Update app.py
Browse files
app.py
CHANGED
@@ -1,7 +1,8 @@
|
|
1 |
import gradio as gr
|
2 |
import torch
|
3 |
from transformers import pipeline, set_seed
|
4 |
-
|
|
|
5 |
import openai
|
6 |
import os
|
7 |
import time
|
@@ -15,7 +16,6 @@ openai_available = False
|
|
15 |
|
16 |
if api_key:
|
17 |
try:
|
18 |
-
openai.api_key = api_key
|
19 |
# Starting with openai v1, client instantiation is preferred
|
20 |
openai_client = openai.OpenAI(api_key=api_key)
|
21 |
# Simple test to check if the key is valid (optional, but good)
|
@@ -39,29 +39,31 @@ asr_pipeline = None
|
|
39 |
try:
|
40 |
print("Loading ASR pipeline (Whisper) on CPU...")
|
41 |
# Force CPU usage with device=-1 or device="cpu"
|
42 |
-
|
|
|
43 |
print("ASR pipeline loaded successfully on CPU.")
|
44 |
except Exception as e:
|
45 |
print(f"Could not load ASR pipeline: {e}. Voice input will be disabled.")
|
46 |
traceback.print_exc() # Print full traceback for debugging
|
47 |
|
48 |
-
# 2. 文本到图像模型 (
|
49 |
image_generator_pipe = None
|
|
|
|
|
50 |
try:
|
51 |
-
print("Loading
|
52 |
-
print("
|
53 |
-
|
54 |
-
|
55 |
-
image_generator_pipe = StableDiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.float32)
|
56 |
image_generator_pipe = image_generator_pipe.to(device)
|
57 |
-
print("
|
58 |
except Exception as e:
|
59 |
-
print(f"CRITICAL: Could not load
|
60 |
traceback.print_exc() # Print full traceback for debugging
|
61 |
# Define a dummy object to prevent crashes later if loading failed
|
62 |
class DummyPipe:
|
63 |
def __call__(self, *args, **kwargs):
|
64 |
-
raise RuntimeError(f"
|
65 |
image_generator_pipe = DummyPipe()
|
66 |
|
67 |
|
@@ -73,17 +75,25 @@ def enhance_prompt_openai(short_prompt, style_modifier="cinematic", quality_boos
|
|
73 |
if not openai_available or not openai_client:
|
74 |
# Fallback or error if OpenAI key is missing/invalid
|
75 |
print("OpenAI not available. Returning original prompt with modifiers.")
|
76 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
77 |
if not short_prompt:
|
78 |
# Return an error message formatted for Gradio output
|
79 |
raise gr.Error("Input description cannot be empty.")
|
80 |
|
81 |
# Construct the prompt for the OpenAI model
|
82 |
system_message = (
|
83 |
-
"You are an expert prompt engineer for AI image generation models
|
84 |
-
"Expand the user's short description into a detailed, vivid, and coherent prompt. "
|
85 |
-
"Focus on
|
86 |
-
"Incorporate the requested style and quality keywords naturally. Avoid conversational text."
|
|
|
87 |
)
|
88 |
user_message = (
|
89 |
f"Enhance this description: \"{short_prompt}\". "
|
@@ -94,13 +104,13 @@ def enhance_prompt_openai(short_prompt, style_modifier="cinematic", quality_boos
|
|
94 |
|
95 |
try:
|
96 |
response = openai_client.chat.completions.create(
|
97 |
-
model="gpt-3.5-turbo", # Cost-effective choice
|
98 |
messages=[
|
99 |
{"role": "system", "content": system_message},
|
100 |
{"role": "user", "content": user_message},
|
101 |
],
|
102 |
temperature=0.7, # Controls creativity vs predictability
|
103 |
-
max_tokens=
|
104 |
n=1, # Generate one response
|
105 |
stop=None # Let the model decide when to stop
|
106 |
)
|
@@ -108,7 +118,7 @@ def enhance_prompt_openai(short_prompt, style_modifier="cinematic", quality_boos
|
|
108 |
print("OpenAI enhancement successful.")
|
109 |
# Basic cleanup: remove potential quotes around the whole response
|
110 |
if enhanced_prompt.startswith('"') and enhanced_prompt.endswith('"'):
|
111 |
-
|
112 |
return enhanced_prompt
|
113 |
except openai.AuthenticationError:
|
114 |
print("OpenAI Authentication Error: Invalid API key?")
