Next-Diffusion-SD-Demo / scripts /prompt_generator.py
svjack's picture
Upload 10 files
5dfbe1d
raw
history blame
9.51 kB
import json
import random
import gradio as gr
import modules
from pathlib import Path
from modules import script_callbacks
import modules.scripts as scripts
result_prompt = ""
base_dir = scripts.basedir()
dropdown_options_file = Path(base_dir, "json/dropdown_options.json")
category_data_file = Path(base_dir, "json/category_data.json")
style_data_file = Path(base_dir, "json/style_data.json")
prefix_data_file = Path(base_dir, "json/prefix_data.json")
lightning_data_file = Path(base_dir, "json/lightning_data.json")
lens_data_file = Path(base_dir, "json/lens_data.json")
class Model:
'''
Small strut to hold data for the text generator
'''
def __init__(self, name) -> None:
self.name = name
pass
def populate_dropdown_options():
path = dropdown_options_file
with open(path, 'r') as f:
data = json.load(f)
category_choices = data["category"]
style_choices = data["style"]
lightning_choices = data["lightning"]
lens_choices = data["lens"]
return tuple(category_choices), tuple(style_choices), tuple(lightning_choices), tuple(lens_choices),
def add_to_prompt(*args):
prompt, use_default_negative_prompt = args
default_negative_prompt = "(worst quality:1.2), (low quality:1.2), (lowres:1.1), (monochrome:1.1), (greyscale), multiple views, comic, sketch, (((bad anatomy))), (((deformed))), (((disfigured))), watermark, multiple_views, mutation hands, mutation fingers, extra fingers, missing fingers, watermark"
if(use_default_negative_prompt):
return prompt, default_negative_prompt
else:
return prompt, ""
def get_random_prompt(data):
random_key = random.choice(list(data.keys()))
random_array = random.choice(data[random_key])
random_strings = random.sample(random_array, 3)
return random_strings
def get_correct_prompt(data, selected_dropdown):
correct_array = data[selected_dropdown]
random_array = random.choice(correct_array)
random_strings = random.sample(random_array, 3)
random_strings.insert(0, selected_dropdown)
return random_strings
def generate_prompt_output(*args):
#all imported files
prefix_path = prefix_data_file
category_path = category_data_file
style_path = style_data_file
lightning_path = lightning_data_file
lens_path = lens_data_file
#destructure args
category, style, lightning, lens, negative_prompt = args
# Convert variables to lowercase
category = category.lower()
style = style.lower()
lightning = lightning.lower()
lens = lens.lower()
# Open category_data.json and grab correct text
with open(prefix_path, 'r') as f:
prefix_data = json.load(f)
prefix_prompt = random.sample(prefix_data, 6)
modified_prefix_prompt = [f"(({item}))" for item in prefix_prompt]
# Open category_data.json and grab correct text
with open(category_path, 'r') as f2:
category_data = json.load(f2)
if category == "none":
category_prompt = ""
elif category == "random":
category_prompt = get_random_prompt(category_data)
else:
category_prompt = get_correct_prompt(category_data, category)
# Open style_data.json and grab correct text
with open(style_path, 'r') as f3:
style_data = json.load(f3)
if style == "none":
style_prompt = ""
elif style == "random":
style_prompt = get_random_prompt(style_data)
else:
style_prompt = get_correct_prompt(style_data, style)
# Open lightning_data.json and grab correct text
with open(lightning_path, 'r') as f4:
lightning_data = json.load(f4)
if lightning == "none":
lightning_prompt = ""
elif lightning == "random":
lightning_prompt = get_random_prompt(lightning_data)
else:
lightning_prompt = get_correct_prompt(lightning_data, lightning)
# Open lens_data.json and grab correct text
with open(lens_path, 'r') as f5:
lens_data = json.load(f5)
if lens == "none":
