File size: 6,276 Bytes
7733cf5
9772569
 
 
 
7e583c2
9772569
 
 
 
 
ffbb513
9772569
ffbb513
 
ec0e4d3
bf9239a
7e583c2
 
 
9772569
 
 
bf9239a
 
 
9772569
bf9239a
9772569
 
 
 
 
 
 
 
bf9239a
 
 
 
 
 
 
2f2cbf0
bf9239a
67720c3
1589ab9
 
9772569
bf9239a
9772569
 
 
 
 
 
 
2a2d6e3
9772569
 
 
1589ab9
 
 
2a2d6e3
9772569
 
 
 
 
 
ffbb513
9772569
ffbb513
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
9772569
ffbb513
9772569
b0c6316
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
9a88fb8
b0c6316
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
import os
import spaces
import gradio as gr
import torch
from diffusers import FluxPipeline, FluxTransformer2DModel, FlowMatchEulerDiscreteScheduler
from huggingface_hub import hf_hub_download, login
from PIL import Image
import requests
from translatepy import Translator
import numpy as np
import random
import logging

# Setup logging
logging.basicConfig(level=logging.DEBUG)

# Get HF_TOKEN from environment
HF_TOKEN = os.getenv("HF_TOKEN")
login(token=HF_TOKEN)

translator = Translator()

# Constants
MODEL_NAME = "black-forest-labs/FLUX.1-dev"
LORA_WEIGHT_NAME = "MegaTronX/CivitAI_Flux_LoRA-SuicideGirlsv1"
WEIGHT_FILE = "SuicideGirlsv1.safetensors"

# CSS for hiding the footer
CSS = """
footer {
    visibility: hidden;
}
"""

MAX_SEED = np.iinfo(np.int32).max

# Load the main model and LoRA weights
def load_model():
    pipe = FluxPipeline.from_pretrained(MODEL_NAME, torch_dtype=torch.bfloat16)
    pipe.load_lora_weights(LORA_WEIGHT_NAME, weight_name=WEIGHT_FILE)
    pipe.fuse_lora(lora_scale=0.8)
    pipe.to("cuda")
    return pipe

pipe = load_model()

def clear_gallery():
    return None

# Function to generate images
@spaces.GPU()
def generate_image(
    prompt,
    width=768,
    height=1024,
    scale=3.5,
    steps=24,
    seed=42,
    nums=1,
    progress=gr.Progress(track_tqdm=True)
):
    # Clears the gallery 
    yield None, seed
    
    if seed == 42:
        seed = random.randint(0, MAX_SEED)
    seed = int(seed)

    generator = torch.Generator().manual_seed(seed)
    
    prompt = str(translator.translate(prompt, 'English'))
    logging.debug(f"Translated Prompt: {prompt}")

    try:
        images = pipe(
            prompt, 
            width=width,
            height=height,
            guidance_scale=scale,
            num_inference_steps=steps,
            generator=generator,
            output_type="pil",
            max_sequence_length=512,
            num_images_per_prompt=nums,
        ).images

        logging.debug(f"Generated Images: {images}")

        # Save the first image as a test
        images[0].save("test_image.png")
        logging.info(f"Image saved to test_image.png")
        
    except Exception as e:
        logging.error(f"Error during image generation: {e}")
    
    return images, seed

examples = [
    "SuicideGirl with black hair and green eyes",
    "close up portrait, Amidst the interplay of light and shadows in a photography studio, a soft spotlight traces the contours of a face, highlighting a figure clad in a sleek black turtleneck. The garment, hugging the skin with subtle luxury, complements the Caucasian model's understated makeup, embodying minimalist elegance. Behind, a pale gray backdrop extends, its fine texture shimmering subtly in the dim light, artfully balancing the composition and focusing attention on the subject. In a palette of black, gray, and skin tones, simplicity intertwines with profundity, as every detail whispers untold stories.",
    "Caucasian, The image features a young woman of European descent standing in an studio setting, surrounded by silk. (She is wearing a silk dress), paired with a bold. Her brown hair is wet and tousled, falling naturally around her face, giving her a raw and edgy look. The woman has an intense and direct gaze, adding to the dramatic feel of the image. The backdrop is flowing silk, big silk. The overall composition blends elements of fashion and nature, creating a striking and powerful visual",
    "A black and white portrait of a young woman with a captivating gaze. She's bundled up in a cozy black sweater, hands gently cupped near her face. The monochromatic tones highlight her delicate features and the contemplative mood of the image",
    "Fashion photography portrait, close up portrait, (a woman of European descent is surrounded by lava rock and magma from head to neck, red magma hair, wear volcanic lava rock magma outfit coat lava rock magma fashion costume with ruffled layers"
]

# Gradio Interface
with gr.Blocks(css=CSS, theme="ocean") as demo:
    with gr.Group():
        with gr.Row():
            prompt = gr.Textbox(label='Enter Your Prompt (multilingual)', scale=6)
            submit = gr.Button(scale=1, variant='primary')
    img = gr.Gallery(label="Gallery", columns=1, preview=True, height=600)
    with gr.Accordion("Advanced Options", open=False):
        with gr.Row():
            width = gr.Slider(
                label="Width",
                minimum=512,
                maximum=1280,
                step=8,
                value=768,
            )
            height = gr.Slider(
                label="Height",
                minimum=512,
                maximum=1280,
                step=8,
                value=1024,
            )
        with gr.Row():
            scale = gr.Slider(
                label="Guidance Scale",
                minimum=0,
                maximum=50,
                step=0.1,
                value=3.0,
            )
            steps = gr.Slider(
                label="Steps",
                minimum=1,
                maximum=50,
                step=1,
                value=28,
            )
        with gr.Row():
            seed = gr.Slider(
                label="Seed (-1 Random)",
                minimum=-1,
                maximum=MAX_SEED,
                step=1,
                value=0,
                visible=True
            )
            nums = gr.Slider(
                label="Image Numbers",
                minimum=1,
                maximum=4,
                step=1,
                value=1,
                scale=1,
            )
    gr.Examples(
        examples=examples,
        inputs=prompt,
        outputs=[img, seed],
        fn=generate_image,
        cache_examples=True,
        cache_mode='lazy'
    )

    gr.on(
        triggers=[
            prompt.submit,
            submit.click,
        ],
        fn=generate_image,
        inputs=[
            prompt, 
            width, 
            height, 
            scale, 
            steps, 
            seed, 
            nums
        ],
        outputs=[img, seed],
        api_name="run",
    )

    # Clears the gallery before generating a new image
    submit.click(fn=clear_gallery, outputs=img, queue=False)
    
demo.queue().launch()