|
@@ -127,38 +137,61 @@ def enhance_prompt_openai(short_prompt, style_modifier="cinematic", quality_boos
|
|
127 |
|
128 |
# Step 2: Prompt-to-Image (CPU)
|
129 |
def generate_image_cpu(prompt, negative_prompt, guidance_scale, num_inference_steps):
|
130 |
-
"""Generates image using
|
131 |
-
|
132 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
133 |
if not prompt or "[Error:" in prompt or "Error:" in prompt:
|
134 |
# Check if the prompt itself is an error message from the previous step
|
135 |
raise gr.Error("Cannot generate image due to invalid or missing prompt.")
|
136 |
|
137 |
print(f"Generating image on CPU for prompt: {prompt[:100]}...") # Log truncated prompt
|
138 |
-
|
139 |
-
|
|
|
|
|
|
|
140 |
start_time = time.time()
|
141 |
|
142 |
try:
|
143 |
# Use torch.inference_mode() or torch.no_grad() for efficiency
|
144 |
with torch.no_grad():
|
145 |
# Seed for reproducibility (optional, but good practice)
|
146 |
-
generator = torch.Generator(device=device).manual_seed(int(time.time()))
|
147 |
-
|
148 |
-
|
149 |
-
|
150 |
-
|
151 |
-
|
152 |
-
|
153 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
154 |
end_time = time.time()
|
155 |
-
print(f"Image generated successfully on CPU in {end_time - start_time:.2f} seconds.")
|
156 |
return image
|
157 |
except Exception as e:
|
158 |
-
print(f"Error during image generation on CPU: {e}")
|
159 |
traceback.print_exc()
|
160 |
# Propagate error to Gradio UI
|
161 |
-
raise gr.Error(f"Image generation failed on CPU: {e}")
|
162 |
|
163 |
|
164 |
# Bonus: Voice-to-Text (CPU)
|
@@ -174,6 +207,7 @@ def transcribe_audio(audio_file_path):
|
|
174 |
start_time = time.time()
|
175 |
try:
|
176 |
# Ensure the pipeline uses the correct device (should be CPU based on loading)
|
|
|
177 |
transcription = asr_pipeline(audio_file_path)["text"]
|
178 |
end_time = time.time()
|
179 |
print(f"Transcription successful in {end_time - start_time:.2f} seconds.")
|
@@ -201,20 +235,37 @@ def process_input(input_text, audio_file, style_choice, quality_choice, neg_prom
|
|
201 |
print(f"Using text input: '{final_text_input}'")
|
202 |
elif audio_file is not None:
|
203 |
print("Processing audio input...")
|
204 |
-
|
205 |
-
|
206 |
-
#
|
207 |
-
|
208 |
-
|
209 |
-
|
210 |
-
|
211 |
-
|
212 |
-
|
213 |
-
|
214 |
-
|
215 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
216 |
print(status_message)
|
|
|
217 |
return status_message, None # Return error
|
|
|
218 |
else:
|
219 |
status_message = "[Error: No input provided. Please enter text or record audio.]"
|
220 |
print(status_message)
|
@@ -224,7 +275,7 @@ def process_input(input_text, audio_file, style_choice, quality_choice, neg_prom
|
|
224 |
if final_text_input:
|
225 |
try:
|
226 |
enhanced_prompt = enhance_prompt_openai(final_text_input, style_choice, quality_choice)
|
227 |
-
status_message = enhanced_prompt # Display the prompt
|
228 |
print(f"Enhanced prompt: {enhanced_prompt}")
|
229 |
except gr.Error as e:
|
230 |
# Catch Gradio-specific errors from enhancement function
|
@@ -240,22 +291,31 @@ def process_input(input_text, audio_file, style_choice, quality_choice, neg_prom
|
|
240 |
return status_message, None
|
241 |
|
242 |
# 3. Generate Image (if prompt is valid)
|
|
|
243 |
if enhanced_prompt and not status_message.startswith("[Error:") and not status_message.startswith("[Prompt Enhancement Error:"):
|
244 |
try:
|
245 |
# Show "Generating..." message while waiting
|
246 |
-
gr.Info("Starting image generation on CPU
|
247 |
generated_image = generate_image_cpu(enhanced_prompt, neg_prompt, guidance, steps)
|
248 |
gr.Info("Image generation complete!")
