Spaces:
Running
on
Zero
Running
on
Zero
Dimensions
Browse files
app.py
CHANGED
@@ -37,6 +37,15 @@ FIXED_FPS = 24
|
|
37 |
MIN_FRAMES_MODEL = 8
|
38 |
MAX_FRAMES_MODEL = 121
|
39 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
40 |
# Instantiate the pipeline in the global scope
|
41 |
print("Initializing WanTI2V pipeline...")
|
42 |
device = "cuda" if torch.cuda.is_available() else "cpu"
|
@@ -55,44 +64,52 @@ pipeline = wan.WanTI2V(
|
|
55 |
)
|
56 |
print("Pipeline initialized and ready.")
|
57 |
|
58 |
-
# --- Helper Functions ---
|
59 |
-
def
|
60 |
-
|
61 |
-
|
62 |
-
|
63 |
-
|
64 |
-
|
65 |
-
|
66 |
-
|
67 |
-
|
68 |
-
best_diff = float('inf')
|
69 |
-
|
70 |
-
for size_str in available_sizes:
|
71 |
-
# Parse size string like "704*1280"
|
72 |
-
height, width = map(int, size_str.split('*'))
|
73 |
-
size_aspect_ratio = height / width
|
74 |
-
diff = abs(img_aspect_ratio - size_aspect_ratio)
|
75 |
-
|
76 |
-
if diff < best_diff:
|
77 |
-
best_diff = diff
|
78 |
-
best_size = size_str
|
79 |
|
80 |
-
|
|
|
81 |
|
82 |
-
|
83 |
-
|
84 |
-
if image is None:
|
85 |
-
return gr.update()
|
86 |
|
87 |
-
|
88 |
-
|
89 |
-
best_size = select_best_size_for_image(pil_image, available_sizes)
|
90 |
|
91 |
-
return
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
92 |
|
93 |
def get_duration(image,
|
94 |
prompt,
|
95 |
-
|
|
|
96 |
duration_seconds,
|
97 |
sampling_steps,
|
98 |
guide_scale,
|
@@ -112,7 +129,8 @@ def get_duration(image,
|
|
112 |
def generate_video(
|
113 |
image,
|
114 |
prompt,
|
115 |
-
|
|
|
116 |
duration_seconds,
|
117 |
sampling_steps,
|
118 |
guide_scale,
|
@@ -124,21 +142,27 @@ def generate_video(
|
|
124 |
if seed == -1:
|
125 |
seed = random.randint(0, sys.maxsize)
|
126 |
|
|
|
|
|
|
|
|
|
127 |
input_image = None
|
128 |
if image is not None:
|
129 |
input_image = Image.fromarray(image).convert("RGB")
|
130 |
-
# Resize image to match
|
131 |
-
|
132 |
-
input_image = input_image.resize((target_width, target_height))
|
133 |
|
134 |
# Calculate number of frames based on duration
|
135 |
num_frames = np.clip(int(round(duration_seconds * FIXED_FPS)), MIN_FRAMES_MODEL, MAX_FRAMES_MODEL)
|
136 |
|
|
|
|
|
|
|
137 |
video_tensor = pipeline.generate(
|
138 |
input_prompt=prompt,
|
139 |
img=input_image, # Pass None for T2V, Image for I2V
|
140 |
-
size=SIZE_CONFIGS
|
141 |
-
max_area=MAX_AREA_CONFIGS
|
142 |
frame_num=num_frames, # Use calculated frames instead of cfg.frame_num
|
143 |
shift=shift,
|
144 |
sample_solver='unipc',
|
@@ -170,7 +194,7 @@ with gr.Blocks(css=css, theme=gr.themes.Soft(), delete_cache=(60, 900)) as demo:
|
|
170 |
|
171 |
with gr.Row():
|
172 |
with gr.Column(scale=2):
|
173 |
-
image_input = gr.Image(type="numpy", label="Input Image (Optional)", elem_id="input_image")
|
174 |
prompt_input = gr.Textbox(label="Prompt", value="A beautiful waterfall in a lush jungle, cinematic.", lines=3)
|
175 |
duration_input = gr.Slider(
|
176 |
minimum=round(MIN_FRAMES_MODEL/FIXED_FPS, 1),
|
@@ -180,40 +204,41 @@ with gr.Blocks(css=css, theme=gr.themes.Soft(), delete_cache=(60, 900)) as demo:
|
|
180 |
label="Duration (seconds)",
|
181 |
info=f"Clamped to model's {MIN_FRAMES_MODEL}-{MAX_FRAMES_MODEL} frames at {FIXED_FPS}fps."
