Copy link to clipboard
Copied
There are far too many frustrating things in Firefly.
Why does Firefly not comprehend prompts. No matter what is requested! Or it adds things that nobody wanted, that don't even suit the theme of the current output.
Here is a topical example that beats me completely. I am requesting two WHITE males standing next to each other. There is no racism of any kind involved. How does one know what one can prompt and what not???
I get all sorts of ethnic groups! Any combination. I did not ask for any of them.
My story REQUIRES WHITE MALES. If I wanted any particular combination I would request that particular combination and I expect to get what I asked for. Nothing else. It is not a difficult request, is it now?!
I could write a book about all my frustrations. And I do read updated information.
Do I need ChatGPT-4 sitting on my lap so that it can compose my prompts and I then get what is needed.
Siiiiiiiigh
Please don't give me too many excuses.
Copy link to clipboard
Copied
Caucasian is the term for that.
Copy link to clipboard
Copied
Copy link to clipboard
Copied
Please post the exact prompt and some results.
Copy link to clipboard
Copied
Hello Monika, I hardly ever keep my prompts. I am sending along some results. 2 actually show one of the prompts.
I did also make some side comments, but I kept them limited.
I would appreciate your feedback.
Thank you
Gerd
Copy link to clipboard
Copied
Hi @joziG,
I have found that the best structure for a prompt in text to image is:
Based on that "caucasian males" works better than "both men white".
droopy
Copy link to clipboard
Copied
To add on @droopydog500 great advice, these prompts seems a little complicated. Sometimes you'll have better results when generating the landscape, then using generative fill to add elements to the image.
Copy link to clipboard
Copied
A little bit of additional information. I also got a female person! I certainly never asked for that. I did not even allude to anything female. Amongst the other wrongs that are unacceptable are faces or body parts which are either distorted, disfigured and/or full of artifacts. Could someone please try to explain to me why, on a few occasions, I got ultra-thin and very tall persons?? Why?? That's why I eventually stipulated a height and a profile for the body. Well, that was haphazardly adhered do. Just as an aside, I rendered many more images, all with very varied results. Not one satisfied me completely. I can't wrap my mind around the fact that a specific syntax should be adhered to. That's on the insane side. I am not addressing a human being! BTW, I did use the term Caucasian at times after that was suggested. Result: nothing changed.
Thank you all of you guys for putting up with me.
Copy link to clipboard
Copied
Hey @joziG
Just too add a bit more to what @Nite-Stian and @droopydog500 have mentioned.
1st) create a composition reference
In this case I asked for two caucasian men standing together.
Now I have a Composition Ref image to give FF to work with.
2nd I used this image as a Composition Ref and left the slider 1/2 way.
Prompt
two caucasian men, on a mountain top
This was the result i got.
Ask for the 2 caucasian men and then ask for the setting as Droppy mentioned.
Cheers
O. Nate