Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
|
@@ -1,13 +1,20 @@
|
|
| 1 |
import gradio as gr
|
| 2 |
import requests
|
| 3 |
from PIL import Image
|
|
|
|
| 4 |
from transformers import BlipProcessor, BlipForConditionalGeneration
|
| 5 |
import time
|
| 6 |
from gradio_client import Client
|
| 7 |
|
|
|
|
| 8 |
blipper="Salesforce/blip-image-captioning-large"
|
| 9 |
chatter="K00B404/transcript_image_generator"
|
| 10 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 11 |
# Load BLIP model for image captioning
|
| 12 |
processor = BlipProcessor.from_pretrained(blipper)
|
| 13 |
model = BlipForConditionalGeneration.from_pretrained(blipper)
|
|
@@ -32,9 +39,9 @@ Some characters might be terse while others might be more verbose."""
|
|
| 32 |
return persona
|
| 33 |
|
| 34 |
|
| 35 |
-
def
|
| 36 |
"""Function to interact with the chatbot API using the generated persona"""
|
| 37 |
-
|
| 38 |
try:
|
| 39 |
# Call the API with the current message and system prompt (persona)
|
| 40 |
response = chatbot_client.predict(
|
|
@@ -78,8 +85,11 @@ def generate_persona(img, min_len, max_len, persona_detail_level):
|
|
| 78 |
# Calculate processing time
|
| 79 |
end = time.time()
|
| 80 |
total_time = f"Processing time: {end - start:.2f} seconds"
|
| 81 |
-
|
| 82 |
-
|
|
|
|
|
|
|
|
|
|
| 83 |
|
| 84 |
|
| 85 |
def chat_with_persona(message, history, system_message, max_tokens, temperature, top_p):
|
|
@@ -98,6 +108,48 @@ def chat_with_persona(message, history, system_message, max_tokens, temperature,
|
|
| 98 |
except Exception as e:
|
| 99 |
return f"Error communicating with the chatbot API: {str(e)}"
|
| 100 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 101 |
# Create Gradio interface with tabs
|
| 102 |
with gr.Blocks(title="Image Character Persona Generator") as iface:
|
| 103 |
# Store the generated persona in a state variable to share between tabs
|
|
@@ -129,6 +181,7 @@ with gr.Blocks(title="Image Character Persona Generator") as iface:
|
|
| 129 |
3. Select detail level for the persona
|
| 130 |
4. Click "Generate Character Persona"
|
| 131 |
5. Switch to the "Test Persona" tab to chat with your character
|
|
|
|
| 132 |
""")
|
| 133 |
|
| 134 |
# Second tab: Test Character Chat
|
|
@@ -176,6 +229,31 @@ with gr.Blocks(title="Image Character Persona Generator") as iface:
|
|
| 176 |
[msg, chatbot])
|
| 177 |
|
| 178 |
clear_btn.click(clear_chat, outputs=chatbot)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 179 |
|
| 180 |
# Function to update system prompt in Test tab when persona is generated
|
| 181 |
def update_persona_state(caption, persona, time_output):
|
|
|
|
| 1 |
import gradio as gr
|
| 2 |
import requests
|
| 3 |
from PIL import Image
|
| 4 |
+
import os
|
| 5 |
from transformers import BlipProcessor, BlipForConditionalGeneration
|
| 6 |
import time
|
| 7 |
from gradio_client import Client
|
| 8 |
|
| 9 |
+
token = os.getenv('HF_TOKEN')
|
| 10 |
blipper="Salesforce/blip-image-captioning-large"
|
| 11 |
chatter="K00B404/transcript_image_generator"
|
| 12 |
|
| 13 |
+
# Set your API endpoint and authorization details
|
| 14 |
+
API_URL = "https://api-inference.huggingface.co/models/black-forest-labs/FLUX.1-schnell"
|
| 15 |
+
headers = {fAuthorization": f"Bearer {token}"} # Replace with your actual token
|
| 16 |
+
timeout = 60 # seconds
|
| 17 |
+
|
| 18 |
# Load BLIP model for image captioning
|
| 19 |
processor = BlipProcessor.from_pretrained(blipper)
|
| 20 |
model = BlipForConditionalGeneration.from_pretrained(blipper)
|
|
|
|
| 39 |
return persona
|
| 40 |
|
| 41 |
|
| 42 |
+
def helper_llm(message, system_prompt, max_tokens=256, temperature=0.5, top_p=0.95):
|
| 43 |
"""Function to interact with the chatbot API using the generated persona"""
|
| 44 |
+
|
| 45 |
try:
|
| 46 |
# Call the API with the current message and system prompt (persona)
|
| 47 |
response = chatbot_client.predict(
|
|
|
|
| 85 |
# Calculate processing time
|
| 86 |
end = time.time()
|
| 87 |
total_time = f"Processing time: {end - start:.2f} seconds"
|
| 88 |
+
|
| 89 |
+
# dramaturg to mae a solid role for a actor from pragmatic description
|
| 90 |
+
system_message="You are a Expert Dramaturg and your task is to use the input persona information and write a 'Role' description as compact instuctions for the actor"
|
| 91 |
+
persona = helper_llm(persona, system_prompt=system_prompt)
|
| 92 |
+
return caption, persona, total_time
|
| 93 |
|
| 94 |
|
| 95 |
def chat_with_persona(message, history, system_message, max_tokens, temperature, top_p):
|
|
|
|
| 108 |
except Exception as e:
|
| 109 |
return f"Error communicating with the chatbot API: {str(e)}"
|
| 110 |
|
| 111 |
+
|
| 112 |
+
|
| 113 |
+
|
| 114 |
+
|
| 115 |
+
def generate_flux_image(final_prompt, is_negative, steps, cfg_scale, seed, strength):
|
| 116 |
+
"""
|
| 117 |
+
Generate an image using the FLUX model via Hugging Face's inference API.
