Simple Prompts not understood , even in non-fast mode

[ravi_2912]

Hi, 

I was just having look around and did some testing and started with very simple prompts, I tried a very ordinary prompt to say design a pic that has some simple words written like example "abc" or "xyz", (always a single word) the output pics that will get generated will have text that would be truncated like say only one or 2 alphabets in the picture.
Is there something wrong I am doing with prompt ? or is this a model limitation. 

thanks.  

[Oh.N8]

Hey @ravi_2912 

Could you provide a screenshot and an example of your prompt(s); so, we can better help you? 

Cheers

O. Nate

[ravi_2912]

Hi Nate, 

Here is a sample Screenshot. 

Also the outout rejects my objective, if I dont specifically mention the word ' text ' in the prompt. Example screenshot below. 

I hope the Firefly model improves and becomes more precise in future. 

Thanks

Ravi

[Monika Gause · Accepted solution]

Firefly cannot spell. If you want this to be a specific layout, like in a logo or the like, make a reference image for that that also contains the text in the font you like.

[ravi_2912]

Ok, got it. So, firefly needs a first draft kind of thing to generate image with custom text. 

Also, why does it ignore background color instruction ? The water is more of greener shade than blue. 

[Monika Gause]

I can't see. abackground instruction in your prompt. "Water" can basically have any color. The only color you mention at all is green.