Spaces:
Running
on
L40S
Running
on
L40S
Update app.py
Browse files
app.py
CHANGED
@@ -55,36 +55,39 @@ Other times the user will not want modifications , but instead want a new image
|
|
55 |
Video descriptions must have the same num of words as examples below. Extra words will be ignored.
|
56 |
"""
|
57 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
58 |
def get_video_dimensions(input_video_path):
|
59 |
reader = imageio_ffmpeg.read_frames(input_video_path)
|
60 |
metadata = next(reader)
|
61 |
return metadata['size']
|
62 |
|
63 |
def center_crop_resize(input_video_path, target_width=720, target_height=480):
|
64 |
-
# Open the video file
|
65 |
cap = cv2.VideoCapture(input_video_path)
|
66 |
|
67 |
-
# Get original video properties
|
68 |
orig_width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
|
69 |
orig_height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
|
70 |
orig_fps = cap.get(cv2.CAP_PROP_FPS)
|
71 |
total_frames = int(cap.get(cv2.CAP_PROP_FRAME_COUNT))
|
72 |
|
73 |
-
# Calculate resize factor
|
74 |
width_factor = target_width / orig_width
|
75 |
height_factor = target_height / orig_height
|
76 |
resize_factor = max(width_factor, height_factor)
|
77 |
|
78 |
-
# Calculate intermediate size
|
79 |
inter_width = int(orig_width * resize_factor)
|
80 |
inter_height = int(orig_height * resize_factor)
|
81 |
|
82 |
-
# Calculate frame skip
|
83 |
target_fps = 8
|
84 |
ideal_skip = max(0, math.ceil(orig_fps / target_fps) - 1)
|
85 |
skip = min(5, ideal_skip) # Cap at 5
|
86 |
|
87 |
-
# Adjust skip if not enough frames
|
88 |
while (total_frames / (skip + 1)) < 49 and skip > 0:
|
89 |
skip -= 1
|
90 |
|
@@ -98,10 +101,8 @@ def center_crop_resize(input_video_path, target_width=720, target_height=480):
|
|
98 |
break
|
99 |
|
100 |
if total_read % (skip + 1) == 0:
|
101 |
-
# Resize frame
|
102 |
resized = cv2.resize(frame, (inter_width, inter_height), interpolation=cv2.INTER_AREA)
|
103 |
|
104 |
-
# Center crop
|
105 |
start_x = (inter_width - target_width) // 2
|
106 |
start_y = (inter_height - target_height) // 2
|
107 |
cropped = resized[start_y:start_y+target_height, start_x:start_x+target_width]
|
@@ -113,7 +114,6 @@ def center_crop_resize(input_video_path, target_width=720, target_height=480):
|
|
113 |
|
114 |
cap.release()
|
115 |
|
116 |
-
# Save the processed video to a temporary file
|
117 |
with tempfile.NamedTemporaryFile(suffix='.mp4', delete=False) as temp_file:
|
118 |
temp_video_path = temp_file.name
|
119 |
fourcc = cv2.VideoWriter_fourcc(*'mp4v')
|
@@ -188,13 +188,12 @@ def infer(
|
|
188 |
seed = random.randint(0, 2 ** 8 - 1)
|
189 |
if(video_input):
|
190 |
video = load_video(video_input)[:49] # Limit to 49 frames
|
191 |
-
video_pt =
|
192 |
video=video,
|
193 |
prompt=prompt,
|
194 |
num_inference_steps=num_inference_steps,
|
195 |
num_videos_per_prompt=1,
|
196 |
strength=video_strenght,
|
197 |
-
num_frames=49,
|
198 |
use_dynamic_cfg=True,
|
199 |
output_type="pt",
|
200 |
guidance_scale=guidance_scale,
|
@@ -241,7 +240,7 @@ def delete_old_files():
|
|
241 |
|
242 |
|
243 |
threading.Thread(target=delete_old_files, daemon=True).start()
