Spaces:
Running
on
Zero
Running
on
Zero
Update app.py
#3
by
salso
- opened
app.py
CHANGED
@@ -10,20 +10,14 @@ from PIL import Image
|
|
10 |
BTEN_API_KEY = os.getenv("API_KEY")
|
11 |
URL = os.getenv("URL")
|
12 |
|
|
|
13 |
def image_to_base64(image: Image.Image) -> str:
|
14 |
-
"""Convert a PIL image to a base64-encoded PNG string."""
|
15 |
with io.BytesIO() as buffer:
|
16 |
image.save(buffer, format="PNG")
|
17 |
return base64.b64encode(buffer.getvalue()).decode("utf-8")
|
18 |
|
19 |
|
20 |
def ensure_image(img) -> Image.Image:
|
21 |
-
"""
|
22 |
-
Ensure the input is a PIL Image.
|
23 |
-
If it's already a PIL Image, return it.
|
24 |
-
If it's a string (file path), open it.
|
25 |
-
If it's a dict with a "name" key, open the file at that path.
|
26 |
-
"""
|
27 |
if isinstance(img, Image.Image):
|
28 |
return img
|
29 |
elif isinstance(img, str):
|
@@ -34,19 +28,7 @@ def ensure_image(img) -> Image.Image:
|
|
34 |
raise ValueError("Cannot convert input to a PIL Image.")
|
35 |
|
36 |
|
37 |
-
def call_baseten_generate(
|
38 |
-
image: Image.Image,
|
39 |
-
prompt: str,
|
40 |
-
steps: int,
|
41 |
-
strength: float,
|
42 |
-
height: int,
|
43 |
-
width: int,
|
44 |
-
lora_name: str,
|
45 |
-
remove_bg: bool,
|
46 |
-
) -> Image.Image | None:
|
47 |
-
"""
|
48 |
-
Call the Baseten /predict endpoint with provided parameters and return the generated image.
|
49 |
-
"""
|
50 |
image = ensure_image(image)
|
51 |
b64_image = image_to_base64(image)
|
52 |
payload = {
|
@@ -59,14 +41,10 @@ def call_baseten_generate(
|
|
59 |
"lora_name": lora_name,
|
60 |
"bgrm": remove_bg,
|
61 |
}
|
62 |
-
|
63 |
-
headers = {"Authorization": f"Api-Key {os.getenv('API_KEY')}"}
|
64 |
-
else:
|
65 |
-
headers = {"Authorization": f"Api-Key {BTEN_API_KEY}"}
|
66 |
try:
|
67 |
if not URL:
|
68 |
raise ValueError("The URL environment variable is not set.")
|
69 |
-
|
70 |
response = requests.post(URL, headers=headers, json=payload)
|
71 |
if response.status_code == 200:
|
72 |
data = response.json()
|
@@ -83,20 +61,17 @@ def call_baseten_generate(
|
|
83 |
return None
|
84 |
|
85 |
|
86 |
-
#
|
87 |
|
88 |
-
Mode = TypedDict(
|
89 |
-
"
|
90 |
-
|
91 |
-
|
92 |
-
|
93 |
-
|
94 |
-
|
95 |
-
|
96 |
-
|
97 |
-
"remove_bg": bool,
|
98 |
-
},
|
99 |
-
)
|
100 |
|
101 |
MODE_DEFAULTS: dict[str, Mode] = {
|
102 |
"Subject Generation": {
|
@@ -105,12 +80,7 @@ MODE_DEFAULTS: dict[str, Mode] = {
|
|
105 |
"default_strength": 1.2,
|
106 |
"default_height": 512,
|
107 |
"default_width": 512,
|
108 |
-
"models": [
|
109 |
-
"zendsd_512_146000",
|
110 |
-
"subject_99000_512",
|
111 |
-
# "zen_pers_11000",
|
112 |
-
"zen_26000_512",
|
113 |
-
],
|
114 |
"remove_bg": True,
|
115 |
},
|
116 |
"Background Generation": {
|
@@ -119,20 +89,7 @@ MODE_DEFAULTS: dict[str, Mode] = {
|
|
119 |
"default_strength": 1.2,
|
120 |
"default_height": 1024,
|
121 |
"default_width": 1024,
|
122 |
-
"models": [
|
123 |
-
"bgwlight_15000_1024",
|
124 |
-
# "rmgb_12000_1024",
|
125 |
-
"bg_canny_58000_1024",
|
126 |
-
# "gen_back_3000_1024",
|
127 |
-
"gen_back_7000_1024",
|
128 |
-
# "gen_bckgnd_18000_512",
|
129 |
-
# "gen_bckgnd_18000_512",
|
130 |
-
# "loose_25000_512",
|
131 |
-
# "looser_23000_1024",
|
132 |
-
# "looser_bg_gen_21000_1280",
|
133 |
-
# "old_looser_46000_1024",
|
134 |
-
# "relight_bg_gen_31000_1024",
|
135 |
-
],
|
136 |
"remove_bg": True,
|
137 |
},
|
138 |
"Canny": {
|
@@ -150,9 +107,7 @@ MODE_DEFAULTS: dict[str, Mode] = {
|
|
150 |
"default_strength": 1.2,
|
151 |