|
249 |
except gr.Error as e:
|
250 |
# Catch Gradio errors from generation function
|
251 |
-
|
|
|
252 |
print(f"Image Generation Error: {e}")
|
|
|
253 |
except Exception as e:
|
|
|
254 |
status_message = f"{enhanced_prompt}\n\n[Unexpected Image Generation Error: {e}]"
|
255 |
print(f"Unexpected Image Generation Error: {e}")
|
256 |
traceback.print_exc()
|
257 |
-
#
|
258 |
-
|
|
|
|
|
|
|
|
|
|
|
259 |
|
260 |
# 4. Return results to Gradio UI
|
261 |
# Return the status message (enhanced prompt or error) and the image (or None if error)
|
@@ -267,22 +327,31 @@ def process_input(input_text, audio_file, style_choice, quality_choice, neg_prom
|
|
267 |
style_options = ["cinematic", "photorealistic", "anime", "fantasy art", "cyberpunk", "steampunk", "watercolor", "illustration", "low poly"]
|
268 |
quality_options = ["highly detailed", "sharp focus", "intricate details", "4k", "masterpiece", "best quality", "professional lighting"]
|
269 |
|
270 |
-
#
|
271 |
-
|
272 |
-
|
|
|
273 |
|
274 |
with gr.Blocks(theme=gr.themes.Soft()) as demo:
|
275 |
-
gr.Markdown("# AI Image Generator (CPU Version)")
|
276 |
gr.Markdown(
|
277 |
"**Enter a short description or use voice input.** The app uses OpenAI (if API key is provided) "
|
278 |
-
"to create a detailed prompt, then generates an image using
|
279 |
)
|
280 |
-
# Add specific warning about
|
281 |
-
gr.HTML("<p style='color:orange;font-weight:bold;'>⚠️
|
282 |
|
283 |
# Display OpenAI availability status
|
284 |
if not openai_available:
|
285 |
gr.Markdown("**Note:** OpenAI API key not found or invalid. Prompt enhancement will use a basic fallback.")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
286 |
|
287 |
with gr.Row():
|
288 |
with gr.Column(scale=1):
|
@@ -294,27 +363,31 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
|
|
294 |
inp_audio = gr.Audio(sources=["microphone"], type="filepath", label="Or record your idea (clears text box if used)")
|
295 |
else:
|
296 |
gr.Markdown("**Voice input disabled:** Whisper model failed to load.")
|
297 |
-
|
|
|
298 |
|
299 |
# --- Controls (Step 3 requirements met) ---
|
|
|
|
|
300 |
# Control 1: Dropdown
|
301 |
-
inp_style = gr.Dropdown(label="Base Style", choices=style_options, value="cinematic")
|
302 |
# Control 2: Radio
|
303 |
-
inp_quality = gr.Radio(label="Quality Boost", choices=quality_options, value="highly detailed")
|
304 |
# Control 3: Textbox (Negative Prompt)
|
305 |
-
inp_neg_prompt = gr.Textbox(label="Negative Prompt (optional)", placeholder="e.g., blurry, low quality, text, watermark,
|
306 |
# Control 4: Slider (Guidance Scale)
|
307 |
-
inp_guidance = gr.Slider(minimum=1.0, maximum=15.0, step=0.5, value=
|
308 |
# Control 5: Slider (Inference Steps) - Reduced max/default
|
309 |
-
inp_steps = gr.Slider(minimum=
|
310 |
|
311 |
# --- Action Button ---
|
312 |
-
|
|
|
313 |
|
314 |
with gr.Column(scale=1):
|
315 |
# --- Outputs ---
|
316 |
out_prompt = gr.Textbox(label="Generated Prompt / Status", interactive=False, lines=5) # Show prompt or error status here
|
317 |
-
out_image = gr.Image(label="Generated Image", type="pil")
|
318 |
|
319 |
# --- Event Handling ---
|
320 |
# Define inputs list carefully, handling potentially invisible audio input
|
@@ -322,32 +395,44 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
|
|
322 |
if asr_pipeline:
|
323 |
inputs_list.append(inp_audio)
|
324 |
else:
|
325 |
-
inputs_list.append(
|
|
|
326 |
|
327 |
inputs_list.extend([inp_style, inp_quality, inp_neg_prompt, inp_guidance, inp_steps])
