File size: 7,691 Bytes
537fd2c
 
 
 
 
 
 
 
62eacc2
537fd2c
 
 
 
 
93e795a
537fd2c
 
 
 
 
 
 
 
 
93e795a
 
537fd2c
 
 
93e795a
 
537fd2c
 
 
 
 
93e795a
537fd2c
 
 
93e795a
 
 
 
 
 
537fd2c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4304422
 
 
537fd2c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
93e795a
537fd2c
 
 
 
93e795a
537fd2c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
import os

import gradio as gr

from background_replacer import replace_background

developer_mode = os.getenv('DEV_MODE', False)

DEFAULT_POSITIVE_PROMPT = "on the pavement, poolside, idyllic infinity pool, Hawaiian hilltops, commercial product photography"
DEFAULT_NEGATIVE_PROMPT = ""

EXAMPLES = [
    [
        "examples/black-sneakers-with-white-sole.jpg",
        "on the grass in Central Park, gorgeous summer day with Bethesda fountain in the background, commercial footwear product photography",
        "people, litter, trash, crowds, messy",
    ],
    [
        "examples/DIY-beard-balm.jpg",
        "on a mossy rock, white wood anemone blossoms, Loch Ken, Scotland",
        "purple, wrong proportions",
    ],
    [
        "examples/dj-making-music-on-mixer.jpg",
        "on the turntables with a packed dance floor, epic midnight edm party in Miami Beach, colorful nightlife photography",
        "disfigured, dismembered, mangled, marred",
    ],
    [
        "examples/jean-shorts-woman.jpg",
        "on the beach in Malibu, a five-star beachfront hotel in the background, stark late afternoon light near the dunes, lifestyle photography",
        "blurry background, ripples, soft focus, bokeh",
    ],
]

INTRO = """
# SDXL Background Replacement for Product Images
_from your friends at_ 🛍️[Shopify](https://www.shopify.com/)

Building an online store requires lots of high quality product and marketing images. This is an early demo of a background replacement tool built with Stable Diffusion XL that makes it easy to use your existing product images to make something new. Please be patient during peak demand. 😅

To use it, upload your product photo (.jpg or .png), then describe the background you’d like to see in place of the original. For best results follow the general pattern in the examples below:
1. ❌ _Do not_ describe your product in the prompt (ex: black sneakers)
2. ✅ Do describe the "grounding" for your product (ex: placed on a table)
3. ✅ Do describe the scene you want (ex: in a greek cottage)
4. ✅ Do describe a style of image (ex: side view commercial product photography)
5. 🤔 Optionally, describe what you want to avoid 🙅 in the negative prompt field
"""

MORE_INFO = """
### More information
- You can check our [FAQs here](https://huggingface.co/spaces/Shopify/background-replacement/blob/main/README.md#faqs)!
- We are also gathering resources from the community and sharing ideas [here](https://huggingface.co/spaces/Shopify/background-replacement/discussions).
- Shopify is on a mission to redefine commerce with AI. If you’re an AI or ML engineer looking to build the future of commerce, [join us](https://www.shopify.com/careers)!
"""


def generate(
    image,
    positive_prompt,
    negative_prompt,
    seed,
    depth_map_feather_threshold,
    depth_map_dilation_iterations,
    depth_map_blur_radius,
    progress=gr.Progress(track_tqdm=True)
):
    if image is None:
        return [None, None, None, None]

    options = {
        'seed': seed,
        'depth_map_feather_threshold': depth_map_feather_threshold,
        'depth_map_dilation_iterations': depth_map_dilation_iterations,
        'depth_map_blur_radius': depth_map_blur_radius,
    }

    return replace_background(image, positive_prompt, negative_prompt, options)


custom_css = """
    #image-upload {
        flex-grow: 1;
    }
    #params .tabs {
        display: flex;
        flex-direction: column;
        flex-grow: 1;
    }
    #params .tabitem[style="display: block;"] {
        flex-grow: 1;
        display: flex !important;
    }
    #params .gap {
        flex-grow: 1;
    }
    #params .form {
        flex-grow: 1 !important;
    }
    #params .form > :last-child{
        flex-grow: 1;
    }
    .md ol, .md ul {
        margin-left: 1rem;
    }
"""

with gr.Blocks(css=custom_css) as iface:
    gr.Markdown(INTRO)

    with gr.Row():
        with gr.Column():
            image_upload = gr.Image(
                label="Product image",
                type="pil",
                elem_id="image-upload"
            )
            caption = gr.Label(
                label="Caption",
                visible=developer_mode
            )
        with gr.Column(elem_id="params"):
            with gr.Tab('Prompts'):
                positive_prompt = gr.Textbox(
                    label="Positive Prompt: describe what you'd like to see",
                    lines=3,
                    value=DEFAULT_POSITIVE_PROMPT
                )
                negative_prompt = gr.Textbox(
                    label="Negative Prompt: describe what you want to avoid",
                    lines=3,
                    value=DEFAULT_NEGATIVE_PROMPT
                )
            if developer_mode:
                with gr.Tab('Options'):
                    seed = gr.Number(
                        label="Seed",
                        precision=0,
                        value=0,
                        elem_id="seed",
                        visible=developer_mode
                    )
                    depth_map_feather_threshold = gr.Slider(
                        label="Depth map feather threshold",
                        value=128,
                        minimum=0,
                        maximum=255,
                        visible=developer_mode
                    )
                    depth_map_dilation_iterations = gr.Number(
                        label="Depth map dilation iterations",
                        precision=0,
                        value=10,
                        minimum=0,
                        visible=developer_mode
                    )
                    depth_map_blur_radius = gr.Number(
                        label="Depth map blur radius",
                        precision=0,
                        value=10,
                        minimum=0,
                        visible=developer_mode
                    )
            else:
                seed = gr.Number(value=-1, visible=False)
                depth_map_feather_threshold = gr.Slider(
                    value=128, visible=False)
                depth_map_dilation_iterations = gr.Number(
                    precision=0, value=10, visible=False)
                depth_map_blur_radius = gr.Number(
                    precision=0, value=10, visible=False)

    gen_button = gr.Button(value="Generate!", variant="primary")

    with gr.Tab('Results'):
        results = gr.Gallery(
            show_label=False,
            object_fit="contain",
            columns=4
        )

    if developer_mode:
        with gr.Tab('Generated'):
            generated = gr.Gallery(
                show_label=False,
                object_fit="contain",
                columns=4
            )

        with gr.Tab('Pre-processing'):
            pre_processing = gr.Gallery(
                show_label=False,
                object_fit="contain",
                columns=4
            )
    else:
        generated = gr.Gallery(visible=False)
        pre_processing = gr.Gallery(visible=False)

    gr.Examples(
        examples=EXAMPLES,
        inputs=[image_upload, positive_prompt, negative_prompt],
    )

    gr.Markdown(MORE_INFO)

    gen_button.click(
        fn=generate,
        inputs=[
            image_upload,
            positive_prompt,
            negative_prompt,
            seed,
            depth_map_feather_threshold,
            depth_map_dilation_iterations,
            depth_map_blur_radius
        ],
        outputs=[
            results,
            generated,
            pre_processing,
            caption
        ],
    )

iface.queue(api_open=False).launch(show_api=False)