cutcutcodec 1.6.1

video editing software

Useful links: Binary Installers | Source Repository | Online Documentation |

Description

This video editing software has been designed for speed and to implement some effects that are hard to find elsewhere. The kernel is written in python and C, so it’s easy to integrate it in your own project. Although it allows you to fine-tune many parameters, it’s smart enough to find the settings that are best suited to your project.

This software is fast and highly configurable for the following reasons:

1. Based on PyAV, this software supports an incredible number of formats and codecs.

2. A complete test benchmark guarantees an excelent kernel reliability.

3. Powered by torch and written in C, this software efficiently exploits the CPU and GPU in order to make it very fast.

4. The code is parallelised to take advantage of all the CPU cores, making it extremely fast.

Examples

There are plenty of other examples in the documentation!

In the following stupid example, we blink the blue color and we add a red noise on the bottom.

import cutcutcodec

SETTINGS = [
    {"encodec": "libsvtav1", "rate": 30, "shape": (480, 720)},
    {"encodec": "libvorbis", "rate": 44100, "bitrate": 128000},
]
VIDEO_PATH = cutcutcodec.utils.get_project_root() / "media" / "video" / "intro.webm"

with cutcutcodec.read(VIDEO_PATH) as container:
    video, audio = container.out_select("video")[0], container.out_select("audio")[0]
    noise = cutcutcodec.generation.video.GeneratorVideoNoise().out_streams[0]
    video = cutcutcodec.filter.video.equation.FilterVideoEquation(
        [video, noise], "r0/2 + r1*(i+1)/2", "g0", "b0*sin(2*pi*0.5*t)",
    ).out_streams[0]
    cutcutcodec.write([video, audio], "/tmp/my_video.webm", streams_settings=SETTINGS)

Features

Audio

• General properties

  1. Supports a large number of channels (mono, stereo, 5.1, 7.1, …) with all sampling rate.

  2. Automatic detection of the optimal sample frequency based on shannon theory.

• Generation

  1. Audio-white-noise generation.

  2. Generate any audio signal from any equation.

• Filters

  1. Audio-cutting, Audio-translate and Audio-concatenate.

  2. Add multiple tracks.

  3. Audio arbitrary equation on several channels of several tracks. (dynamic volume, mixing, wouawoua, …)

  4. Finite Impulse Response (FIR) invariant filter. (reverb, equalizer, echo, delay, volume, …)

  5. Denoising based on optimal Winer filtering.

  6. Hight quality anti aliasing low pass filter (based on FIR).

Video

• General properties

  1. Unlimited support of all image resolutions. (SD, FULL HD, 4K, 8K, …)

  2. No limit on fps. (3000/1001 fps, 60 fps, 120 fps, …)

  3. Automatic detection of the optimal resolution and fps.

  4. Support for the alpha transparency layer.

  5. Floating-point image calculation for greater accuracy.

• Generation

  1. Video-white-noise generation.

  2. Generate any video signal from any equation.

  3. Mandelbrot fractal generation.

• Filters

  1. Video-cutting, Video-translate and Video-concatenate.

  2. Resize and crop (high quality, no aliasing).

  3. Overlaying video tracks (with transparency control).

  4. Apply a video arbitrary equation one several video streams.

  5. Fast C and fft differentiable implementation of the lpips, psnr, ssim, uvq and vmaf metrics.

  6. All gamut and gamma colorspace conversion. (sRGB, BT709, BT2020, …)

• Metrics

  Also accessible with a cli: cutcutcodec metric dis.mp4 ref.mp4 --ssim --vmaf > ssim_vmaf.json #. Comparative quality fidelity metrics like PSNR, SSIM and VMAF. #. Comparative quality sementic metrics like LPIPS. #. Non comparative quality metrics like UVQ. #. Non comparative complexity metrics like RMS_SOBEL and RMS_TIME_DIFF.