[Flux1 D only]Are you trying to draw the image in your head?

FrozenGunner had a question about prompting on flux1, so I will address it here. Thank you for your tip.

Step 1
Shut your eyes and envision the drawing you wish to bring to life.


Step 2
Use Google Images to search for the elements in your mind and find the one that best suits your needs.

Step 3

Feed the image you found on

https://huggingface.co/spaces/gokaygokay/Florence-2

with "More Detailed Caption".

Or you can feed the image into ChatGPT and ask it: "describe this image in 200/300/350 words".(350 may cerate many pointless words)

You will got the text. Check it for the correctness.

https://www.youtube.com/watch?v=YRzKY7n6BVU

Step 4

Copy the trigger words of the style lora and into the text you created. The limitation is around 380 words. And generate it.

PS: You may find it boring soon if you don't use those art style Loras, as Flux1D lacks artist style variety.


I am still searching for ways to ensure Flux1D becomes better.