stable_diffusion_videos==0.3.0 🚀

Highlights

Resume Support 🔥

You can now resume unfinished runs instead of starting over. Just pass resume=True and we'll resume the run at <output_dir>/<name>.

from stable_diffusion_videos import walk

video_path = walk(
    output_dir='dreams',
    name='my_unfinished_run',
    resume=True  # All you need to do!
)

Thank you to @codefaux for adding this feature.

Batch Inference 🔥

We've been generating frames one at a time. It's a lot faster to do more than one instead. Play with the batch_size kwarg until you go out of memory, then reduce the value by 1. For me, I am able to do batch_size=4 on a V100 16GB, which generates images ~20% faster.

from stable_diffusion_videos import walk

walk(
    prompts=['a cat', 'a dog'],
    seeds=[42, 123],
    batch_size=4,  # All you need to do!
    num_steps=60,
    make_video=True
)

What's Changed

• adding fps as param to walk by @kennethgoodman in https://github.com/nateraw/stable-diffusion-videos/pull/23
• Add option for lower VRAM by @0x1355 in https://github.com/nateraw/stable-diffusion-videos/pull/31
• Resume support by @codefaux and @nateraw in https://github.com/nateraw/stable-diffusion-videos/pull/42
• Remove unnecessary video file from repo by @nateraw in https://github.com/nateraw/stable-diffusion-videos/pull/43
• Add batch inference by @nateraw in https://github.com/nateraw/stable-diffusion-videos/pull/45
• Save in .png instead of .jpg by @0x1355 in https://github.com/nateraw/stable-diffusion-videos/pull/44
• Add frame_filename_ext to control saving/resuming from .png or .jpg by @nateraw in https://github.com/nateraw/stable-diffusion-videos/pull/47

New Contributors

• @kennethgoodman made their first contribution in https://github.com/nateraw/stable-diffusion-videos/pull/23
• @0x1355 made their first contribution in https://github.com/nateraw/stable-diffusion-videos/pull/31

Full Changelog: https://github.com/nateraw/stable-diffusion-videos/compare/v0.2.0...v0.3.0