Copy link to clipboard
Text prompts are not right in image
All time mistakes in spelling
This is a current limitation for most Generative AI. I recommend trying Adobe Express. Please also add your vote and follow this post: P: Text is wrong in Text to Image
I'm guessing it doesn't like "Fools"
I have the same issue! Did you find a solution?
Thanks for posting..
Can you support your post with a screenshot of your prompt when that happens?
Firefly supports up to 100 languages, make sure that your language is supported https://www.adobe.com/sensei/generative-ai/firefly.html#faqs
See attached image download and screenshot with the prompt "your perfect nest"
This type of generation is not currently supported.
For example I type : "Word CORRECT on a sign"
in a promt - and the picture is good BUT the spelling
mostly (9 out of 10 times) shows a wrong spelled word ...
Words like RROTCE, ORCTERCIC, etc is on the sign instead,
and sometimes there is no word at all on the sign.
Is there maybe a command in a promt to force correct spelling?
When I write "word" before the word in promt, it will -more often-
try to put a name on the sign ...
What to do? ... Or is spelling a limit / bug maybe?
Чому не пише чіткий потрібний текст на зображенні ? А щось схоже уродує.
Ось замість "agrodolina" якесь "acradoiina" ітп
I would like to see FF Txt->Image to enhance the recognition of words.
I used Word "Merry Christmas" in front of christmas scene, watercolor, simple
It created image below.
FF will recognize some words, but gets lost with more complex words
Word "Joyous Christmas" in front of christmas scene, watercolor, simple. (see 2nd Image)
The ability to create formatted phrases would be fantastic
I find it recognizes some simple words
Word "Merry Christmas" in front of christmas scene, watercolor, simple (image 1)
Word "Joyous Christmas" in front of christmas scene, watercolor, simple(image2)
Ah, this has been the case from day dot pretty much.
I wonder if Adobe will be enabling correct type in FF or if the preference is for text to be done in Express.
This an example of what DALL-E3 does with prompt
Malheureusemnt quuand on est écrit en arabe ça ne forme pas typographyquement les mots, le site met seulement des lettres d'alpahbet arabe cote a cote et ce n'est pas du tout le résulat qu'on souhaitent. Svp veillez résoudre ce souci pour qu'on puisse former des mots entier, merci a vous l'équipe je sais que vous êtes capable de résoudre cela & force a vous 😉
In every AI generator, they never use the words or image I request. WHY??... it all ends up being a waste of time trying to get a result that is useful. Hours wasted with nothing notable.
Hi @SteveLanc ,
It's something that Firefly has to work on.
It's not just Arabic. Adobe Firefly does not do any text properly. See this post as one example:
Trying to get Firefly to add three (3) words in English to my generated image. No can do?
Would be great to add a section with a word that could be created onto the photo like a logo etc. The prompt does not spell the words I write in it to create the image.
I am trying to generate an image with English words in there, but it creates the words in another language
So i asked the ai to make a banner with the word Rules written on it for my server..The ai always showed me results with spelling errors in it . The spelling of rules was shown as Rulles , Rues , Ruels not a single image was correct and i wasted my tokens .I hope you look into this ,Thank you .
Proof : https://firefly.adobe.com/public/t2i?id=urn%3Aaaid%3Asc%3AAP%3A11598cd9-db1b-41da-b357-171a99c4586b&...