|
328 |
|
|
|
329 |
btn_generate.click(
|
330 |
fn=process_input,
|
331 |
inputs=inputs_list,
|
332 |
outputs=[out_prompt, out_image]
|
333 |
)
|
334 |
|
335 |
-
# Clear text input if audio is used
|
336 |
if asr_pipeline:
|
337 |
-
def
|
338 |
-
|
339 |
-
|
340 |
-
|
341 |
-
|
|
|
|
|
|
|
|
|
|
|
342 |
|
343 |
|
344 |
# ---- Application Launch ----
|
345 |
if __name__ == "__main__":
|
346 |
-
#
|
347 |
-
if not isinstance(image_generator_pipe,
|
348 |
-
print("
|
349 |
-
|
350 |
-
|
|
|
|
|
|
|
|
|
351 |
|
352 |
# Launch the Gradio app
|
353 |
-
|
|
|
|
1 |
import gradio as gr
|
2 |
import torch
|
3 |
from transformers import pipeline, set_seed
|
4 |
+
# 导入 AutoPipelineForText2Image 以便兼容不同模型
|
5 |
+
from diffusers import AutoPipelineForText2Image
|
6 |
import openai
|
7 |
import os
|
8 |
import time
|
|
|
16 |
|
17 |
if api_key:
|
18 |
try:
|
|
|
19 |
# Starting with openai v1, client instantiation is preferred
|
20 |
openai_client = openai.OpenAI(api_key=api_key)
|
21 |
# Simple test to check if the key is valid (optional, but good)
|
|
|
39 |
try:
|
40 |
print("Loading ASR pipeline (Whisper) on CPU...")
|
41 |
# Force CPU usage with device=-1 or device="cpu"
|
42 |
+
# 使用 fp16 会更快但需要GPU,CPU上用 float32
|
43 |
+
asr_pipeline = pipeline("automatic-speech-recognition", model="openai/whisper-base", device=device, torch_dtype=torch.float32)
|
44 |
print("ASR pipeline loaded successfully on CPU.")
|
45 |
except Exception as e:
|
46 |
print(f"Could not load ASR pipeline: {e}. Voice input will be disabled.")
|
47 |
traceback.print_exc() # Print full traceback for debugging
|
48 |
|
49 |
+
# 2. 文本到图像模型 (Tiny Text-to-Image) - 资源友好模型
|
50 |
image_generator_pipe = None
|
51 |
+
# 使用资源需求极低的 Tiny Text-to-Image 模型
|
52 |
+
model_id = "hf-internal-testing/tiny-text-to-image"
|
53 |
try:
|
54 |
+
print(f"Loading Text-to-Image pipeline ({model_id}) on CPU...")
|
55 |
+
print("NOTE: Using a very small model for resource efficiency. Image quality will be lower than Stable Diffusion.")
|
56 |
+
# 使用 AutoPipelineForText2Image 自动识别模型类型
|
57 |
+
image_generator_pipe = AutoPipelineForText2Image.from_pretrained(model_id, torch_dtype=torch.float32)
|
|
|
58 |
image_generator_pipe = image_generator_pipe.to(device)
|
59 |
+
print(f"Text-to-Image pipeline ({model_id}) loaded successfully on CPU.")
|
60 |
except Exception as e:
|
61 |
+
print(f"CRITICAL: Could not load Text-to-Image pipeline ({model_id}): {e}. Image generation will fail.")
|
62 |
traceback.print_exc() # Print full traceback for debugging
|
63 |
# Define a dummy object to prevent crashes later if loading failed
|
64 |
class DummyPipe:
|
65 |
def __call__(self, *args, **kwargs):
|
66 |
+
raise RuntimeError(f"Text-to-Image model failed to load: {e}")
|
67 |
image_generator_pipe = DummyPipe()
|
68 |
|
69 |
|
|
|
75 |
if not openai_available or not openai_client:
|
76 |
# Fallback or error if OpenAI key is missing/invalid
|
77 |
print("OpenAI not available. Returning original prompt with modifiers.")
|
78 |
+
# Basic fallback prompt enhancement
|
79 |
+
if short_prompt:
|
80 |
+
return f"{short_prompt}, {style_modifier}, {quality_boost}"
|
81 |
+
else:
|
82 |
+
# If short prompt is empty, fallback should also indicate error
|
83 |
+
raise gr.Error("Input description cannot be empty.")
|
84 |
+
|
85 |
+
|
86 |
if not short_prompt:
|
87 |
# Return an error message formatted for Gradio output
|
88 |
raise gr.Error("Input description cannot be empty.")
