Adobe FireFly Beta : Praying Mantis

 I finally have some time to test Adobe Firefly Beta. In this exercise, I hope to understand the process of Adobe Firefly workflow as well as to learn how to communicate with AI using prompts that AI can understand.

Lets starts simple . I was thinking of combining two ideas , praying mantis and martial arts. 

I must have confused the AI , "two hands" is treated as a separate object and not the hands of the praying mantis. The image generated on the top left is an interesting perspective.

 I am unsure if the hands are actually legs of of the praying mantis . I join the two sentence in the previous prompt choosing to use the verb "wielding" and not mentioning hands or legs.

PROMPT :  " Praying mantis in a big white room practicing martial arts wielding a long sword "

The one pose that I like is the mantis holding a long stick instead of a sword. I use it as a reference image and adjust the bar more towards the meaning of the prompt to regenerate till I found the image I like on the lower bottom right.

There are quite a bit of things to correct in the picture which I use generative fill to do it.

PROMPT : " Remove feeler "  apparently the AI understood this works better than "remove line".

PROMPT : "  "  for sword handle , neck area , shadow for it to regenerate alternatives images.

PROMPT " Join to body "  

Use the word body rather than the word leg.

To add extension of the leg use the following prompt

PROMPT " Add body "

PROMPT " add broken wing "

The prompt does not work it is against user agreement , no generation of blood or violent images. 

PROMPT: "Splatter liquid on the floor"

Final collaborative human + AI image