|
182 |
)
|
183 |
-
size_input = gr.Dropdown(label="Output Resolution", choices=list(SUPPORTED_SIZES[TASK_NAME]), value="704*1280")
|
184 |
-
with gr.Column(scale=2):
|
185 |
-
video_output = gr.Video(label="Generated Video", elem_id="output_video")
|
186 |
|
187 |
-
|
188 |
with gr.Accordion("Advanced Settings", open=False):
|
|
|
|
|
|
|
189 |
steps_input = gr.Slider(label="Sampling Steps", minimum=10, maximum=50, value=38, step=1)
|
190 |
scale_input = gr.Slider(label="Guidance Scale", minimum=1.0, maximum=10.0, value=cfg.sample_guide_scale, step=0.1)
|
191 |
shift_input = gr.Slider(label="Sample Shift", minimum=1.0, maximum=20.0, value=cfg.sample_shift, step=0.1)
|
192 |
seed_input = gr.Number(label="Seed (-1 for random)", value=-1, precision=0)
|
193 |
|
|
|
|
|
194 |
run_button = gr.Button("Generate Video", variant="primary")
|
195 |
|
196 |
# Add image upload handler
|
197 |
image_input.upload(
|
198 |
-
fn=
|
199 |
-
inputs=[image_input],
|
200 |
-
outputs=[
|
201 |
)
|
202 |
|
203 |
image_input.clear(
|
204 |
-
fn=
|
205 |
-
inputs=[image_input],
|
206 |
-
outputs=[
|
207 |
)
|
208 |
|
209 |
example_image_path = os.path.join(os.path.dirname(__file__), "examples/i2v_input.JPG")
|
210 |
gr.Examples(
|
211 |
examples=[
|
212 |
-
[example_image_path, "The cat removes the glasses from its eyes.",
|
213 |
-
[None, "A cinematic shot of a boat sailing on a calm sea at sunset.",
|
214 |
-
[None, "Drone footage flying over a futuristic city with flying cars.",
|
215 |
],
|
216 |
-
inputs=[image_input, prompt_input,
|
217 |
outputs=video_output,
|
218 |
fn=generate_video,
|
219 |
cache_examples=False,
|
@@ -221,7 +246,7 @@ with gr.Blocks(css=css, theme=gr.themes.Soft(), delete_cache=(60, 900)) as demo:
|
|
221 |
|
222 |
run_button.click(
|
223 |
fn=generate_video,
|
224 |
-
inputs=[image_input, prompt_input,
|
225 |
outputs=video_output
|
226 |
)
|
227 |
|
|
|
37 |
MIN_FRAMES_MODEL = 8
|
38 |
MAX_FRAMES_MODEL = 121
|
39 |
|
40 |
+
# Dimension calculation constants (from Wan 2.1 Fast demo)
|
41 |
+
MOD_VALUE = 32
|
42 |
+
DEFAULT_H_SLIDER_VALUE = 512
|
43 |
+
DEFAULT_W_SLIDER_VALUE = 896
|
44 |
+
NEW_FORMULA_MAX_AREA = 480.0 * 832.0
|
45 |
+
|
46 |
+
SLIDER_MIN_H, SLIDER_MAX_H = 128, 896
|
47 |
+
SLIDER_MIN_W, SLIDER_MAX_W = 128, 896
|
48 |
+
|
49 |
# Instantiate the pipeline in the global scope
|
50 |
print("Initializing WanTI2V pipeline...")
|
51 |
device = "cuda" if torch.cuda.is_available() else "cpu"
|
|
|
64 |
)
|
65 |
print("Pipeline initialized and ready.")
