Expand

clean_latents_4x = clean_latents_4x.repeat(1, 1, 2, 1, 1)

Optimize indices

HunyuanVideo-I2V

No stream

Revert

New model

Restore param

Upload 4 files

Upload 4 files

Add logs

Upload 4 files

Upload 4 files

concat_latent()

pillow_heif.register_avif_opener()

Upload 4 files

Upload 4 files

Upload 4 files

Remove logs

End Generation

Fix indentation

Add logs

block.launch(share=True, mcp_server=True)

Try return

End button

set_mp4_comments_imageio_ffmpeg(output_filename, f"Prompt: {prompts} | Negative Prompt: {n_prompt}");

Upload 4 files

Upload 4 files

Better order

Concat

Timeless prompt

Teacache factor

Optimization

Upload 4 files

Upload 4 files

After final_prompt

final_prompt

Revert

Remove logs

timeless_prompt.change(fn=handle_timeless_prompt_change, inputs=[timeless_prompt], outputs=[final_prompt])

Timeless prompt

Optimize the other function

Optimize function

Optimize code

Placeholder

Upload 4 files

Upload 4 files

from diffusers_helper.memory import cpu, gpu, get_cuda_free_memory_gb, move_model_to_device_with_memory_preservation, offload_model_from_device_for_memory_preservation, fake_diffusers_current_device, DynamicSwapInstaller, unload_complete_models, load_model_as_complete

handle_timed_prompt_change

Timed prompts