example-prompts.txt

# *******************************************************************************
# Example prompts.txt file
# These settings should be a good starting point for a GPU with 8GB of VRAM
# You can try these out by typing:
# python make_art.py example-prompts.txt
# see docs for settings explanations: https://github.com/rbbrdckybk/ai-art-generator#usage
# *******************************************************************************

# *******************************************************************************
# subjects go here, one per line
# subjects are required
# setting directives are also allowed here and affect everything that follows them
# *******************************************************************************
[subjects]

# VQGAN+CLIP settings
!PROCESS = vqgan
!WIDTH = 384
!HEIGHT = 384
!ITERATIONS = 500
!LEARNING_RATE = 0.1
!CUTS = 32
!CLIP_MODEL = ViT-B/32
!TRANSFORMER = 
!INPUT_IMAGE =

# prompts here will use the VQGAN+CLIP settings above
a bright red apple
a monkey on a motorcycle

# CLIP-guided diffusion settings
!PROCESS = diffusion
!WIDTH = 768
!HEIGHT = 448
!ITERATIONS = 250
!CUTS = 4
!D_VITB32 = yes
!D_VITB16 = yes
!D_VITL14 = no
!D_RN101 = no
!D_RN50 = yes
!D_RN50x4 = no
!D_RN50x16 = no
!D_RN50x64 = no
!INPUT_IMAGE =

# prompts here will use the CLIP-guided diffusion settings above
a shining lighthouse on the shore of a tropical island in a raging storm

# stable diffusion settings
!PROCESS = stablediff
!SD_LOW_MEMORY = no
!WIDTH = 384
!HEIGHT = 384
!STEPS = 50
!SCALE = 7.5
!SAMPLES = 1
!BATCH_SIZE = 1
!USE_UPSCALE = no
!UPSCALE_AMOUNT = 2.0
!UPSCALE_FACE_ENH = no
!UPSCALE_KEEP_ORG = no

# prompts here will use the stable diffusion settings above
a proud corgi astronaut
Obama as an avocado

# *******************************************************************************
# styles go here - think of these as directives in how to render the subject
# every style listed here will be combined with every subject above to create an image
# these are optional
# *******************************************************************************
[styles]

watercolor
by picasso
pencil sketch


# *******************************************************************************
# prefixes will be placed in front of your subject
# these are optional
# if you list multiple lines, one will be chosen at random
#   e.g. if you're making faces, you might put something like:
#   a beautiful portrait of
# *******************************************************************************
[prefixes]


# *******************************************************************************
# suffixes will be appended after your subject or style depending on process used
# these are optional
# if you list multiple lines, one will be chosen at random
#   e.g. if you have a bunch of keywords relevant to all of your subjects, like:
#   trending on artstation, unreal engine, 4k, masterpiece, vivid colors
# *******************************************************************************
[suffixes]