cogvideox1.5-5b-prompt-camera-motion

🎥 CogvideoX-5b LoRa: Camera Movement Control

🚀 Try it here: Interactive Demo

---

Description

This LoRa (Low-Rank Adapter) model extends CogVideoX to control camera movement in 6 directions: left, right, up, down, zoom_in, and zoom_out. The LoRa can generate smooth camera motions for enhanced video creation.

<video controls autoplay src="https://cdn-uploads.huggingface.co/production/uploads/63fde49f6315a264aba6a7ed/sFer38aCSYiWtqQIrVJm9.mp4"></video>

<video controls autoplay src="https://cdn-uploads.huggingface.co/production/uploads/63fde49f6315a264aba6a7ed/pwyqKU0xLHZ6TuUda7hez.mp4"></video>

---

Usage

Usage

The LoRa was trained to control camera movement in 6 directions: left, right, up, down, zoom_in, zoom_out.

Prompt Format

Start prompt with text like this:

'Сamera moves to the {}...',
'Сamera is moving to the {}...',
'{} camera movement...',
'{} camera turn...',

Inference examples

ComfyUI example

<a href="https://huggingface.co/NimVideo/cogvideox1.5-5b-prompt-camera-motion/blob/main/cogvideox_1_5_5b_I2V_prompt_motion_lora_example.json">JSON File Example</a>

Minimal code example

import torch
from diffusers import CogVideoXImageToVideoPipeline
from diffusers.utils import export_to_video, load_image


pipe = CogVideoXImageToVideoPipeline.from_pretrained(
    "THUDM/CogVideoX1.5-5B-I2V", torch_dtype=torch.bfloat16
)

pipe.load_lora_weights("NimVideo/cogvideox1.5-5b-prompt-camera-motion", adapter_name="cogvideox-lora")
pipe.set_adapters(["cogvideox-lora"], [1.0])

pipe.enable_sequential_cpu_offload()
pipe.vae.enable_slicing()
pipe.vae.enable_tiling()

height = 768 
width = 1360
image = load_image("resources/car.jpg").resize((width, height))
prompt = "Camera is moving to the left. A red sports car driving on a winding road."

video_generate = pipe(
    image=image,
    prompt=prompt,
    height=height, 
    width=width, 
    num_inference_steps=50,  
    num_frames=81,  
    guidance_scale=6.0,
    generator=torch.Generator().manual_seed(42), 
).frames[0]

export_to_video(video_generate, output_path, fps=8)

Inference with cli and jupyter-notebook examlple you can find <a href="https://github.com/Nim-Video/cogvideox1.5-5b-prompt-camera-motion">on our Github</a>

Acknowledgements

Original code and models CogVideoX.

Contacts

<p>Issues should be raised directly in the repository.</p>