Скачать с ютуб First AI Codec Running in FFmpeg & VLC – Deep Render’s Breakthrough в хорошем качестве

First AI Codec Running in FFmpeg & VLC – Deep Render’s Breakthrough

What if an AI-powered video codec could deliver 50% better compression than AV1, encode in FFmpeg in near real time, decode on a Mac or iPhone at 2X real-time, and play seamlessly in VLC? Sound too good to be true? Well, that’s exactly what Deep Render has built. This might be the first AI codec truly ready for deployment—running efficiently on today’s hardware with no need for dedicated decoder chips. In this interview, Jan Ozer speaks with Arsalan Zafar, CTO and co-founder of Deep Render, and Sebastjan Cizel, Head of Engineering, to break down what makes Deep Render’s codec different, why AI-driven compression is finally practical, and how this could change video delivery as we know it. The conversation is divided into three sections: 1️⃣ Arsalan explains what makes an AI codec different and how Deep Render outperforms traditional codecs. 2️⃣ Sebastjan showcases Deep Render encoding in FFmpeg and playing back in VLC. 3️⃣ The team discusses real-world deployment, training, and the future of AI-driven video compression. 📌 Topics Covered: 00:00 Introduction to Deep Render’s AI-based codec 02:08 Traditional vs. AI-based video compression paradigms 03:34 Benchmark comparisons: Deep Render vs. AV1, HEVC, and VP9 07:46 Real-time encoding and decoding performance 08:40 Hardware acceleration and use of NPUs 10:15 AI codec research and comparison with Microsoft, Qualcomm, and WaveOne 14:40 Disrupting the traditional codec development model 17:16 Demonstration of Deep Render’s codec in FFmpeg and VLC 25:59 Training process and model optimization 30:08 AI codec deployment strategy and supported devices 32:14 Market rollout: smartphones, laptops, smart TVs, and dongles Deep Render’s AI codec marks a major shift in video compression by eliminating legacy encoding components and relying entirely on neural networks. Unlike traditional codecs that require years of standardization and dedicated hardware, Deep Render’s solution runs efficiently on NPUs in modern devices, enabling rapid deployment. The demo proves this isn’t just theory—Deep Render’s AI codec is encoding and decoding in FFmpeg and VLC, running in real-time on consumer hardware. With 40-50% better compression efficiency than AV1, HEVC, and VP9, plus real-time speeds on devices like the M4 MacBook Pro, this is the first AI codec even close to ready for production. 🔗 Learn more about Deep Render: www.deeprender.ai/ 🔗 Follow Jan Ozer: www.streaminglearningcenter.com #AI #VideoCompression #DeepRender #FFmpeg #VLC #AV1 #HEVC #TechInterview