Spaces:
Running
on
Zero
Running
on
Zero
Commit
•
b832af5
1
Parent(s):
702754c
Update app.py
Browse files
app.py
CHANGED
@@ -15,6 +15,27 @@ MODEL_REPO = "rain1011/pyramid-flow-sd3"
|
|
15 |
MODEL_VARIANT = "diffusion_transformer_768p"
|
16 |
MODEL_DTYPE = "bf16"
|
17 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
18 |
# Download and load the model
|
19 |
def load_model():
|
20 |
if not os.path.exists(MODEL_PATH):
|
@@ -67,12 +88,13 @@ def generate_video_from_image(image, prompt, duration, video_guidance_scale):
|
|
67 |
torch_dtype = torch.bfloat16 if MODEL_DTYPE == "bf16" else torch.float32
|
68 |
|
69 |
target_size = (1280, 720)
|
70 |
-
|
|
|
71 |
|
72 |
with torch.no_grad(), torch.cuda.amp.autocast(enabled=True, dtype=torch_dtype):
|
73 |
frames = model.generate_i2v(
|
74 |
prompt=prompt,
|
75 |
-
input_image=
|
76 |
num_inference_steps=[10, 10, 10],
|
77 |
temp=temp,
|
78 |
guidance_scale=7.0,
|
|
|
15 |
MODEL_VARIANT = "diffusion_transformer_768p"
|
16 |
MODEL_DTYPE = "bf16"
|
17 |
|
18 |
+
def center_crop(image, target_width, target_height):
|
19 |
+
width, height = image.size
|
20 |
+
aspect_ratio_target = target_width / target_height
|
21 |
+
aspect_ratio_image = width / height
|
22 |
+
|
23 |
+
if aspect_ratio_image > aspect_ratio_target:
|
24 |
+
# Crop the width (left and right)
|
25 |
+
new_width = int(height * aspect_ratio_target)
|
26 |
+
left = (width - new_width) // 2
|
27 |
+
right = left + new_width
|
28 |
+
top, bottom = 0, height
|
29 |
+
else:
|
30 |
+
# Crop the height (top and bottom)
|
31 |
+
new_height = int(width / aspect_ratio_target)
|
32 |
+
top = (height - new_height) // 2
|
33 |
+
bottom = top + new_height
|
34 |
+
left, right = 0, width
|
35 |
+
|
36 |
+
image = image.crop((left, top, right, bottom))
|
37 |
+
return image
|
38 |
+
|
39 |
# Download and load the model
|
40 |
def load_model():
|
41 |
if not os.path.exists(MODEL_PATH):
|
|
|
88 |
torch_dtype = torch.bfloat16 if MODEL_DTYPE == "bf16" else torch.float32
|
89 |
|
90 |
target_size = (1280, 720)
|
91 |
+
cropped_image = center_crop(image, 1280, 720)
|
92 |
+
resized_image = cropped_image.resize((1280, 720))
|
93 |
|
94 |
with torch.no_grad(), torch.cuda.amp.autocast(enabled=True, dtype=torch_dtype):
|
95 |
frames = model.generate_i2v(
|
96 |
prompt=prompt,
|
97 |
+
input_image=resized_image,
|
98 |
num_inference_steps=[10, 10, 10],
|
99 |
temp=temp,
|
100 |
guidance_scale=7.0,
|