Spaces:
Running
on
L40S
Running
on
L40S
Update app.py
Browse files
app.py
CHANGED
@@ -19,11 +19,50 @@ from trellis.representations import Gaussian, MeshExtractResult
|
|
19 |
from trellis.utils import render_utils, postprocessing_utils
|
20 |
from diffusers import FluxPipeline
|
21 |
from typing import Tuple, Dict, Any # Tuple import ์ถ๊ฐ
|
22 |
-
|
23 |
-
# ํ์ผ ์๋จ์ import ๋ฌธ ์์
|
24 |
import transformers
|
25 |
from transformers import pipeline as transformers_pipeline
|
26 |
-
from transformers import Pipeline
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
27 |
# CUDA ๋ฉ๋ชจ๋ฆฌ ๊ด๋ฆฌ ์ค์
|
28 |
torch.cuda.empty_cache()
|
29 |
torch.backends.cuda.matmul.allow_tf32 = True
|
@@ -71,7 +110,7 @@ class timer:
|
|
71 |
|
72 |
def preprocess_image(image: Image.Image) -> Tuple[str, Image.Image]:
|
73 |
trial_id = str(uuid.uuid4())
|
74 |
-
processed_image =
|
75 |
processed_image.save(f"{TMP_DIR}/{trial_id}.png")
|
76 |
return trial_id, processed_image
|
77 |
|
@@ -169,7 +208,7 @@ def text_to_image(prompt: str, height: int, width: int, steps: int, scales: floa
|
|
169 |
|
170 |
# ํ๋กฌํํธ ์ ์ฒ๋ฆฌ
|
171 |
if contains_korean(prompt):
|
172 |
-
translated = translator(prompt)[0]['translation_text']
|
173 |
prompt = translated
|
174 |
|
175 |
# ํ๋กฌํํธ ํ์ ๊ฐ์
|
@@ -177,7 +216,7 @@ def text_to_image(prompt: str, height: int, width: int, steps: int, scales: floa
|
|
177 |
|
178 |
with torch.inference_mode(), torch.autocast("cuda", dtype=torch.bfloat16):
|
179 |
try:
|
180 |
-
generated_image = flux_pipe(
|
181 |
prompt=[formatted_prompt],
|
182 |
generator=torch.Generator().manual_seed(int(seed)),
|
183 |
num_inference_steps=int(steps),
|
@@ -330,35 +369,8 @@ if __name__ == "__main__":
|
|
330 |
print(f"Using device: {device}")
|
331 |
|
332 |
try:
|
333 |
-
#
|
334 |
-
|
335 |
-
"JeffreyXiang/TRELLIS-image-large"
|
336 |
-
)
|
337 |
-
trellis_pipeline.to(device)
|
338 |
-
|
339 |
-
# ์ด๋ฏธ์ง ์์ฑ ํ์ดํ๋ผ์ธ
|
340 |
-
flux_pipe = FluxPipeline.from_pretrained(
|
341 |
-
"black-forest-labs/FLUX.1-dev",
|
342 |
-
torch_dtype=torch.bfloat16,
|
343 |
-
device_map="balanced"
|
344 |
-
)
|
345 |
-
|
346 |
-
# Hyper-SD LoRA ๋ก๋
|
347 |
-
lora_path = hf_hub_download(
|
348 |
-
"ByteDance/Hyper-SD",
|
349 |
-
"Hyper-FLUX.1-dev-8steps-lora.safetensors",
|
350 |
-
use_auth_token=HF_TOKEN
|
351 |
-
)
|
352 |
-
flux_pipe.load_lora_weights(lora_path)
|
353 |
-
flux_pipe.fuse_lora(lora_scale=0.125)
|
354 |
-
|
355 |
-
# ๋ฒ์ญ๊ธฐ ์ด๊ธฐํ
|
356 |
-
global translator
|
357 |
-
translator = transformers_pipeline(
|
358 |
-
"translation",
|
359 |
-
model="Helsinki-NLP/opus-mt-ko-en",
|
360 |
-
device=device
|
361 |
-
)
|
362 |
|
363 |
# CUDA ๋ฉ๋ชจ๋ฆฌ ์ด๊ธฐํ
|
364 |
if torch.cuda.is_available():
|
@@ -367,7 +379,7 @@ if __name__ == "__main__":
|
|
367 |
# ์ด๊ธฐ ์ด๋ฏธ์ง ์ ์ฒ๋ฆฌ ํ
์คํธ
|
368 |
try:
|
369 |
test_image = Image.fromarray(np.zeros((512, 512, 3), dtype=np.uint8))
|
370 |
-
trellis_pipeline.preprocess_image(test_image)
|
371 |
except Exception as e:
|
372 |
print(f"Warning: Initial preprocessing test failed: {e}")
