# Netflix/vmaf

> VMAF is Netflix's Emmy-winning perceptual video quality assessment algorithm, shipped as a C library plus Python tooling. It scores how close an encoded video looks to its source and supports model training.

- Magnitude: 3.3 out of 10 — Breakout
- Stars: 5,557 total · +1 star measured 2026-10-19, 03:33–05:38 UTC, ≈ 16 by evening
- Star trust: star growth looks organic
- Category: Generative media · Language: C · Created: 2016-02-08 · Last push: 2026-10-17
- GitHub: https://github.com/Netflix/vmaf · Page: https://gitnova.dev/en/r/Netflix/vmaf

## Useful for

- Score encoded video against its source to compare codec or bitrate settings
- Run VMAF as an FFmpeg filter (--enable-libvmaf) in an encoding pipeline
- Train and validate a custom VMAF model on your own dataset with the Python library

## Why it’s here

- Star-counter measurements on 2026-10-19 (UTC), 03:33–05:38: 5556 → 5557 stars (+1). This is the change over that interval.
- Estimated end-of-day forecast: about +16 stars, using observed gains and the previous day.
- Before this, the repository barely got any stars — about 2 per day.
- Hacker News: “Netflix vmaf: Perceptual video quality assessment based on multi-method fusion” — 22 points, 9 h ago.

## Star trust

Star growth looks organic.

Star-trust labels are heuristics based on the repository’s behavior, not a check of every stargazer.

## Numbers

- Forks: 842
- Issues and pull requests: 1,687
- Watchers: 480
- Average over the last week: 7 per day
- Usual pace: 2 per day
- Stars in the last hour (measured): 1
- Latest release: v3.2.1 (2026-09-14)

## Stars per day, last 30 days (oldest → newest, today is partial)

2026-09-20 … 2026-10-19: 3, 2, 2, 3, 2, 3, 1, 0, 2, 4, 2, 1, 1, 1, 1, 1, 4, 2, 0, 1, 2, 1, 3, 1, 2, 0, 3, 1, 24, 1

## Hacker News

- Netflix vmaf: Perceptual video quality assessment based on multi-method fusion — 22 points, 8 comments: https://news.ycombinator.com/item?id=50145295

## Spotted in now

- Spotted on Hacker News

## More in this category

1. **Kruk2/jasna** — 7.2 · Breakout · Generative media · Python · +161 stars measured 2026-10-19, 00:05–05:36 UTC, ≈ 505 by evening
   A tool that restores mosaic areas in videos and images: it detects mosaics and restores only those regions, leaving the rest of the picture untouched. Runs on GPU with a GUI app and a CLI.
   Full card: https://gitnova.dev/en/r/Kruk2/jasna.md
2. **Arekejoker/qwen21-nsfw-pose-skill** — 6.9 · Peaking · Generative media · Python · +329 stars measured 2026-10-19, 00:06–05:36 UTC, ≈ 812 by evening
   A set of recipes and scripts for local NSFW image generation and pose control with Qwen-Image 2.1 via ComfyUI, including an NSFW LoRA and a library of 136 3D poses.
   Full card: https://gitnova.dev/en/r/Arekejoker/qwen21-nsfw-pose-skill.md
3. **Comfy-Org/ComfyUI** — 6.3 · Breakout · Generative media · Python · +248 stars measured 2026-10-19, 00:06–05:36 UTC, ≈ 881 by evening
   ComfyUI is a diffusion-model engine for generating images, video, audio and 3D with a visual node-graph interface. It lets you build and reuse pipelines without code, run them locally or via API.
   Full card: https://gitnova.dev/en/r/Comfy-Org/ComfyUI.md
4. **OpenMOSS/MOSS-Image** — 5.9 · Early signal · Generative media · Python · +68 stars measured 2026-10-19, 00:05–05:36 UTC, ≈ 270 by evening
   MOSS-Image is a 5B-parameter DiT text-to-image base model with strong text rendering, JSON layout control, 4K-token prompts and LoRA fine-tuning support.
   Full card: https://gitnova.dev/en/r/OpenMOSS/MOSS-Image.md
5. **debpalash/VoiceStudio** — 5.7 · Peaking · Generative media · Python · +667 stars measured 2026-10-19, 00:06–05:36 UTC, ≈ 2,448 by evening
   A fully local, open-source ElevenLabs alternative for voice cloning, voice design, video dubbing, dictation, transcription and audiobook creation on your own hardware, with no account or API key. Supports 16 TTS and 11 ASR engines, a…
   Full card: https://gitnova.dev/en/r/debpalash/VoiceStudio.md

---

Magnitude (0–10) measures how fast and how unusually interest in a repository is growing right now. It is not a quality score.
Days are UTC. “So far today” is a fact; “expected by the end of the day” is a forecast.
Summaries and use cases are written by an LLM (DeepSeek V4.1 Flash) from the README and may be inaccurate: verify specific claims (benchmarks, speed, hardware) in the repository itself.
Data as of 2026-10-19 05:46 UTC, updated every 30 minutes.