|
66 |
|
67 |
+
# --- Helper Functions (from Wan 2.1 Fast demo) ---
|
68 |
+
def _calculate_new_dimensions_wan(pil_image, mod_val, calculation_max_area,
|
69 |
+
min_slider_h, max_slider_h,
|
70 |
+
min_slider_w, max_slider_w,
|
71 |
+
default_h, default_w):
|
72 |
+
orig_w, orig_h = pil_image.size
|
73 |
+
if orig_w <= 0 or orig_h <= 0:
|
74 |
+
return default_h, default_w
|
75 |
+
|
76 |
+
aspect_ratio = orig_h / orig_w
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
77 |
|
78 |
+
calc_h = round(np.sqrt(calculation_max_area * aspect_ratio))
|
79 |
+
calc_w = round(np.sqrt(calculation_max_area / aspect_ratio))
|
80 |
|
81 |
+
calc_h = max(mod_val, (calc_h // mod_val) * mod_val)
|
82 |
+
calc_w = max(mod_val, (calc_w // mod_val) * mod_val)
|
|
|
|
|
83 |
|
84 |
+
new_h = int(np.clip(calc_h, min_slider_h, (max_slider_h // mod_val) * mod_val))
|
85 |
+
new_w = int(np.clip(calc_w, min_slider_w, (max_slider_w // mod_val) * mod_val))
|
|
|
86 |
|
87 |
+
return new_h, new_w
|
88 |
+
|
89 |
+
def handle_image_upload_for_dims_wan(uploaded_pil_image, current_h_val, current_w_val):
|
90 |
+
if uploaded_pil_image is None:
|
91 |
+
return gr.update(value=DEFAULT_H_SLIDER_VALUE), gr.update(value=DEFAULT_W_SLIDER_VALUE)
|
92 |
+
try:
|
93 |
+
# Convert numpy array to PIL Image if needed
|
94 |
+
if hasattr(uploaded_pil_image, 'shape'): # numpy array
|
95 |
+
pil_image = Image.fromarray(uploaded_pil_image).convert("RGB")
|
96 |
+
else: # already PIL Image
|
97 |
+
pil_image = uploaded_pil_image
|
98 |
+
|
99 |
+
new_h, new_w = _calculate_new_dimensions_wan(
|
100 |
+
pil_image, MOD_VALUE, NEW_FORMULA_MAX_AREA,
|
101 |
+
SLIDER_MIN_H, SLIDER_MAX_H, SLIDER_MIN_W, SLIDER_MAX_W,
|
102 |
+
DEFAULT_H_SLIDER_VALUE, DEFAULT_W_SLIDER_VALUE
|
103 |
+
)
|
104 |
+
return gr.update(value=new_h), gr.update(value=new_w)
|
105 |
+
except Exception as e:
|
106 |
+
gr.Warning("Error attempting to calculate new dimensions")
|
107 |
+
return gr.update(value=DEFAULT_H_SLIDER_VALUE), gr.update(value=DEFAULT_W_SLIDER_VALUE)
|
108 |
|
109 |
def get_duration(image,
|
110 |
prompt,
|
111 |
+
height,
|
112 |
+
width,
|
113 |
duration_seconds,
|
114 |
sampling_steps,
|
115 |
guide_scale,
|
|
|
129 |
def generate_video(
|
130 |
image,
|
131 |
prompt,
|
132 |
+
height,
|
133 |
+
width,
|
134 |
duration_seconds,
|
135 |
sampling_steps,
|
136 |
guide_scale,
|
|
|
142 |
if seed == -1:
|
143 |
seed = random.randint(0, sys.maxsize)
|
144 |
|
145 |
+
# Ensure dimensions are multiples of MOD_VALUE
|
146 |
+
target_h = max(MOD_VALUE, (int(height) // MOD_VALUE) * MOD_VALUE)
|
147 |
+
target_w = max(MOD_VALUE, (int(width) // MOD_VALUE) * MOD_VALUE)
|
148 |
+
|
149 |
input_image = None
|
150 |
if image is not None:
|
151 |
input_image = Image.fromarray(image).convert("RGB")
|
152 |
+
# Resize image to match target dimensions
|
153 |
+
input_image = input_image.resize((target_w, target_h))
|
|
|
154 |
|
155 |
# Calculate number of frames based on duration
|
156 |
num_frames = np.clip(int(round(duration_seconds * FIXED_FPS)), MIN_FRAMES_MODEL, MAX_FRAMES_MODEL)
|
157 |
|
158 |
+
# Create size string for the pipeline
|
159 |
+
size_str = f"{target_h}*{target_w}"
|
160 |
+
|
161 |
video_tensor = pipeline.generate(
|
162 |
input_prompt=prompt,
|
163 |
img=input_image, # Pass None for T2V, Image for I2V
|
164 |
+
size=SIZE_CONFIGS.get(size_str, (target_h, target_w)),
|
165 |
+
max_area=MAX_AREA_CONFIGS.get(size_str, target_h * target_w),
|
166 |
frame_num=num_frames, # Use calculated frames instead of cfg.frame_num
|
167 |
shift=shift,
|
168 |
sample_solver='unipc',
|
|
|
194 |
|
195 |
with gr.Row():
|
196 |
with gr.Column(scale=2):
|
197 |
+
image_input = gr.Image(type="numpy", label="Input Image (Optional, auto-resized to target H/W)", elem_id="input_image")
|
198 |
prompt_input = gr.Textbox(label="Prompt", value="A beautiful waterfall in a lush jungle, cinematic.", lines=3)
|
199 |
duration_input = gr.Slider(
|
200 |
minimum=round(MIN_FRAMES_MODEL/FIXED_FPS, 1),
|
|
|
204 |
label="Duration (seconds)",
|
205 |
info=f"Clamped to model's {MIN_FRAMES_MODEL}-{MAX_FRAMES_MODEL} frames at {FIXED_FPS}fps."
