commerce. They are a company of footwear and apparel experts helping independent retailers sell online." 

The logo text is not Modular Commerce. The same result occurs when Modular Commerce is within ""

Merging all similar threads here so the team can access all requests on this issue

[myfc21]

Hello everyone,
i report a problem that i find in using. if i ask for a caption on the image, for example a writing on a t-shirt, it does not take it correct, but doesun my of the letters or integer the writing.

Example: THAKA becomes: TAAKAA, TAKAA, AHHHAAATKKA

Merging all similar threads here so the team can access all requests on this issue

[Ukiya]

Firefly tend to generate images with typo in the text

Merging all similar threads here so we can track the number of reports for this

[always_clicking]

Playing around with "text-to-image",  the images that are created with the prompt are correct, however,  when asked to include text in the image, most times the text is misspelled even with in quotes.  I have included a screen shot of my prompt and the results.  This happens in all of my prompts.

Merging all similar threads here so the team can access all requests on this issue