|
89 |
|
90 |
# Construct the prompt for the OpenAI model
|
91 |
system_message = (
|
92 |
+
"You are an expert prompt engineer for AI image generation models. "
|
93 |
+
"Expand the user's short description into a detailed, vivid, and coherent prompt, suitable for smaller, faster text-to-image models. "
|
94 |
+
"Focus on clear subjects, objects, and main scene elements. "
|
95 |
+
"Incorporate the requested style and quality keywords naturally, but keep the overall prompt concise enough for smaller models. Avoid conversational text."
|
96 |
+
# Adjusting guidance for smaller models
|
97 |
)
|
98 |
user_message = (
|
99 |
f"Enhance this description: \"{short_prompt}\". "
|
|
|
104 |
|
105 |
try:
|
106 |
response = openai_client.chat.completions.create(
|
107 |
+
model="gpt-3.5-turbo", # Cost-effective choice
|
108 |
messages=[
|
109 |
{"role": "system", "content": system_message},
|
110 |
{"role": "user", "content": user_message},
|
111 |
],
|
112 |
temperature=0.7, # Controls creativity vs predictability
|
113 |
+
max_tokens=100, # Limit output length - reduced for potentially shorter prompts for smaller models
|
114 |
n=1, # Generate one response
|
115 |
stop=None # Let the model decide when to stop
|
116 |
)
|
|
|
118 |
print("OpenAI enhancement successful.")
|
119 |
# Basic cleanup: remove potential quotes around the whole response
|
120 |
if enhanced_prompt.startswith('"') and enhanced_prompt.endswith('"'):
|
121 |
+
enhanced_prompt = enhanced_prompt[1:-1]
|
122 |
return enhanced_prompt
|
123 |
except openai.AuthenticationError:
|
124 |
print("OpenAI Authentication Error: Invalid API key?")
|
|
|
137 |
|
138 |
# Step 2: Prompt-to-Image (CPU)
|
139 |
def generate_image_cpu(prompt, negative_prompt, guidance_scale, num_inference_steps):
|
140 |
+
"""Generates image using the loaded model on CPU."""
|
141 |
+
# 检查加载的模型是否是期望的pipeline类型或DummyPipe
|
142 |
+
if not isinstance(image_generator_pipe, AutoPipelineForText2Image):
|
143 |
+
# If it's a DummyPipe or None for some reason
|
144 |
+
if isinstance(image_generator_pipe, DummyPipe):
|
145 |
+
# DummyPipe will raise its own error when called, so just let it
|
146 |
+
pass # The call below will raise the intended error
|
147 |
+
else:
|
148 |
+
# Handle unexpected case where pipe is not loaded correctly
|
149 |
+
raise gr.Error("Image generation pipeline is not available (failed to load model).")
|
150 |
+
|
151 |
+
|
152 |
if not prompt or "[Error:" in prompt or "Error:" in prompt:
|
153 |
# Check if the prompt itself is an error message from the previous step
|
154 |
raise gr.Error("Cannot generate image due to invalid or missing prompt.")
|
155 |
|
156 |
print(f"Generating image on CPU for prompt: {prompt[:100]}...") # Log truncated prompt
|
157 |
+
# Note: Negative prompt and guidance scale might have less impact or behave differently
|
158 |
+
# on very small models like tiny-text-to-image.
|
159 |
+
print(f"Negative prompt: {negative_prompt}") # Will likely be ignored by tiny model
|
160 |
+
print(f"Guidance scale: {guidance_scale}, Steps: {num_inference_steps}") # Steps might be fixed internally by tiny model
|
161 |
+
|
162 |
start_time = time.time()
|
163 |
|
164 |
try:
|
165 |
# Use torch.inference_mode() or torch.no_grad() for efficiency
|
166 |
with torch.no_grad():