"default_height": 1024,
|
152 |
"default_width": 1024,
|
153 |
-
"models": [
|
154 |
-
"depth_9800_1024",
|
155 |
-
],
|
156 |
"remove_bg": True,
|
157 |
},
|
158 |
"Deblurring": {
|
@@ -161,63 +116,87 @@ MODE_DEFAULTS: dict[str, Mode] = {
|
|
161 |
"default_strength": 1.2,
|
162 |
"default_height": 1024,
|
163 |
"default_width": 1024,
|
164 |
-
"models": ["deblurr_1024_10000"],
|
165 |
"remove_bg": False,
|
166 |
},
|
167 |
}
|
168 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
169 |
|
170 |
header = """
|
171 |
<h1>π ZenCtrl / FLUX</h1>
|
172 |
<div align="center" style="line-height: 1;">
|
173 |
-
<a href="https://github.com/FotographerAI/ZenCtrl/tree/main" target="_blank"
|
174 |
-
<a href="https://huggingface.co/spaces/fotographerai/ZenCtrl" target="_blank"
|
175 |
-
<a href="https://discord.com/invite/b9RuYQ3F8k" target="_blank"
|
176 |
-
<a href="https://fotographer.ai/" target="_blank" style="margin: 2px;" name="lp_link"><img src="https://img.shields.io/badge/Website-Landing_Page-blue" alt="LP" style="display: inline-block; vertical-align: middle;"></a>
|
177 |
-
<a href="https://x.com/FotographerAI" target="_blank" style="margin: 2px;" name="twitter_link"><img src="https://img.shields.io/twitter/follow/FotographerAI?style=social" alt="X" style="display: inline-block; vertical-align: middle;"></a>
|
178 |
</div>
|
179 |
"""
|
180 |
|
181 |
-
defaults = MODE_DEFAULTS["Subject Generation"]
|
182 |
-
|
183 |
-
|
184 |
with gr.Blocks(title="π ZenCtrl") as demo:
|
185 |
gr.HTML(header)
|
186 |
-
gr.Markdown(
|
187 |
-
|
188 |
-
# ZenCtrl Demo
|
189 |
-
[WIP] One Agent to Generate multi-view, diverse-scene, and task-specific high-resolution images from a single subject imageβwithout fine-tuning.
|
190 |
-
We are first releasing some of the task specific weights and will release the codes soon.
|
191 |
-
The goal is to unify all of the visual content generation tasks with a single LLM...
|
192 |
-
|
193 |
-
**Modes:**
|
194 |
-
- **Subject Generation:** Focuses on generating detailed subject portraits.
|
195 |
-
- **Background Generation:** Creates dynamic, vibrant backgrounds:
|
196 |
-
You can generate part of the image from sketch while keeping part of it as it is.
|
197 |
-
- **Canny:** Emphasizes strong edge detection.
|
198 |
-
- **Depth:** Produces images with realistic depth and perspective.
|
199 |
-
|
200 |
-
For more details, shoot us a message on discord.
|
201 |
-
"""
|
202 |
-
)
|
203 |
with gr.Tabs():
|
204 |
for mode in MODE_DEFAULTS:
|
205 |
with gr.Tab(mode):
|
206 |
defaults = MODE_DEFAULTS[mode]
|
207 |
gr.Markdown(f"### {mode} Mode")
|
208 |
-
gr.Markdown(f"**Default Model:** {defaults['model']}")
|
209 |
|
210 |
with gr.Row():
|
211 |
-
with gr.Column(scale=2
|
212 |
-
input_image = gr.Image(
|
213 |
-
label="Upload Image",
|
214 |
-
type="pil",
|
215 |
-
scale=3,
|
216 |
-
height=370,
|
217 |
-
min_width=100,
|
218 |
-
)
|
219 |
generate_button = gr.Button("Generate")
|
220 |
-
with gr.Blocks(
|
221 |
model_dropdown = gr.Dropdown(
|
222 |
label="Model",
|
223 |
choices=defaults["models"],
|
@@ -229,20 +208,12 @@ with gr.Blocks(title="π ZenCtrl") as demo:
|
|
229 |
)
|
230 |
|
231 |
with gr.Column(scale=2):
|
232 |
-
output_image = gr.Image(
|
233 |
-
label="Generated Image",
|
234 |
-
type="pil",
|
235 |
-
height=573,
|
236 |
-
scale=4,
|
237 |
-
min_width=100,
|
238 |
-
)
|
239 |
|
240 |
-
gr.Markdown("#### Prompt")
|
241 |
prompt_box = gr.Textbox(
|
242 |
label="Prompt", value=defaults["prompt"], lines=2
|
243 |
)
