Spaces:
Running
on
Zero
Running
on
Zero
MohamedRashad
commited on
Commit
·
836dd96
1
Parent(s):
c0ec201
Add prompt enhancement functionality and integrate Gradio client in app.py; update requirements.txt
Browse files- app.py +68 -5
- requirements.txt +3 -1
app.py
CHANGED
@@ -13,6 +13,7 @@ import re
|
|
13 |
import random
|
14 |
from pathlib import Path
|
15 |
from typing import List
|
|
|
16 |
|
17 |
import cv2
|
18 |
import numpy as np
|
@@ -29,8 +30,10 @@ import spaces
|
|
29 |
from models.infinity import Infinity
|
30 |
from models.basic import *
|
31 |
from utils.dynamic_resolution import dynamic_resolution_h_w, h_div_w_templates
|
|
|
32 |
|
33 |
torch._dynamo.config.cache_size_limit = 64
|
|
|
34 |
|
35 |
# Define a function to download weights if not present
|
36 |
def download_infinity_weights(weights_path):
|
@@ -357,6 +360,60 @@ def load_transformer(vae, args):
|
|
357 |
)
|
358 |
return infinity
|
359 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
360 |
# Set up paths
|
361 |
weights_path = Path(__file__).parent / 'weights'
|
362 |
weights_path.mkdir(exist_ok=True)
|
@@ -380,7 +437,6 @@ args = argparse.Namespace(
|
|
380 |
rope2d_normalized_by_hw=2,
|
381 |
use_scale_schedule_embedding=0,
|
382 |
sampling_per_bits=1,
|
383 |
-
text_encoder_ckpt=str(weights_path / 'flan-t5-xl'),
|
384 |
text_channels=2048,
|
385 |
apply_spatial_patchify=0,
|
386 |
h_div_w_template=1.000,
|
@@ -400,7 +456,7 @@ infinity = load_transformer(vae, args)
|
|
400 |
|
401 |
# Define the image generation function
|
402 |
@spaces.GPU
|
403 |
-
def generate_image(prompt, cfg, tau, h_div_w, seed, enable_positive_prompt):
|
404 |
try:
|
405 |
args.prompt = prompt
|
406 |
args.cfg = cfg
|
@@ -454,8 +510,8 @@ with gr.Blocks() as demo:
|
|
454 |
# Prompt Settings
|
455 |
gr.Markdown("### Prompt Settings")
|
456 |
prompt = gr.Textbox(label="Prompt", value="alien spaceship enterprise", placeholder="Enter your prompt here...")
|
457 |
-
|
458 |
-
|
459 |
# Image Settings
|
460 |
gr.Markdown("### Image Settings")
|
461 |
with gr.Row():
|
@@ -477,10 +533,17 @@ with gr.Blocks() as demo:
|
|
477 |
# Error Handling
|
478 |
error_message = gr.Textbox(label="Error Message", visible=False)
|
479 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
480 |
# Link the generate button to the image generation function
|
481 |
generate_button.click(
|
482 |
generate_image,
|
483 |
-
inputs=[prompt, cfg, tau, h_div_w, seed
|
484 |
outputs=output_image
|
485 |
)
|
486 |
|
|
|
13 |
import random
|
14 |
from pathlib import Path
|
15 |
from typing import List
|
16 |
+
import json
|
17 |
|
18 |
import cv2
|
19 |
import numpy as np
|
|
|
30 |
from models.infinity import Infinity
|
31 |
from models.basic import *
|
32 |
from utils.dynamic_resolution import dynamic_resolution_h_w, h_div_w_templates
|
33 |
+
from gradio_client import Client
|
34 |
|
35 |
torch._dynamo.config.cache_size_limit = 64
|
36 |
+
client = Client("Qwen/Qwen2.5-72B-Instruct")
|
37 |
|
38 |
# Define a function to download weights if not present
|
39 |
def download_infinity_weights(weights_path):
|
|
|
360 |
)
|
361 |
return infinity
|
362 |
|
363 |
+
def enhance_prompt(prompt):
|
364 |
+
SYSTEM = """You are part of a team of bots that creates images. You work with an assistant bot that will draw anything you say.
|
365 |
+
|
366 |
+
When given a user prompt, your role is to transform it into a creative, detailed, and vivid image description. Additionally, you will assign a configuration value (`cfg`) based on the type of image.
|
367 |
+
|
368 |
+
### Guidelines for Generating the Output:
|
369 |
+
|
370 |
+
1. **Output Format:**
|
371 |
+
Your response must be in the following dictionary format:
|
372 |
+
```json
|
373 |
+
{
|
374 |
+
"prompt": "<enhanced image description>",
|
375 |
+
"cfg": <cfg value>
|
376 |
+
}
|
377 |
+
```
|
378 |
+
|
379 |
+
2. **Enhancing the "prompt" field:**
|
380 |
+
- Use your creativity to transform short or vague prompts into highly detailed, descriptive, and imaginative image descriptions.