|
244 |
-
examples = [["horse.mp4", "
|
245 |
|
246 |
with gr.Blocks() as demo:
|
247 |
gr.Markdown("""
|
@@ -265,12 +264,11 @@ with gr.Blocks() as demo:
|
|
265 |
|
266 |
""")
|
267 |
with gr.Row():
|
268 |
-
with gr.Accordion("Video-to-video", open=False):
|
269 |
-
video_input = gr.Video(label="Input Video (will be cropped to 49 frames, 6 seconds at 8fps)")
|
270 |
-
strength = gr.Slider(0.1, 1.0, value=0.8, step=0.01, label="Strength")
|
271 |
-
examples_component = gr.Examples(examples, fn=process_video, inputs=[input_video, prompt], outputs=output_video, cache_examples="lazy")
|
272 |
-
examples_component.dataset._components = [input_video]
|
273 |
with gr.Column():
|
|
|
|
|
|
|
|
|
274 |
prompt = gr.Textbox(label="Prompt (Less than 200 Words)", placeholder="Enter your prompt here", lines=5)
|
275 |
|
276 |
with gr.Row():
|
@@ -366,14 +364,18 @@ with gr.Blocks() as demo:
|
|
366 |
|
367 |
|
368 |
def generate(prompt,
|
|
|
|
|
369 |
seed_value,
|
370 |
scale_status,
|
371 |
rife_status,
|
372 |
-
progress=gr.Progress(track_tqdm=True)
|
373 |
):
|
374 |
|
375 |
latents, seed = infer(
|
376 |
prompt,
|
|
|
|
|
377 |
num_inference_steps=50, # NOT Changed
|
378 |
guidance_scale=7.0, # NOT Changed
|
379 |
seed=seed_value,
|
@@ -409,17 +411,17 @@ with gr.Blocks() as demo:
|
|
409 |
|
410 |
generate_button.click(
|
411 |
generate,
|
412 |
-
inputs=[prompt, seed_param, enable_scale, enable_rife],
|
413 |
outputs=[video_output, download_video_button, download_gif_button, seed_text],
|
414 |
)
|
415 |
|
416 |
enhance_button.click(enhance_prompt_func, inputs=[prompt], outputs=[prompt])
|
417 |
|
418 |
-
|
419 |
resize_if_unfit,
|
420 |
-
inputs=[
|
421 |
-
outputs=[
|
422 |
)
|
423 |
if __name__ == "__main__":
|
424 |
demo.queue(max_size=15)
|
425 |
-
demo.launch()
|
|
|
55 |
Video descriptions must have the same num of words as examples below. Extra words will be ignored.
|
56 |
"""
|
57 |
|
58 |
+
def resize_if_unfit(input_video, progress=gr.Progress(track_tqdm=True)):
|
59 |
+
width, height = get_video_dimensions(input_video)
|
60 |
+
|
61 |
+
if width == 720 and height == 480:
|
62 |
+
processed_video = input_video
|
63 |
+
else:
|
64 |
+
processed_video = center_crop_resize(input_video)
|
65 |
+
return processed_video
|
66 |
+
|
67 |
def get_video_dimensions(input_video_path):
|
68 |
reader = imageio_ffmpeg.read_frames(input_video_path)
|
69 |
metadata = next(reader)
|
70 |
return metadata['size']
|
71 |
|
72 |
def center_crop_resize(input_video_path, target_width=720, target_height=480):
|
|
|
73 |
cap = cv2.VideoCapture(input_video_path)
|
74 |
|
|
|
75 |
orig_width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
|
76 |
orig_height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
|
77 |
orig_fps = cap.get(cv2.CAP_PROP_FPS)
|
78 |
total_frames = int(cap.get(cv2.CAP_PROP_FRAME_COUNT))
|
79 |
|
|
|
80 |
width_factor = target_width / orig_width
|
81 |
height_factor = target_height / orig_height
|
82 |
resize_factor = max(width_factor, height_factor)
|
83 |
|
|
|
84 |
inter_width = int(orig_width * resize_factor)
|
85 |