|
244 |
|
245 |
-
# Wrap generation parameters in an Accordion for collapsible view.
|
246 |
with gr.Accordion("Generation Parameters", open=False):
|
247 |
with gr.Row():
|
248 |
step_slider = gr.Slider(
|
@@ -271,26 +242,8 @@ with gr.Blocks(title="π ZenCtrl") as demo:
|
|
271 |
label="Width",
|
272 |
)
|
273 |
|
274 |
-
def on_generate_click(
|
275 |
-
model_name,
|
276 |
-
prompt,
|
277 |
-
steps,
|
278 |
-
strength,
|
279 |
-
height,
|
280 |
-
width,
|
281 |
-
remove_bg,
|
282 |
-
image,
|
283 |
-
):
|
284 |
-
return call_baseten_generate(
|
285 |
-
image,
|
286 |
-
prompt,
|
287 |
-
steps,
|
288 |
-
strength,
|
289 |
-
height,
|
290 |
-
width,
|
291 |
-
model_name,
|
292 |
-
remove_bg,
|
293 |
-
)
|
294 |
|
295 |
generate_button.click(
|
296 |
fn=on_generate_click,
|
@@ -305,9 +258,16 @@ with gr.Blocks(title="π ZenCtrl") as demo:
|
|
305 |
input_image,
|
306 |
],
|
307 |
outputs=[output_image],
|
308 |
-
concurrency_limit=None
|
309 |
)
|
310 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
311 |
|
312 |
if __name__ == "__main__":
|
313 |
demo.launch()
|
|
|
10 |
BTEN_API_KEY = os.getenv("API_KEY")
|
11 |
URL = os.getenv("URL")
|
12 |
|
13 |
+
|
14 |
def image_to_base64(image: Image.Image) -> str:
|
|
|
15 |
with io.BytesIO() as buffer:
|
16 |
image.save(buffer, format="PNG")
|
17 |
return base64.b64encode(buffer.getvalue()).decode("utf-8")
|
18 |
|
19 |
|
20 |
def ensure_image(img) -> Image.Image:
|
|
|
|
|
|
|
|
|
|
|
|
|
21 |
if isinstance(img, Image.Image):
|
22 |
return img
|
23 |
elif isinstance(img, str):
|
|
|
28 |
raise ValueError("Cannot convert input to a PIL Image.")
|
29 |
|
30 |
|
31 |
+
def call_baseten_generate(image: Image.Image, prompt: str, steps: int, strength: float, height: int, width: int, lora_name: str, remove_bg: bool) -> Image.Image | None:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
32 |
image = ensure_image(image)
|
33 |
b64_image = image_to_base64(image)
|
34 |
payload = {
|
|
|
41 |
"lora_name": lora_name,
|
42 |
"bgrm": remove_bg,
|
43 |
}
|
44 |
+
headers = {"Authorization": f"Api-Key {BTEN_API_KEY or os.getenv('API_KEY')}"}
|
|
|
|
|
|
|
45 |
try:
|
46 |
if not URL:
|
47 |
raise ValueError("The URL environment variable is not set.")
