Netflix's bet on advanced video encoding
Netflix’s push to market its “advanced” video encoding prompts widespread skepticism from viewers who say real‑world streaming quality has declined, especially in 4K and dark scenes. Commenters argue that aggressive bitrate reduction, DRM constraints, and device limitations mean subscribers often see heavily compressed video that compares poorly to Blu‑ray, torrents, or Apple TV+. Some acknowledge genuine technical innovation in per‑title and AI‑assisted encoding, but contend it primarily serves to cut bandwidth costs rather than improve the viewing experience.
Overall sentiment on Netflix video quality
- Many commenters say Netflix’s encoding looks bad, especially:
- 4K streams.
- Dark scenes (banding, macroblocking, “pixel porridge”).
- Fast motion, fine textures, and complex scenes (forests, waves, particles).
- Some report no problems on TVs/streaming boxes, suggesting tolerance or device differences.
- Several note that Netflix appears worse than Apple TV+, Blu‑ray, and some other services.
Bitrate, codecs, and device limitations
- Perceived decline since pandemic-era bitrate cuts; people doubt those were fully reversed.
- Apple TV+ is repeatedly cited as using much higher bitrates than Netflix for 4K.
- Netflix uses different maximum resolutions/bitrates depending on:
- Browser, DRM level, and OS (desktop browsers often limited to 720p/1080p).
- Device certification (many Android boxes/TVs get a web-app or lower quality).
- Subscription tier (4K/highest bitrate restricted to premium plans).
- Auto-scaling often drops quality mid-stream; some suspect this is aggressive cost-saving, not just bandwidth adaptation.
“Advanced encoding” vs long-standing techniques
- Several people argue the article overhypes Netflix’s work:
- Per-title and per-shot optimization is framed as new, but seen as an extension of long-known VBR/multipass ideas.
- Some call decreasing color gamut or aggressive compression “ghetto encoding,” not “advanced.”
- Others note Netflix does use modern codecs (e.g., not 4K in H.264) and stress testing material, but say users rarely see those best encodes.
AI/ML-driven encoding
- Discussion of using object detection / attention models to allocate bits to important regions (faces, foreground).
- Some note that encoders already do perceptual weighting, but ML is being explored to improve predictions and decisions.
- Barriers include computational cost and model deployment at scale.
Business incentives, regulation, and “enshittification”
- Strong view that bandwidth cost optimization is prioritized over viewer quality.
- Debate over whether this is “monopolization” vs broader “market power” and shareholder incentives.
- Ideas floated:
- Quality/bitrate standards for using labels like “4K”.
- More regulation, though many see it as unlikely.
- Some see degradation as part of a wider pattern of services getting worse over time.
Piracy, local media, and user control
- Many say torrents, Blu‑ray rips, and personal servers deliver clearly better quality at similar or lower nominal resolutions.
- Technical discussion on how high-quality WEB-DLs are obtained (DRM key extraction, HDCP stripping, vulnerable devices).
- Frustration that users cannot force specific resolutions/bitrates or trade more buffering for stable high quality.
- Some prefer high-bitrate 1080p over low-bitrate 4K and wish industry had prioritized framerate/bitrate over resolution.