|
167 |
# Seed for reproducibility (optional, but good practice)
|
168 |
+
# generator = torch.Generator(device=device).manual_seed(int(time.time())) # Tiny model might not use generator param
|
169 |
+
# Tiny Text-to-Image pipeline call structure might be simpler
|
170 |
+
# Check model specific documentation if parameters like guidance_scale, num_inference_steps, negative_prompt
|
171 |
+
# are actually supported. They might be ignored.
|
172 |
+
# Using a simple call that is generally compatible
|
173 |
+
output = image_generator_pipe(prompt=prompt) # Tiny model might only take prompt
|
174 |
+
|
175 |
+
# The output structure varies between pipelines, assuming it has .images
|
176 |
+
# if hasattr(output, 'images') and isinstance(output.images, list) and len(output.images) > 0:
|
177 |
+
# image = output.images[0] # Access the first image
|
178 |
+
# else:
|
179 |
+
# # Handle cases where output format is different
|
180 |
+
# print("Warning: Pipeline output format unexpected. Assuming the output itself is the image.")
|
181 |
+
# image = output # Assume output is the image if no .images
|
182 |
+
|
183 |
+
# Based on tiny-text-to-image, the output is likely a tuple where the first element is a list of images
|
184 |
+
image = output[0][0] # Access the first image in the first list of the tuple output structure
|
185 |
+
|
186 |
+
|
187 |
end_time = time.time()
|
188 |
+
print(f"Image generated successfully on CPU in {end_time - start_time:.2f} seconds (using {model_id}).")
|
189 |
return image
|
190 |
except Exception as e:
|
191 |
+
print(f"Error during image generation on CPU ({model_id}): {e}")
|
192 |
traceback.print_exc()
|
193 |
# Propagate error to Gradio UI
|
194 |
+
raise gr.Error(f"Image generation failed on CPU ({model_id}): {e}")
|
195 |
|
196 |
|
197 |
# Bonus: Voice-to-Text (CPU)
|
|
|
207 |
start_time = time.time()
|
208 |
try:
|
209 |
# Ensure the pipeline uses the correct device (should be CPU based on loading)
|
210 |
+
# Ensure input is in expected format for Whisper pipeline (filepath or audio array)
|
211 |
transcription = asr_pipeline(audio_file_path)["text"]
|
212 |
end_time = time.time()
|
213 |
print(f"Transcription successful in {end_time - start_time:.2f} seconds.")
|
|
|
235 |
print(f"Using text input: '{final_text_input}'")
|
236 |
elif audio_file is not None:
|
237 |
print("Processing audio input...")
|
238 |
+
try:
|
239 |
+
# Gradio might pass a tuple (samplerate, audio_data) or a filepath depending on type="filepath" vs "numpy"
|
240 |
+
# transcribe_audio expects a filepath based on the Gradio component config
|
241 |
+
if isinstance(audio_file, tuple):
|
242 |
+
# If Gradio gives tuple for some reason, try to save to temp file or adjust transcribe_audio
|
243 |
+
# Assuming type="filepath" works as expected and passes filepath
|
244 |
+
audio_filepath_to_transcribe = audio_file[0] # This might be incorrect depending on Gradio version/config
|
245 |
+
print(f"Warning: Gradio audio input was tuple, attempting to use first element as path: {audio_filepath_to_transcribe}")
|
246 |
+
else:
|
247 |
+
audio_filepath_to_transcribe = audio_file # This is expected for type="filepath"
|
248 |
+
|
249 |
+
transcribed_text, _ = transcribe_audio(audio_filepath_to_transcribe)
|
250 |
+
|
251 |
+
if "[Error:" in transcribed_text:
|
252 |
+
# Display transcription error clearly
|
253 |
+
status_message = transcribed_text
|
254 |
+
print(status_message)
|
255 |
+
return status_message, None # Return error in prompt field, no image
|
256 |
+
elif transcribed_text:
|
257 |
+
final_text_input = transcribed_text
|
258 |
+
print(f"Using transcribed audio input: '{final_text_input}'")
|
259 |
+
else:
|
260 |
+
status_message = "[Error: Audio input received but transcription was empty.]"
|
261 |
+
print(status_message)
|
262 |
+
return status_message, None # Return error
|
263 |
+
except Exception as e:
|
264 |
+
status_message = f"[Unexpected Audio Transcription Error: {e}]"
|
265 |
print(status_message)
|
266 |
+
traceback.print_exc()
|
267 |
return status_message, None # Return error
|
268 |
+
|
269 |
else:
|
270 |
status_message = "[Error: No input provided. Please enter text or record audio.]"
|
271 |
print(status_message)
|
|
|
275 |
if final_text_input:
|
276 |
try:
|
277 |
enhanced_prompt = enhance_prompt_openai(final_text_input, style_choice, quality_choice)
|
278 |
+
status_message = enhanced_prompt # Display the prompt initially
|
279 |
print(f"Enhanced prompt: {enhanced_prompt}")
|
280 |
except gr.Error as e:
|
281 |
# Catch Gradio-specific errors from enhancement function
|
|
|
291 |
return status_message, None
|
292 |
|
293 |
# 3. Generate Image (if prompt is valid)
|
294 |
+
# Check if the enhanced prompt step resulted in an error message
|
295 |
if enhanced_prompt and not status_message.startswith("[Error:") and not status_message.startswith("[Prompt Enhancement Error:"):
|
296 |
try:
|
297 |
# Show "Generating..." message while waiting
|
298 |
+
gr.Info(f"Starting image generation on CPU using {model_id}. This should be fast but quality is low.")
