tips please to make sure no text gets added to the image…

[worthy kayak]

…because it doesn’t get the spellling correxxtt !?

[kindred oracle]

Are you using Bing, ChatGPT Plus, or something else? Can you share an example of a prompt and an image?

[worthy kayak]

Chatgpt plus and its not a specific prompt. I often get text in images. I want a prompt addendum that I can use to always get images without text….if possible

[kindred oracle]

Negative prompting is not a good tactic at present. That is, if you tell it not to do something, you're likely to just confuse it by bringing it up in the first place. So, just never mention text or a lack of text, you should have better results. Just focus on what you do want!

[worthy kayak]

Ok thanks for cool tips. Just curious how to adjust a prompt that gives@me text in image if I otherwise happy with the results.

[kindred oracle]

Can you share an example of one you've had problems with? It might help just as a starting point. But in general, if you just focus only on what you do want and not at all on what you don't want, the model will work best.

[worthy kayak]

Ok let me find an example we can use. But usually what I do is produce images for presentations etc and then I just use the context from the article or slide and prompt something like…create an image representing the following content.

[worthy kayak]

I just inputted this. It gave me some odd texts here and there:

A library with books where scifi books are no longer deemed scifi due to speed of technology development

The prompt generated by Dalle:

A futuristic library interior with towering shelves filled with books. The books once categorized as science fiction now have labels indicating they are general knowledge or history, reflecting the rapid advancement of technology. The library is sleek with glass and metal accents, interactive displays, and floating holographic book titles. It's well-lit, with a spacious layout, and there are people of diverse descents and genders browsing the shelves, discussing the once speculative concepts now realized. Some are using advanced gadgets to read or access information. The atmosphere is one of enlightenment and progress, merging classical library aesthetics with a modern, high-tech twist.

[kindred oracle]

I'm not surprised that a prompt about a library and books resulted in text being generated! But, let's unpack this a little further:

• "A library with books where scifi books are no longer deemed scifi due to speed of technology development"
• When you describe the situation as "no longer deemed scifi," what do you imagine that looking like in the picture? In other words, do you just mean "the scene should take place in a futuristic library"? In other words, this is kind of an abstract idea, and DALLE 3 will work best if you give it visual ideas--unless you want it to come up with its own.
• Which is what it did in this case. As you can see in the image prompt you also provided: it interpreted "no longer deemed scifi" to mean "I gotta put labels on these scifi books to say something else so people know." So in other words, it probably added even more text than a normal library/books image would have, simply because it got confused.
  What do you think? Maybe you could try something simpler like just "A futuristic library." Then, based off of what you see, you can fine-tune your images from there!

[prime atlas]

So generally prompts which have a lot of non-visual details will generate text. And as other users are pointing out, if you have things like "library" or "book" you are more likely to get words.