lens_prompt = ""
elif lens == "random":
lens_prompt = get_random_prompt(lens_data)
else:
lens_prompt = get_correct_prompt(lens_data, lens)
prompt_output = modified_prefix_prompt, category_prompt, style_prompt, lightning_prompt, lens_prompt
prompt_strings = []
for sublist in prompt_output:
# Join the sublist elements into a single string
prompt_string = ", ".join(str(item) for item in sublist)
if prompt_string: # Check if the prompt_string is not empty
prompt_strings.append(prompt_string)
# Join the non-empty prompt_strings
final_output = ", ".join(prompt_strings)
return final_output
def on_ui_tabs():
# UI structure
txt2img_prompt = modules.ui.txt2img_paste_fields[0][0]
img2img_prompt = modules.ui.img2img_paste_fields[0][0]
txt2img_negative_prompt = modules.ui.txt2img_paste_fields[1][0]
img2img_negative_prompt = modules.ui.img2img_paste_fields[1][0]
with gr.Blocks(analytics_enabled=False) as prompt_generator:
with gr.Tab("Prompt Generator"):
with gr.Row(): # Use Row to arrange two columns side by side
with gr.Column(): # Left column for dropdowns
category_choices, style_choices, lightning_choices, lens_choices = populate_dropdown_options()
with gr.Row():
gr.HTML('''<h2 id="input_header">Input 👇</h2>''')
with gr.Row().style(equal_height=True): # Place dropdowns side by side
# Create a dropdown to select
category_dropdown = gr.Dropdown(
choices=category_choices,
value=category_choices[1],
label="Category", show_label=True
)
style_dropdown = gr.Dropdown(
choices=style_choices,
value=style_choices[1],
label="Style", show_label=True
)
with gr.Row():
lightning_dropdown = gr.Dropdown(
choices=lightning_choices,
value=lightning_choices[1],
label="Lightning", show_label=True
)
lens_dropdown = gr.Dropdown(
choices=lens_choices,
value=lens_choices[1],
label="Lens", show_label=True
)
with gr.Row():
gr.HTML('''
<hr class="rounded" id="divider">
''')
with gr.Row():
gr.HTML('''<h2 id="input_header">Links</h2>''')
with gr.Row():
gr.HTML('''
<h3>Stable Diffusion Tutorials⚡</h3>
<container>
<a href="https://nextdiffusion.ai" target="_blank">
<button id="website_button" class="external-link">Website</button>
</a>
<a href="https://www.youtube.com/channel/UCd9UIUkLnjE-Fj-CGFdU74Q?sub_confirmation=1" target="_blank">
<button id="youtube_button" class="external-link">YouTube</button>
</a>
</container>
''')
with gr.Column(): # Right column for result_textbox and generate_button
# Add a Textbox to display the generated text
with gr.Row():
gr.HTML('''<h2 id="output_header">Output 👋</h2>''')
result_textbox = gr.Textbox(label="Generated Prompt", lines=3)
use_default_negative_prompt = gr.Checkbox(label="Include Negative Prompt", value=True, interactive=True, elem_id="negative_prompt_checkbox")
setattr(use_default_negative_prompt,"do_not_save_to_config",True)
with gr.Row():
txt2img = gr.Button("Send to txt2img")
img2img = gr.Button("Send to img2img")
# Create a button to trigger text generation
txt2img.click(add_to_prompt, inputs=[result_textbox, use_default_negative_prompt], outputs=[txt2img_prompt, txt2img_negative_prompt ]).then(None, _js='switch_to_txt2img',inputs=None, outputs=None)
img2img.click(add_to_prompt, inputs=[result_textbox, use_default_negative_prompt], outputs=[img2img_prompt, img2img_negative_prompt]).then(None, _js='switch_to_img2img',inputs=None, outputs=None)
generate_button = gr.Button(value="Generate", elem_id="generate_button")
# Register the callback for the Generate button
generate_button.click(fn=generate_prompt_output, inputs=[category_dropdown, style_dropdown, lightning_dropdown, lens_dropdown, use_default_negative_prompt], outputs=[result_textbox])
return (prompt_generator, "Next Diffusion ⚡", "Next Diffusion ⚡"),
script_callbacks.on_ui_tabs(on_ui_tabs)