|
373 |
|
|
|
19 |
from trellis.utils import render_utils, postprocessing_utils
|
20 |
from diffusers import FluxPipeline
|
21 |
from typing import Tuple, Dict, Any # Tuple import ์ถ๊ฐ
|
22 |
+
# ํ์ผ ์๋จ์ import ๋ฌธ
|
|
|
23 |
import transformers
|
24 |
from transformers import pipeline as transformers_pipeline
|
25 |
+
from transformers import Pipeline
|
26 |
+
|
27 |
+
# ์ ์ญ ๋ณ์ ์ด๊ธฐํ
|
28 |
+
class GlobalVars:
|
29 |
+
def __init__(self):
|
30 |
+
self.translator = None
|
31 |
+
self.trellis_pipeline = None
|
32 |
+
self.flux_pipe = None
|
33 |
+
|
34 |
+
g = GlobalVars()
|
35 |
+
|
36 |
+
def initialize_models(device):
|
37 |
+
# 3D ์์ฑ ํ์ดํ๋ผ์ธ
|
38 |
+
g.trellis_pipeline = TrellisImageTo3DPipeline.from_pretrained(
|
39 |
+
"JeffreyXiang/TRELLIS-image-large"
|
40 |
+
)
|
41 |
+
g.trellis_pipeline.to(device)
|
42 |
+
|
43 |
+
# ์ด๋ฏธ์ง ์์ฑ ํ์ดํ๋ผ์ธ
|
44 |
+
g.flux_pipe = FluxPipeline.from_pretrained(
|
45 |
+
"black-forest-labs/FLUX.1-dev",
|
46 |
+
torch_dtype=torch.bfloat16,
|
47 |
+
device_map="balanced"
|
48 |
+
)
|
49 |
+
|
50 |
+
# Hyper-SD LoRA ๋ก๋
|
51 |
+
lora_path = hf_hub_download(
|
52 |
+
"ByteDance/Hyper-SD",
|
53 |
+
"Hyper-FLUX.1-dev-8steps-lora.safetensors",
|
54 |
+
use_auth_token=HF_TOKEN
|
55 |
+
)
|
56 |
+
g.flux_pipe.load_lora_weights(lora_path)
|
57 |
+
g.flux_pipe.fuse_lora(lora_scale=0.125)
|
58 |
+
|
59 |
+
# ๋ฒ์ญ๊ธฐ ์ด๊ธฐํ
|
60 |
+
g.translator = transformers_pipeline(
|
61 |
+
"translation",
|
62 |
+
model="Helsinki-NLP/opus-mt-ko-en",
|
63 |
+
device=device
|
64 |
+
)
|
65 |
+
|
66 |
# CUDA ๋ฉ๋ชจ๋ฆฌ ๊ด๋ฆฌ ์ค์
|
67 |
torch.cuda.empty_cache()
|
68 |
torch.backends.cuda.matmul.allow_tf32 = True
|
|
|
110 |
|
111 |
def preprocess_image(image: Image.Image) -> Tuple[str, Image.Image]:
|
112 |
trial_id = str(uuid.uuid4())
|
113 |
+
processed_image = g.trellis_pipeline.preprocess_image(image)
|
114 |
processed_image.save(f"{TMP_DIR}/{trial_id}.png")
|
115 |
return trial_id, processed_image
|
116 |
|
|
|
208 |
|
209 |
# ํ๋กฌํํธ ์ ์ฒ๋ฆฌ
|
210 |
if contains_korean(prompt):
|
211 |
+
translated = g.translator(prompt)[0]['translation_text']
|
212 |
prompt = translated
|
213 |
|
214 |
# ํ๋กฌํํธ ํ์ ๊ฐ์
|
|
|
216 |
|
217 |
with torch.inference_mode(), torch.autocast("cuda", dtype=torch.bfloat16):
|
218 |
try:
|
219 |
+
generated_image = g.flux_pipe(
|
220 |
prompt=[formatted_prompt],
|
221 |
generator=torch.Generator().manual_seed(int(seed)),
|
222 |
num_inference_steps=int(steps),
|
|
|
369 |
print(f"Using device: {device}")
|
370 |
|
371 |
try:
|
372 |
+
# ๋ชจ๋ธ ์ด๊ธฐํ
|
373 |
+
initialize_models(device)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
374 |
|
375 |
# CUDA ๋ฉ๋ชจ๋ฆฌ ์ด๊ธฐํ
|
376 |
if torch.cuda.is_available():
|
|
|
379 |
# ์ด๊ธฐ ์ด๋ฏธ์ง ์ ์ฒ๋ฆฌ ํ
์คํธ
|
380 |
try:
|
381 |
test_image = Image.fromarray(np.zeros((512, 512, 3), dtype=np.uint8))
|
382 |
+
g.trellis_pipeline.preprocess_image(test_image)
|
383 |
except Exception as e:
|
384 |
print(f"Warning: Initial preprocessing test failed: {e}")
|
385 |
|