|
206 |
)
|
|
|
|
|
|
|
207 |
|
|
|
208 |
with gr.Accordion("Advanced Settings", open=False):
|
209 |
+
with gr.Row():
|
210 |
+
height_input = gr.Slider(minimum=SLIDER_MIN_H, maximum=SLIDER_MAX_H, step=MOD_VALUE, value=DEFAULT_H_SLIDER_VALUE, label=f"Output Height (multiple of {MOD_VALUE})")
|
211 |
+
width_input = gr.Slider(minimum=SLIDER_MIN_W, maximum=SLIDER_MAX_W, step=MOD_VALUE, value=DEFAULT_W_SLIDER_VALUE, label=f"Output Width (multiple of {MOD_VALUE})")
|
212 |
steps_input = gr.Slider(label="Sampling Steps", minimum=10, maximum=50, value=38, step=1)
|
213 |
scale_input = gr.Slider(label="Guidance Scale", minimum=1.0, maximum=10.0, value=cfg.sample_guide_scale, step=0.1)
|
214 |
shift_input = gr.Slider(label="Sample Shift", minimum=1.0, maximum=20.0, value=cfg.sample_shift, step=0.1)
|
215 |
seed_input = gr.Number(label="Seed (-1 for random)", value=-1, precision=0)
|
216 |
|
217 |
+
with gr.Column(scale=2):
|
218 |
+
video_output = gr.Video(label="Generated Video", elem_id="output_video")
|
219 |
run_button = gr.Button("Generate Video", variant="primary")
|
220 |
|
221 |
# Add image upload handler
|
222 |
image_input.upload(
|
223 |
+
fn=handle_image_upload_for_dims_wan,
|
224 |
+
inputs=[image_input, height_input, width_input],
|
225 |
+
outputs=[height_input, width_input]
|
226 |
)
|
227 |
|
228 |
image_input.clear(
|
229 |
+
fn=handle_image_upload_for_dims_wan,
|
230 |
+
inputs=[image_input, height_input, width_input],
|
231 |
+
outputs=[height_input, width_input]
|
232 |
)
|
233 |
|
234 |
example_image_path = os.path.join(os.path.dirname(__file__), "examples/i2v_input.JPG")
|
235 |
gr.Examples(
|
236 |
examples=[
|
237 |
+
[example_image_path, "The cat removes the glasses from its eyes.", 704, 1280, 1.5],
|
238 |
+
[None, "A cinematic shot of a boat sailing on a calm sea at sunset.", 704, 1280, 2.0],
|
239 |
+
[None, "Drone footage flying over a futuristic city with flying cars.", 704, 1280, 2.0],
|
240 |
],
|
241 |
+
inputs=[image_input, prompt_input, height_input, width_input, duration_input],
|
242 |
outputs=video_output,
|
243 |
fn=generate_video,
|
244 |
cache_examples=False,
|
|
|
246 |
|
247 |
run_button.click(
|
248 |
fn=generate_video,
|
249 |
+
inputs=[image_input, prompt_input, height_input, width_input, duration_input, steps_input, scale_input, shift_input, seed_input],
|
250 |
outputs=video_output
|
251 |
)
|
252 |
|