inter_height = int(orig_height * resize_factor)
|
86 |
|
|
|
87 |
target_fps = 8
|
88 |
ideal_skip = max(0, math.ceil(orig_fps / target_fps) - 1)
|
89 |
skip = min(5, ideal_skip) # Cap at 5
|
90 |
|
|
|
91 |
while (total_frames / (skip + 1)) < 49 and skip > 0:
|
92 |
skip -= 1
|
93 |
|
|
|
101 |
break
|
102 |
|
103 |
if total_read % (skip + 1) == 0:
|
|
|
104 |
resized = cv2.resize(frame, (inter_width, inter_height), interpolation=cv2.INTER_AREA)
|
105 |
|
|
|
106 |
start_x = (inter_width - target_width) // 2
|
107 |
start_y = (inter_height - target_height) // 2
|
108 |
cropped = resized[start_y:start_y+target_height, start_x:start_x+target_width]
|
|
|
114 |
|
115 |
cap.release()
|
116 |
|
|
|
117 |
with tempfile.NamedTemporaryFile(suffix='.mp4', delete=False) as temp_file:
|
118 |
temp_video_path = temp_file.name
|
119 |
fourcc = cv2.VideoWriter_fourcc(*'mp4v')
|
|
|
188 |
seed = random.randint(0, 2 ** 8 - 1)
|
189 |
if(video_input):
|
190 |
video = load_video(video_input)[:49] # Limit to 49 frames
|
191 |
+
video_pt = pipe_video(
|
192 |
video=video,
|
193 |
prompt=prompt,
|
194 |
num_inference_steps=num_inference_steps,
|
195 |
num_videos_per_prompt=1,
|
196 |
strength=video_strenght,
|
|
|
197 |
use_dynamic_cfg=True,
|
198 |
output_type="pt",
|
199 |
guidance_scale=guidance_scale,
|
|
|
240 |
|
241 |
|
242 |
threading.Thread(target=delete_old_files, daemon=True).start()
|
243 |
+
examples = [["horse.mp4"], ["kitten.mp4"], ["train_running.mp4"]]
|
244 |
|
245 |
with gr.Blocks() as demo:
|
246 |
gr.Markdown("""
|
|
|
264 |
|
265 |
""")
|
266 |
with gr.Row():
|
|
|
|
|
|
|
|
|
|
|
267 |
with gr.Column():
|
268 |
+
with gr.Accordion("Video-to-video", open=False):
|
269 |
+
video_input = gr.Video(label="Input Video (will be cropped to 49 frames, 6 seconds at 8fps)")
|
270 |
+
strength = gr.Slider(0.1, 1.0, value=0.8, step=0.01, label="Strength")
|
271 |
+
examples_component = gr.Examples(examples, inputs=[video_input], cache_examples=False)
|
272 |
prompt = gr.Textbox(label="Prompt (Less than 200 Words)", placeholder="Enter your prompt here", lines=5)
|
273 |
|
274 |
with gr.Row():
|
|
|
364 |
|
365 |
|
366 |
def generate(prompt,
|
367 |
+
video_input,
|
368 |
+
video_strenght,
|
369 |
seed_value,
|
370 |
scale_status,
|
371 |
rife_status,
|
372 |
+
#progress=gr.Progress(track_tqdm=True)
|
373 |
):
|
374 |
|
375 |
latents, seed = infer(
|
376 |
prompt,
|
377 |
+
video_input,
|
378 |
+
video_strenght,
|
379 |
num_inference_steps=50, # NOT Changed
|
380 |
guidance_scale=7.0, # NOT Changed
|
381 |
seed=seed_value,
|
|
|
411 |
|
412 |
generate_button.click(
|
413 |
generate,
|
414 |
+
inputs=[prompt, video_input, strength, seed_param, enable_scale, enable_rife],
|
415 |
outputs=[video_output, download_video_button, download_gif_button, seed_text],
|
416 |
)
|
417 |
|
418 |
enhance_button.click(enhance_prompt_func, inputs=[prompt], outputs=[prompt])
|
419 |
|
420 |
+
video_input.upload(
|
421 |
resize_if_unfit,
|
422 |
+
inputs=[video_input],
|
423 |
+
outputs=[video_input]
|
424 |
)
|
425 |
if __name__ == "__main__":
|
426 |
demo.queue(max_size=15)
|
427 |
+
demo.launch()
|