File size: 6,552 Bytes
cab0202
 
17e4caa
cab0202
 
 
 
 
 
 
 
 
 
 
 
 
 
3fc4e02
77b6bae
cab0202
3fc4e02
3decda6
17e4caa
3decda6
 
 
 
 
 
cab0202
 
 
 
3fc4e02
4411eb3
cab0202
 
 
 
 
21de9e6
cab0202
 
 
 
 
 
 
 
 
 
 
21de9e6
cab0202
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
5f0f463
cab0202
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
import gradio as gr
import torch
from diffusers import StableDiffusionPipeline,DiffusionPipeline

from diffusion_webui.utils.model_list import stable_model_list
from diffusion_webui.utils.scheduler_list import (
    SCHEDULER_MAPPING,
    get_scheduler,
)


class StableDiffusionText2ImageGenerator:
    def __init__(self):
        self.pipe = None

    def load_model(
        self,
        stable_model_path,
        scheduler,
    ):
        if self.pipe is None or self.pipe.model_name != stable_model_path or self.pipe.scheduler_name != scheduler:
            if stable_model_path == "stabilityai/stable-diffusion-xl-base-0.9":
                self.pipe = DiffusionPipeline.from_pretrained(
                    stable_model_path, safety_checker=None, torch_dtype=torch.float16
                )
            else:
                self.pipe = StableDiffusionPipeline.from_pretrained(
                    stable_model_path, safety_checker=None, torch_dtype=torch.float16
                )

        self.pipe = get_scheduler(pipe=self.pipe, scheduler=scheduler)
        self.pipe.to("cuda")
        self.pipe.enable_xformers_memory_efficient_attention()
        self.pipe.model_name = stable_model_path
        self.pipe.scheduler_name = scheduler

        return self.pipe

    def generate_image(
        self,
        stable_model_path: str,
        prompt: str,
        negative_prompt: str,
        num_images_per_prompt: int,
        scheduler: str,
        guidance_scale: int,
        num_inference_step: int,
        height: int,
        width: int,
        seed_generator=0,
    ):
        pipe = self.load_model(
            stable_model_path=stable_model_path,
            scheduler=scheduler,
        )
        if seed_generator == 0:
            random_seed = torch.randint(0, 1000000, (1,))
            generator = torch.manual_seed(random_seed)
        else:
            generator = torch.manual_seed(seed_generator)

        images = pipe(
            prompt=prompt,
            height=height,
            width=width,
            negative_prompt=negative_prompt,
            num_images_per_prompt=num_images_per_prompt,
            num_inference_steps=num_inference_step,
            guidance_scale=guidance_scale,
            generator=generator,
        ).images

        return images

    def app():
        with gr.Blocks():
            with gr.Row():
                with gr.Column():
                    text2image_prompt = gr.Textbox(
                        lines=1,
                        placeholder="Prompt",
                        show_label=False,
                    )

                    text2image_negative_prompt = gr.Textbox(
                        lines=1,
                        placeholder="Negative Prompt",
                        show_label=False,
                    )
                    with gr.Row():
                        with gr.Column():
                            text2image_model_path = gr.Dropdown(
                                choices=stable_model_list,
                                value=stable_model_list[0],
                                label="Text-Image Model Id",
                            )

                            text2image_guidance_scale = gr.Slider(
                                minimum=0.1,
                                maximum=15,
                                step=0.1,
                                value=7.5,
                                label="Guidance Scale",
                            )

                            text2image_num_inference_step = gr.Slider(
                                minimum=1,
                                maximum=100,
                                step=1,
                                value=50,
                                label="Num Inference Step",
                            )
                            text2image_num_images_per_prompt = gr.Slider(
                                minimum=1,
                                maximum=4,
                                step=1,
                                value=1,
                                label="Number Of Images",
                            )
                        with gr.Row():
                            with gr.Column():
                                text2image_scheduler = gr.Dropdown(
                                    choices=list(SCHEDULER_MAPPING.keys()),
                                    value=list(SCHEDULER_MAPPING.keys())[0],
                                    label="Scheduler",
                            )

                                text2image_height = gr.Slider(
                                    minimum=128,
                                    maximum=1280,
                                    step=32,
                                    value=512,
                                    label="Image Height",
                                )

                                text2image_width = gr.Slider(
                                    minimum=128,
                                    maximum=1280,
                                    step=32,
                                    value=512,
                                    label="Image Width",
                                )
                                text2image_seed_generator = gr.Slider(
                                    label="Seed(0 for random)",
                                    minimum=0,
                                    maximum=1000000,
                                    value=0,
                                )
                    text2image_predict = gr.Button(value="Generator")

                with gr.Column():
                    output_image = gr.Gallery(
                        label="Generated images",
                        show_label=False,
                        elem_id="gallery",
                    ).style(grid=(1, 2), height=200)

            text2image_predict.click(
                fn=StableDiffusionText2ImageGenerator().generate_image,
                inputs=[
                    text2image_model_path,
                    text2image_prompt,
                    text2image_negative_prompt,
                    text2image_num_images_per_prompt,
                    text2image_scheduler,
                    text2image_guidance_scale,
                    text2image_num_inference_step,
                    text2image_height,
                    text2image_width,
                    text2image_seed_generator,
                ],
                outputs=output_image,
            )