sd-wav2lip-uhq icon indicating copy to clipboard operation
sd-wav2lip-uhq copied to clipboard

moov atom not found / float division by zero

Open GsHeri opened this issue 1 year ago • 2 comments

Using cuda for inference.
[mov,mp4,m4a,3gp,3g2,mj2 @ 000001a212354e00] moov atom not found
Reading video frames...
Number of frames available for inference: 0
(80, 24515)
Traceback (most recent call last):
  File "F:\Projects\sd-next\venv\lib\site-packages\gradio\queueing.py", line 388, in call_prediction
    output = await route_utils.call_process_api(
  File "F:\Projects\sd-next\venv\lib\site-packages\gradio\route_utils.py", line 219, in call_process_api
    output = await app.get_blocks().process_api(
  File "F:\Projects\sd-next\venv\lib\site-packages\gradio\blocks.py", line 1437, in process_api
    result = await self.call_function(
  File "F:\Projects\sd-next\venv\lib\site-packages\gradio\blocks.py", line 1109, in call_function
    prediction = await anyio.to_thread.run_sync(
  File "F:\Projects\sd-next\venv\lib\site-packages\anyio\to_thread.py", line 33, in run_sync
    return await get_asynclib().run_sync_in_worker_thread(
  File "F:\Projects\sd-next\venv\lib\site-packages\anyio\_backends\_asyncio.py", line 877, in run_sync_in_worker_thread
    return await future
  File "F:\Projects\sd-next\venv\lib\site-packages\anyio\_backends\_asyncio.py", line 807, in run
    result = context.run(func, *args)
  File "F:\Projects\sd-next\venv\lib\site-packages\gradio\utils.py", line 641, in wrapper
    response = f(*args, **kwargs)
  File "F:\Projects\sd-next\extensions\sd-wav2lip-uhq\scripts\ui.py", line 160, in generate
    w2l.execute()
  File "F:\Projects\sd-next\extensions\sd-wav2lip-uhq\scripts\wav2lip\w2l.py", line 234, in execute
    mel_idx_multiplier = 80. / fps
ZeroDivisionError: float division by zero

i don't know what the problem is here. something seems to be wrong with my input video, but the error doesn't give me any more information. it's a mp4 with 5:06 length. i have started to work with wav2lip yesterday and successfully processed one other video (which was captured from the same camera, same resolution etc, but it was a lot shorter (9sec for testing).

for this video, i used handbrake to crop the part where the actual mouth movements will be, then saved it with constant framerate and web-optimized mp4. i've tried it a couple of times now, with new versions of the video, better quality, web-optimized yes/no, different croppings. the audio is exactly as long as the video.

i know moov atom has something to do with corrupted / unfinished video files, but this is definitely not corrupt, since i've re-rendered it multiple times and made sure it completed.

also what's conspicuous: the preview image/player for the clip is black and it can't be played. with my other clip, a thumbnail was displayed and i could preview it. with this, it's black and i can't play it

could it be just too long? is there a max length? what else could i try to find out what's wrong here? like i said, a different video clip from the same camera with the same settings worked fine i appreciate any help

edit: tried again with the same video, re-exported via adobe premiere. same result..

GsHeri avatar Dec 04 '23 22:12 GsHeri

Hi, Sorry about that It seems to come from the video input file. Can you senf me the file extension or eventually the file itself?

Thanks

numz avatar Dec 06 '23 11:12 numz

of course, i'd love to. if you give me your email, i can send it via wetransfer

GsHeri avatar Dec 06 '23 13:12 GsHeri