File size: 4,246 Bytes
882e052
c8f1f54
4fe456a
61b9893
df1e443
 
c8f1f54
61b9893
df1e443
882e052
df1e443
 
 
 
a346948
 
 
df1e443
a346948
 
 
 
 
 
 
 
 
 
 
 
 
 
df1e443
 
 
61b9893
df1e443
61b9893
882e052
5d4b17c
61b9893
 
 
 
 
 
 
9aeab3c
61b9893
 
 
 
 
 
 
 
 
 
 
 
 
df1e443
c8f1f54
61b9893
882e052
 
087c578
 
df1e443
a346948
 
df1e443
 
 
 
 
a346948
61b9893
df1e443
 
 
 
61b9893
e3617aa
82c86ca
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
import os
import gradio as gr
from PIL import Image, ImageDraw, ImageFont
import io
import torch
from diffusers import DiffusionPipeline

# ===== CONFIGURATION =====
MODEL_NAME = "HiDream-ai/HiDream-I1-Full"
WATERMARK_TEXT = "SelamGPT"
DEVICE = "cuda" if torch.cuda.is_available() else "cpu"
TORCH_DTYPE = torch.float16 if DEVICE == "cuda" else torch.float32

# ===== MODEL LOADING =====
# Global variable for model caching (alternative to @gr.Cache)
pipe = None

def load_model():
    global pipe
    if pipe is None:
        pipe = DiffusionPipeline.from_pretrained(
            MODEL_NAME,
            torch_dtype=TORCH_DTYPE
        ).to(DEVICE)
        
        # Optimizations
        if DEVICE == "cuda":
            try:
                pipe.enable_xformers_memory_efficient_attention()
            except:
                print("Xformers not available, using default attention")
            pipe.enable_attention_slicing()
    
    return pipe

# ===== WATERMARK FUNCTION =====
def add_watermark(image):
    """Add watermark with optimized PNG output"""
    try:
        draw = ImageDraw.Draw(image)
        
        font_size = 24
        try:
            font = ImageFont.truetype("Roboto-Bold.ttf", font_size)
        except:
            font = ImageFont.load_default(font_size)
        
        text_width = draw.textlength(WATERMARK_TEXT, font=font)
        x = image.width - text_width - 10
        y = image.height - 34
        
        draw.text((x+1, y+1), WATERMARK_TEXT, font=font, fill=(0, 0, 0, 128))
        draw.text((x, y), WATERMARK_TEXT, font=font, fill=(255, 255, 255))
        
        # Convert to optimized PNG
        img_byte_arr = io.BytesIO()
        image.save(img_byte_arr, format='PNG', optimize=True, quality=85)
        img_byte_arr.seek(0)
        return Image.open(img_byte_arr)
    except Exception as e:
        print(f"Watermark error: {str(e)}")
        return image

# ===== IMAGE GENERATION =====
def generate_image(prompt):
    if not prompt.strip():
        return None, "⚠️ Please enter a prompt"
    
    try:
        model = load_model()
        image = model(
            prompt,
            num_inference_steps=30,
            guidance_scale=7.5
        ).images[0]
        
        return add_watermark(image), "✔️ Generation successful"
    
    except torch.cuda.OutOfMemoryError:
        return None, "⚠️ Out of memory! Try a simpler prompt"
    except Exception as e:
        return None, f"⚠️ Error: {str(e)[:200]}"

# ===== GRADIO INTERFACE =====
with gr.Blocks(theme=theme, title="SelamGPT Image Generator") as demo:
    gr.Markdown("""
    # 🎨 SelamGPT Image Generator
    *Powered by HiDream-I1-Full (1024x1024 PNG output)*
    """)
    
    with gr.Row():
        with gr.Column(scale=3):
            prompt_input = gr.Textbox(
                label="Describe your image",
                placeholder="A futuristic Ethiopian city with flying cars...",
                lines=3,
                max_lines=5
            )
            with gr.Row():
                generate_btn = gr.Button("Generate Image", variant="primary")
                clear_btn = gr.Button("Clear")
            
            gr.Examples(
                examples=[
                    ["An ancient Aksumite warrior in cyberpunk armor, 4k detailed"],
                    ["Traditional Ethiopian coffee ceremony in zero gravity"],
                    ["Portrait of a Habesha queen with golden jewelry"]
                ],
                inputs=prompt_input
            )
            
        with gr.Column(scale=2):
            output_image = gr.Image(
                label="Generated Image",
                type="pil",
                format="png",
                height=512
            )
            status_output = gr.Textbox(
                label="Status",
                interactive=False
            )
    
    generate_btn.click(
        fn=generate_image,
        inputs=prompt_input,
        outputs=[output_image, status_output],
        queue=True
    )
    
    clear_btn.click(
        fn=lambda: [None, ""],
        outputs=[output_image, status_output]
    )

if __name__ == "__main__":
    demo.queue(max_size=2)
    demo.launch(server_name="0.0.0.0", server_port=7860)