|
381 |
+
- Preserve the original intent and meaning of the user’s input.
|
382 |
+
- Focus on vivid imagery, sensory details, and emotional resonance in your descriptions.
|
383 |
+
- For particularly long user prompts (over 50 words), output them directly without refinement.
|
384 |
+
- Image descriptions must remain between 8-512 words. Any excess text will be ignored.
|
385 |
+
- If the user's request involves rendering specific text in the image, enclose that text in single quotation marks and prefix it with "the text".
|
386 |
+
|
387 |
+
3. **Determining the "cfg" field:**
|
388 |
+
- If the image to be generated is likely to feature a clear face, set `"cfg": 1`.
|
389 |
+
- If the image does not prominently feature a face, set `"cfg": 3`.
|
390 |
+
|
391 |
+
4. **Examples of Enhanced Prompts:**
|
392 |
+
- **User prompt:** "a tree"
|
393 |
+
**Enhanced prompt:** "A photo of a majestic oak tree stands proudly in the middle of a sunlit meadow, its branches stretching out like welcoming arms. The leaves shimmer in shades of vibrant green, casting dappled shadows on the soft grass below."
|
394 |
+
**Cfg:** `3`
|
395 |
+
|
396 |
+
- **User prompt:** "a cat by the window"
|
397 |
+
**Enhanced prompt:** "A serene scene of a fluffy tabby cat perched on the windowsill, gazing out at the golden hues of a sunset. The soft light filters through lace curtains, highlighting the cat’s delicate whiskers and its relaxed posture."
|
398 |
+
**Cfg:** `3`
|
399 |
+
|
400 |
+
5. **Your Output:**
|
401 |
+
Always return a single dictionary containing both `"prompt"` and `"cfg"` fields. Avoid any additional commentary or explanations.
|
402 |
+
|
403 |
+
Don't write anything except the dictionary in the output. (Don't start with ```)
|
404 |
+
"""
|
405 |
+
result = client.predict(
|
406 |
+
query=prompt,
|
407 |
+
history=[],
|
408 |
+
system=SYSTEM,
|
409 |
+
api_name="/model_chat"
|
410 |
+
)
|
411 |
+
|
412 |
+
dict_of_inputs = json.loads(result[1][-1][-1])
|
413 |
+
print(dict_of_inputs)
|
414 |
+
|
415 |
+
return gr.update(value=dict_of_inputs["prompt"]), gr.update(value=float(dict_of_inputs['cfg']))
|
416 |
+
|
417 |
# Set up paths
|
418 |
weights_path = Path(__file__).parent / 'weights'
|
419 |
weights_path.mkdir(exist_ok=True)
|
|
|
437 |
rope2d_normalized_by_hw=2,
|
438 |
use_scale_schedule_embedding=0,
|
439 |
sampling_per_bits=1,
|
|
|
440 |
text_channels=2048,
|
441 |
apply_spatial_patchify=0,
|
442 |
h_div_w_template=1.000,
|
|
|
456 |
|
457 |
# Define the image generation function
|
458 |
@spaces.GPU
|
459 |
+
def generate_image(prompt, cfg, tau, h_div_w, seed, enable_positive_prompt=False):
|
460 |
try:
|
461 |
args.prompt = prompt
|
462 |
args.cfg = cfg
|
|
|
510 |
# Prompt Settings
|
511 |
gr.Markdown("### Prompt Settings")
|
512 |
prompt = gr.Textbox(label="Prompt", value="alien spaceship enterprise", placeholder="Enter your prompt here...")
|
513 |
+
enhance_prompt_button = gr.Button("Enhance Prompt", variant="secondary")
|
514 |
+
|
515 |
# Image Settings
|
516 |
gr.Markdown("### Image Settings")
|
517 |
with gr.Row():
|
|
|
533 |
# Error Handling
|
534 |
error_message = gr.Textbox(label="Error Message", visible=False)
|
535 |
|
536 |
+
# Link the enhance prompt button to the prompt enhancement function
|
537 |
+
enhance_prompt_button.click(
|
538 |
+
enhance_prompt,
|
539 |
+
inputs=prompt,
|
540 |
+
outputs=[prompt, cfg],
|
541 |
+
)
|
542 |
+
|
543 |
# Link the generate button to the image generation function
|
544 |
generate_button.click(
|
545 |
generate_image,
|
546 |
+
inputs=[prompt, cfg, tau, h_div_w, seed],
|
547 |
outputs=output_image
|
548 |
)
|
549 |
|
requirements.txt
CHANGED
@@ -6,4 +6,6 @@ transformers
|
|
6 |
argparse
|
7 |
spaces
|
8 |
torchvision
|
9 |
-
timm
|
|
|
|
|
|
6 |
argparse
|
7 |
spaces
|
8 |
torchvision
|
9 |
+
timm
|
10 |
+
gradio_client
|
11 |
+
imageio
|