|
| 118 |
+
The function sends a POST request with the given payload and returns the image,
|
| 119 |
+
along with the seed and prompt used.
|
| 120 |
+
"""
|
| 121 |
+
payload = {
|
| 122 |
+
"inputs": final_prompt,
|
| 123 |
+
"is_negative": is_negative,
|
| 124 |
+
"steps": steps,
|
| 125 |
+
"cfg_scale": cfg_scale,
|
| 126 |
+
"seed": seed,
|
| 127 |
+
"strength": strength
|
| 128 |
+
}
|
| 129 |
+
|
| 130 |
+
response = requests.post(API_URL, headers=headers, json=payload, timeout=timeout)
|
| 131 |
+
|
| 132 |
+
if response.status_code != 200:
|
| 133 |
+
print(f"Error: Failed to get image. Response status: {response.status_code}")
|
| 134 |
+
print(f"Response content: {response.text}")
|
| 135 |
+
if response.status_code == 503:
|
| 136 |
+
raise gr.Error(f"{response.status_code} : The model is being loaded")
|
| 137 |
+
raise gr.Error(f"{response.status_code}")
|
| 138 |
+
|
| 139 |
+
try:
|
| 140 |
+
image_bytes = response.content
|
| 141 |
+
image = Image.open(io.BytesIO(image_bytes))
|
| 142 |
+
# Optionally save the image to a file (filename based on seed)
|
| 143 |
+
output_path = f"./output_{seed}.png"
|
| 144 |
+
image.save(output_path)
|
| 145 |
+
print(f'\033[1mGeneration completed!\033[0m (Prompt: {final_prompt})')
|
| 146 |
+
return output_path, str(seed), final_prompt
|
| 147 |
+
except Exception as e:
|
| 148 |
+
print(f"Error when trying to open the image: {e}")
|
| 149 |
+
return None, None, None
|
| 150 |
+
|
| 151 |
+
|
| 152 |
+
|
| 153 |
# Create Gradio interface with tabs
|
| 154 |
with gr.Blocks(title="Image Character Persona Generator") as iface:
|
| 155 |
# Store the generated persona in a state variable to share between tabs
|
|
|
|
| 181 |
3. Select detail level for the persona
|
| 182 |
4. Click "Generate Character Persona"
|
| 183 |
5. Switch to the "Test Persona" tab to chat with your character
|
| 184 |
+
6. create similar images inspired by the 'role'
|
| 185 |
""")
|
| 186 |
|
| 187 |
# Second tab: Test Character Chat
|
|
|
|
| 229 |
[msg, chatbot])
|
| 230 |
|
| 231 |
clear_btn.click(clear_chat, outputs=chatbot)
|
| 232 |
+
|
| 233 |
+
|
| 234 |
+
# New Tab 3: Flux Image Generation
|
| 235 |
+
with gr.Tab("Flux Image Generation"):
|
| 236 |
+
gr.Markdown("### Flux Image Generation")
|
| 237 |
+
final_prompt = gr.Textbox(label="Prompt", lines=2, placeholder="Enter your prompt for Flux...")
|
| 238 |
+
is_negative = gr.Checkbox(label="Use Negative Prompt", value=False)
|
| 239 |
+
steps = gr.Slider(minimum=10, maximum=100, step=1, value=50, label="Steps")
|
| 240 |
+
cfg_scale = gr.Slider(minimum=1, maximum=20, step=1, value=7, label="CFG Scale")
|
| 241 |
+
seed = gr.Number(value=42, label="Seed")
|
| 242 |
+
strength = gr.Slider(minimum=0.0, maximum=1.0, step=0.05, value=0.8, label="Strength")
|
| 243 |
+
|
| 244 |
+
generate_button = gr.Button("Generate Flux Image")
|
| 245 |
+
output_image = gr.Image(label="Generated Image")
|
| 246 |
+
output_seed = gr.Textbox(label="Seed Used")
|
| 247 |
+
output_prompt = gr.Textbox(label="Prompt Used")
|
| 248 |
+
|
| 249 |
+
generate_button.click(
|
| 250 |
+
fn=generate_flux_image,
|
| 251 |
+
inputs=[final_prompt, is_negative, steps, cfg_scale, seed, strength],
|
| 252 |
+
outputs=[output_image, output_seed, output_prompt]
|
| 253 |
+
)
|
| 254 |
+
|
| 255 |
+
|
| 256 |
+
|
| 257 |
|
| 258 |
# Function to update system prompt in Test tab when persona is generated
|
| 259 |
def update_persona_state(caption, persona, time_output):
|