005
01-24-2026, 10:19 PM
Gemini is not able to correct spelling mistakes in posters created by itself.
I asked Gemini to create a poster and add a text given by me. It creates the poster with altered text, ie with spelling mistakes. Even after repeatedly giving instructions to correct the spelling it is unable to do it. And slowly it starts adding more and more weird words rather correcting the spelling mistake.like in the below image the spelling is "other"but it has printed as "oather". Actually I had asked it to print "each other"
https://storage.googleapis.com/support-forums-api/attachment/thread-383700289-13316534063551704491.jpg
Fred SR
Platinum Product Expert
I love helping users solve their issues.
Oct 28, 2025
Hello,
My name is Fred SR, and I'm a volunteer Product Expert in the Gemini Apps Community. I understand how frustrating it must be when you ask for specific text on a poster and it comes out with spelling errors like "oather" instead of "each other." Thank you for sharing that image; it's a perfect example of the issue.
This is actually a known limitation with the current image generation models. When Gemini adds text to an image, it's not "typing" the words as you and I do. It's "drawing" the shapes of the letters as part of the overall image. This is why it can sometimes misinterpret the text or struggle to make precise corrections, as it's treating the text as a visual pattern rather than as language.
The team is aware of this, and feedback like yours is crucial for making improvements. While I can't provide a direct fix, here are a couple of workarounds you can try:
1. Be Very Specific in a New Chat
Sometimes, starting a fresh chat and being extremely clear in your first prompt can help.
Try starting a new chat.
Use quotation marks for the exact text you want. For example: Create a poster of a couple in a cafe. Add the exact text: "we looked at each other a little too deeply to be 'just friends'".
This can sometimes help the model prioritize the exact spelling from the beginning.
2. Generate the Image and Add Text Separately
This is often the most reliable method for getting perfect text.
Ask Gemini to create the poster without any text (e.g., Create a realistic image of a couple looking at each other in a cozy cafe by a window).
Once you have an image you like, save it to your device.
Use a photo editor (like Google Photos, which has a "Markup" tool) to add the text yourself. This gives you full control over the spelling, font, and placement.
Help Improve Gemini
The best way to get this issue in front of the engineering team is to send your feedback directly through the app. The example you provided is exactly what they need to see.
Here’s how to send feedback on your Android app:
At the top, tap your Profile picture or initial.
Tap Feedback.
Enter your description of the issue (you can even mention the "oather" vs. "each other" problem).
Make sure the "Include screenshots and logs" option checked.
Tap Send.
You can find more info on this process here: Send feedback or report a problem with Gemini Apps. (https://support.google.com/gemini/answer/13275746?hl=en&co=GENIE.Platform%3DAndroid)
I know this isn't a perfect solution, but I hope these workarounds help you get the results you're looking for.
Would you like me to help you draft a new prompt to try and get that image without the text?
/* src.: https://support.google.com/gemini/thread/383700289/gemini-is-not-able-to-correct-spelling-mistakes-in-posters-created-by-itself?hl=en-GB
I asked Gemini to create a poster and add a text given by me. It creates the poster with altered text, ie with spelling mistakes. Even after repeatedly giving instructions to correct the spelling it is unable to do it. And slowly it starts adding more and more weird words rather correcting the spelling mistake.like in the below image the spelling is "other"but it has printed as "oather". Actually I had asked it to print "each other"
https://storage.googleapis.com/support-forums-api/attachment/thread-383700289-13316534063551704491.jpg
Fred SR
Platinum Product Expert
I love helping users solve their issues.
Oct 28, 2025
Hello,
My name is Fred SR, and I'm a volunteer Product Expert in the Gemini Apps Community. I understand how frustrating it must be when you ask for specific text on a poster and it comes out with spelling errors like "oather" instead of "each other." Thank you for sharing that image; it's a perfect example of the issue.
This is actually a known limitation with the current image generation models. When Gemini adds text to an image, it's not "typing" the words as you and I do. It's "drawing" the shapes of the letters as part of the overall image. This is why it can sometimes misinterpret the text or struggle to make precise corrections, as it's treating the text as a visual pattern rather than as language.
The team is aware of this, and feedback like yours is crucial for making improvements. While I can't provide a direct fix, here are a couple of workarounds you can try:
1. Be Very Specific in a New Chat
Sometimes, starting a fresh chat and being extremely clear in your first prompt can help.
Try starting a new chat.
Use quotation marks for the exact text you want. For example: Create a poster of a couple in a cafe. Add the exact text: "we looked at each other a little too deeply to be 'just friends'".
This can sometimes help the model prioritize the exact spelling from the beginning.
2. Generate the Image and Add Text Separately
This is often the most reliable method for getting perfect text.
Ask Gemini to create the poster without any text (e.g., Create a realistic image of a couple looking at each other in a cozy cafe by a window).
Once you have an image you like, save it to your device.
Use a photo editor (like Google Photos, which has a "Markup" tool) to add the text yourself. This gives you full control over the spelling, font, and placement.
Help Improve Gemini
The best way to get this issue in front of the engineering team is to send your feedback directly through the app. The example you provided is exactly what they need to see.
Here’s how to send feedback on your Android app:
At the top, tap your Profile picture or initial.
Tap Feedback.
Enter your description of the issue (you can even mention the "oather" vs. "each other" problem).
Make sure the "Include screenshots and logs" option checked.
Tap Send.
You can find more info on this process here: Send feedback or report a problem with Gemini Apps. (https://support.google.com/gemini/answer/13275746?hl=en&co=GENIE.Platform%3DAndroid)
I know this isn't a perfect solution, but I hope these workarounds help you get the results you're looking for.
Would you like me to help you draft a new prompt to try and get that image without the text?
/* src.: https://support.google.com/gemini/thread/383700289/gemini-is-not-able-to-correct-spelling-mistakes-in-posters-created-by-itself?hl=en-GB