|
299 |
generated_image = generate_image_cpu(enhanced_prompt, neg_prompt, guidance, steps)
|
300 |
gr.Info("Image generation complete!")
|
301 |
except gr.Error as e:
|
302 |
# Catch Gradio errors from generation function
|
303 |
+
# Prepend original enhanced prompt to the error message for context
|
304 |
+
status_message = f"{enhanced_prompt}\n\n[Image Generation Error: {e}]"
|
305 |
print(f"Image Generation Error: {e}")
|
306 |
+
generated_image = None # Ensure image is None on error
|
307 |
except Exception as e:
|
308 |
+
# Catch any other unexpected errors
|
309 |
status_message = f"{enhanced_prompt}\n\n[Unexpected Image Generation Error: {e}]"
|
310 |
print(f"Unexpected Image Generation Error: {e}")
|
311 |
traceback.print_exc()
|
312 |
+
generated_image = None # Ensure image is None on error
|
313 |
+
|
314 |
+
else:
|
315 |
+
# If prompt enhancement failed, status_message already contains the error
|
316 |
+
# In this case, we just return the existing status_message and None image
|
317 |
+
print("Skipping image generation due to prompt enhancement failure.")
|
318 |
+
|
319 |
|
320 |
# 4. Return results to Gradio UI
|
321 |
# Return the status message (enhanced prompt or error) and the image (or None if error)
|
|
|
327 |
style_options = ["cinematic", "photorealistic", "anime", "fantasy art", "cyberpunk", "steampunk", "watercolor", "illustration", "low poly"]
|
328 |
quality_options = ["highly detailed", "sharp focus", "intricate details", "4k", "masterpiece", "best quality", "professional lighting"]
|
329 |
|
330 |
+
# Tiny model is very fast, steps/guidance might be ignored or have less effect
|
331 |
+
# Keep sliders but note their limited impact on this specific model
|
332 |
+
default_steps = 10 # Tiny model often uses few steps internally
|
333 |
+
max_steps = 20 # Limit max steps as they might not matter much
|
334 |
|
335 |
with gr.Blocks(theme=gr.themes.Soft()) as demo:
|
336 |
+
gr.Markdown("# AI Image Generator (Resource-Friendly CPU Version)")
|
337 |
gr.Markdown(
|
338 |
"**Enter a short description or use voice input.** The app uses OpenAI (if API key is provided) "
|
339 |
+
f"to create a detailed prompt, then generates an image using a **small, fast model ({model_id}) on the CPU**."
|
340 |
)
|
341 |
+
# Add specific warning about image quality for the tiny model
|
342 |
+
gr.HTML("<p style='color:orange;font-weight:bold;'>⚠️ Note: Using a small model for compatibility. Image quality and resolution will be significantly lower than models like Stable Diffusion.</p>")
|
343 |
|
344 |
# Display OpenAI availability status
|
345 |
if not openai_available:
|
346 |
gr.Markdown("**Note:** OpenAI API key not found or invalid. Prompt enhancement will use a basic fallback.")
|
347 |
+
else:
|
348 |
+
gr.Markdown("**Note:** OpenAI API key found. Prompt will be enhanced using OpenAI.")
|
349 |
+
|
350 |
+
|
351 |
+
# Display Model loading status
|
352 |
+
if not isinstance(image_generator_pipe, AutoPipelineForText2Image):
|
353 |
+
gr.Markdown(f"**CRITICAL:** Image generation model ({model_id}) failed to load. Image generation is disabled. Check logs.")
|
354 |
+
|
355 |
|
356 |
with gr.Row():
|
357 |
with gr.Column(scale=1):
|
|
|
363 |
inp_audio = gr.Audio(sources=["microphone"], type="filepath", label="Or record your idea (clears text box if used)")
|
364 |
else:
|
365 |
gr.Markdown("**Voice input disabled:** Whisper model failed to load.")