|
|
|
48 |
response = requests.post(URL, headers=headers, json=payload)
|
49 |
if response.status_code == 200:
|
50 |
data = response.json()
|
|
|
61 |
return None
|
62 |
|
63 |
|
64 |
+
# ================== MODE CONFIG =====================
|
65 |
|
66 |
+
Mode = TypedDict("Mode", {
|
67 |
+
"model": str,
|
68 |
+
"prompt": str,
|
69 |
+
"default_strength": float,
|
70 |
+
"default_height": int,
|
71 |
+
"default_width": int,
|
72 |
+
"models": list[str],
|
73 |
+
"remove_bg": bool,
|
74 |
+
})
|
|
|
|
|
|
|
75 |
|
76 |
MODE_DEFAULTS: dict[str, Mode] = {
|
77 |
"Subject Generation": {
|
|
|
80 |
"default_strength": 1.2,
|
81 |
"default_height": 512,
|
82 |
"default_width": 512,
|
83 |
+
"models": ["zendsd_512_146000", "subject_99000_512", "zen_26000_512"],
|
|
|
|
|
|
|
|
|
|
|
84 |
"remove_bg": True,
|
85 |
},
|
86 |
"Background Generation": {
|
|
|
89 |
"default_strength": 1.2,
|
90 |
"default_height": 1024,
|
91 |
"default_width": 1024,
|
92 |
+
"models": ["bgwlight_15000_1024", "bg_canny_58000_1024", "gen_back_7000_1024"],
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
93 |
"remove_bg": True,
|
94 |
},
|
95 |
"Canny": {
|
|
|
107 |
"default_strength": 1.2,
|
108 |
"default_height": 1024,
|
109 |
"default_width": 1024,
|
110 |
+
"models": ["depth_9800_1024"],
|
|
|
|
|
111 |
"remove_bg": True,
|
112 |
},
|
113 |
"Deblurring": {
|
|
|
116 |
"default_strength": 1.2,
|
117 |
"default_height": 1024,
|
118 |
"default_width": 1024,
|
119 |
+
"models": ["deblurr_1024_10000"],
|
120 |
"remove_bg": False,
|
121 |
},
|
122 |
}
|
123 |
|
124 |
+
# ================== PRESET EXAMPLES =====================
|
125 |
+
|
126 |
+
MODE_EXAMPLES = {
|
127 |
+
"Subject Generation": [
|
128 |
+
["assets/subj1.jpg", "Close-up portrait of a fruit bowl", "assets/subj1_out.jpg"],
|
129 |
+
["assets/subj2.jpg", "A penguin standing in snow", "assets/subj2_out.jpg"],
|
130 |
+
["assets/subj3.jpg", "A cat with glowing eyes", "assets/subj3_out.jpg"],
|
131 |
+
["assets/subj4.jpg", "A child playing with bubbles", "assets/subj4_out.jpg"],
|
132 |
+
["assets/subj5.jpg", "A stylish young man in neon lights", "assets/subj5_out.jpg"],
|
133 |
+
["assets/subj6.jpg", "Old man with a mysterious look", "assets/subj6_out.jpg"],
|
134 |
+
],
|
135 |
+
|
136 |
+
"Background Generation": [
|
137 |
+
["assets/bg1.jpg", "Modern living room with plants", "assets/bg1_out.jpg"],
|
138 |
+
["assets/bg2.jpg", "Fantasy forest background", "assets/bg2_out.jpg"],
|
139 |
+
["assets/bg3.jpg", "Futuristic cityscape", "assets/bg3_out.jpg"],
|
140 |
+
["assets/bg4.jpg", "Minimalist white studio", "assets/bg4_out.jpg"],
|
141 |
+
["assets/bg5.jpg", "Snowy mountain landscape", "assets/bg5_out.jpg"],
|
142 |
+
["assets/bg6.jpg", "Golden sunset over the sea", "assets/bg6_out.jpg"],
|
143 |
+
],
|
144 |
+
|
145 |
+
"Canny": [
|
146 |
+
["assets/canny1.jpg", "A neon cyberpunk city skyline", "assets/canny1_out.jpg"],
|
147 |
+
["assets/canny2.jpg", "A robot walking in the fog", "assets/canny2_out.jpg"],
|
148 |
+
["assets/canny3.jpg", "A futuristic vehicle parked under a bridge", "assets/canny3_out.jpg"],
|
149 |
+
["assets/canny4.jpg", "Sci-fi lab interior with glowing machinery", "assets/canny4_out.jpg"],
|
150 |
+
["assets/canny5.jpg", "A portrait of a woman outlined in neon", "assets/canny5_out.jpg"],
|
151 |
+
["assets/canny6.jpg", "Post-apocalyptic abandoned street", "assets/canny6_out.jpg"],
|
152 |
+
],
|
153 |
+
|
154 |
+
"Depth": [
|
155 |
+
["assets/depth1.jpg", "A narrow alleyway with deep perspective", "assets/depth1_out.jpg"],
|
156 |
+