|
366 |
+
# Using gr.State as a placeholder that holds None
|
367 |
+
inp_audio = gr.State(None)
|
368 |
|
369 |
# --- Controls (Step 3 requirements met) ---
|
370 |
+
# Note: These controls might have limited effect on the small model
|
371 |
+
gr.Markdown("*(Optional controls - Note: These may have limited or no effect on the small model used)*")
|
372 |
# Control 1: Dropdown
|
373 |
+
inp_style = gr.Dropdown(label="Base Style", choices=style_options, value="cinematic", interactive=True)
|
374 |
# Control 2: Radio
|
375 |
+
inp_quality = gr.Radio(label="Quality Boost", choices=quality_options, value="highly detailed", interactive=True)
|
376 |
# Control 3: Textbox (Negative Prompt)
|
377 |
+
inp_neg_prompt = gr.Textbox(label="Negative Prompt (optional)", placeholder="e.g., blurry, low quality, text, watermark", interactive=True)
|
378 |
# Control 4: Slider (Guidance Scale)
|
379 |
+
inp_guidance = gr.Slider(minimum=1.0, maximum=15.0, step=0.5, value=3.0, label="Guidance Scale (CFG)", interactive=True) # Lower default for small model
|
380 |
# Control 5: Slider (Inference Steps) - Reduced max/default
|
381 |
+
inp_steps = gr.Slider(minimum=1, maximum=max_steps, step=1, value=default_steps, label=f"Inference Steps (lower = faster but less detail, max {max_steps})", interactive=True)
|
382 |
|
383 |
# --- Action Button ---
|
384 |
+
# Disable button if model failed to load
|
385 |
+
btn_generate = gr.Button("Generate Image", variant="primary", interactive=isinstance(image_generator_pipe, AutoPipelineForText2Image))
|
386 |
|
387 |
with gr.Column(scale=1):
|
388 |
# --- Outputs ---
|
389 |
out_prompt = gr.Textbox(label="Generated Prompt / Status", interactive=False, lines=5) # Show prompt or error status here
|
390 |
+
out_image = gr.Image(label="Generated Image", type="pil", show_label=True) # Ensure label is shown
|
391 |
|
392 |
# --- Event Handling ---
|
393 |
# Define inputs list carefully, handling potentially invisible audio input
|
|
|
395 |
if asr_pipeline:
|
396 |
inputs_list.append(inp_audio)
|
397 |
else:
|
398 |
+
inputs_list.append(inp_audio) # Pass the gr.State(None) placeholder
|
399 |
+
|
400 |
|
401 |
inputs_list.extend([inp_style, inp_quality, inp_neg_prompt, inp_guidance, inp_steps])
|
402 |
|
403 |
+
# Link button click to processing function
|
404 |
btn_generate.click(
|
405 |
fn=process_input,
|
406 |
inputs=inputs_list,
|
407 |
outputs=[out_prompt, out_image]
|
408 |
)
|
409 |
|
410 |
+
# Clear text input if audio is used (only if ASR is available)
|
411 |
if asr_pipeline:
|
412 |
+
def clear_text_on_audio_change(audio_data):
|
413 |
+
# Check if audio_data is not None or empty (depending on how Gradio signals recording)
|
414 |
+
if audio_data is not None:
|
415 |
+
print("Audio input detected, clearing text box.")
|
416 |
+
return "" # Clear text box
|
417 |
+
# If audio_data becomes None (e.g., recording cleared), don't clear text
|
418 |
+
return gr.update()
|
419 |
+
|
420 |
+
# .change event fires when the value changes, including becoming None if cleared
|
421 |
+
inp_audio.change(fn=clear_text_on_audio_change, inputs=inp_audio, outputs=inp_text, api_name="clear_text_on_audio")
|
422 |
|
423 |
|
424 |
# ---- Application Launch ----
|
425 |
if __name__ == "__main__":
|
426 |
+
# Final check before launch
|
427 |
+
if not isinstance(image_generator_pipe, AutoPipelineForText2Image):
|
428 |
+
print("\n" + "="*50)
|
429 |
+
print("CRITICAL WARNING:")
|
430 |
+
print(f"Image generation model ({model_id}) failed to load during startup.")
|
431 |
+
print("The Gradio UI will launch, but the 'Generate Image' button will be disabled.")
|
432 |
+
print("Check the logs above for the specific model loading error.")
|
433 |
+
print("="*50 + "\n")
|
434 |
+
|
435 |
|
436 |
# Launch the Gradio app
|
437 |
+
# Running on 0.0.0.0 is necessary for Hugging Face Spaces
|
438 |
+
demo.launch(share=False, server_name="0.0.0.0", server_port=7860)
|