["assets/depth2.jpg", "A mountain road vanishing into the distance", "assets/depth2_out.jpg"],
|
157 |
+
["assets/depth3.jpg", "A hallway with strong depth of field", "assets/depth3_out.jpg"],
|
158 |
+
["assets/depth4.jpg", "A misty forest path stretching far away", "assets/depth4_out.jpg"],
|
159 |
+
["assets/depth5.jpg", "A bridge over a deep canyon", "assets/depth5_out.jpg"],
|
160 |
+
["assets/depth6.jpg", "An underground tunnel with receding arches", "assets/depth6_out.jpg"],
|
161 |
+
],
|
162 |
+
|
163 |
+
"Deblurring": [
|
164 |
+
["assets/deblur1.jpg", "", "assets/deblur1_out.jpg"],
|
165 |
+
["assets/deblur2.jpg", "", "assets/deblur2_out.jpg"],
|
166 |
+
["assets/deblur3.jpg", "", "assets/deblur3_out.jpg"],
|
167 |
+
["assets/deblur4.jpg", "", "assets/deblur4_out.jpg"],
|
168 |
+
["assets/deblur5.jpg", "", "assets/deblur5_out.jpg"],
|
169 |
+
["assets/deblur6.jpg", "", "assets/deblur6_out.jpg"],
|
170 |
+
],
|
171 |
+
}
|
172 |
+
|
173 |
+
|
174 |
+
# ================== UI =====================
|
175 |
|
176 |
header = """
|
177 |
<h1>π ZenCtrl / FLUX</h1>
|
178 |
<div align="center" style="line-height: 1;">
|
179 |
+
<a href="https://github.com/FotographerAI/ZenCtrl/tree/main" target="_blank"><img src="https://img.shields.io/badge/GitHub-Repo-181717.svg"></a>
|
180 |
+
<a href="https://huggingface.co/spaces/fotographerai/ZenCtrl" target="_blank"><img src="https://img.shields.io/badge/π€_HuggingFace-Space-ffbd45.svg"></a>
|
181 |
+
<a href="https://discord.com/invite/b9RuYQ3F8k" target="_blank"><img src="https://img.shields.io/badge/Discord-Join-7289da.svg?logo=discord"></a>
|
|
|
|
|
182 |
</div>
|
183 |
"""
|
184 |
|
|
|
|
|
|
|
185 |
with gr.Blocks(title="π ZenCtrl") as demo:
|
186 |
gr.HTML(header)
|
187 |
+
gr.Markdown("# ZenCtrl Demo")
|
188 |
+
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
189 |
with gr.Tabs():
|
190 |
for mode in MODE_DEFAULTS:
|
191 |
with gr.Tab(mode):
|
192 |
defaults = MODE_DEFAULTS[mode]
|
193 |
gr.Markdown(f"### {mode} Mode")
|
|
|
194 |
|
195 |
with gr.Row():
|
196 |
+
with gr.Column(scale=2):
|
197 |
+
input_image = gr.Image(label="Input Image", type="pil")
|
|
|
|
|
|
|
|
|
|
|
|
|
198 |
generate_button = gr.Button("Generate")
|
199 |
+
with gr.Blocks():
|
200 |
model_dropdown = gr.Dropdown(
|
201 |
label="Model",
|
202 |
choices=defaults["models"],
|
|
|
208 |
)
|
209 |
|
210 |
with gr.Column(scale=2):
|
211 |
+
output_image = gr.Image(label="Generated Image", type="pil")
|
|
|
|
|
|
|
|
|
|
|
|
|
212 |
|
|
|
213 |
prompt_box = gr.Textbox(
|
214 |
label="Prompt", value=defaults["prompt"], lines=2
|
215 |
)
|
216 |
|
|
|
217 |
with gr.Accordion("Generation Parameters", open=False):
|
218 |
with gr.Row():
|
219 |
step_slider = gr.Slider(
|
|
|
242 |
label="Width",
|
243 |
)
|
244 |
|
245 |
+
def on_generate_click(model_name, prompt, steps, strength, height, width, remove_bg, image):
|
246 |
+
return call_baseten_generate(image, prompt, steps, strength, height, width, model_name, remove_bg)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
247 |
|
248 |
generate_button.click(
|
249 |
fn=on_generate_click,
|
|
|
258 |
input_image,
|
259 |
],
|
260 |
outputs=[output_image],
|
|
|
261 |
)
|
262 |
|
263 |
+
# ---------------- Templates --------------------
|
264 |
+
gr.Dataset(
|
265 |
+
label="Presets (Input / Prompt / Output)",
|
266 |
+
headers=["Input", "Prompt", "Output"],
|
267 |
+
components=[input_image, prompt_box, output_image],
|
268 |
+
samples=MODE_EXAMPLES.get(mode, []),
|
269 |
+
samples_per_page=6,
|
270 |
+
)
|
271 |
|
272 |
if __name__ == "__main__":
|
273 |
demo.launch()
|