Bitrate

Bitrate is measured in bits per second (bps), typically expressed as Mbps (megabits per second) or Gbps for high-end formats. It tells you how much data the codec is allowed to use per second of video. But raw bitrate numbers are meaningless without codec context. 50 Mbps in H.264 Long-GOP looks significantly worse than 50 Mbps in ProRes 422, because H.264 achieves compression ratios of 50:1 to 100:1 through inter-frame prediction, while ProRes operates at roughly 5:1 to 7:1 with no inter-frame dependency. The same 50 Mbps means ProRes is using every bit for intra-frame fidelity, while H.264 is splitting it across predictive frames that approximate rather than describe the image. There are three bitrate modes: CBR (constant bitrate) maintains a fixed data rate regardless of content complexity — used for broadcast where transmission bandwidth is fixed. VBR (constrained variable bitrate) allows fluctuations within a ceiling — the encoder allocates more bits to complex scenes (fast motion, detailed foliage, water reflections) and fewer to simple scenes (talking head against a wall). CQ/VQ (constant quality / variable bitrate) modes like x264's CRF or x265's qp maintain consistent visual quality by letting the bitrate fluctuate freely — a CRF of 18 in x264 produces visually transparent output for most content at roughly 8–20 Mbps for 1080p, but complex scenes will spike higher. Then there's the All-I vs Long-GOP distinction that confuses people. All-I means every frame is an independently compressed keyframe — ProRes, DNxHR, and most cinema RAW formats are All-I. Long-GOP (H.264, H.265) uses a group-of-pictures structure where only the first frame (I-frame) is fully encoded, and subsequent frames reference previous ones. This is why a 100 Mbps All-I codec preserves more latitude for grading than a 100 Mbps Long-GOP codec — the Long-GOP bits are being spent on storing motion vectors and prediction residuals, not actual pixel data.

Low bitrate manifests differently depending on the codec and resolution. On H.264 at 1080p, anything below 10 Mbps starts showing visible artifacts — mosquito noise around edges, color bleeding in saturated areas, and blocky gradients. At 4K H.264, you need roughly 35–50 Mbps for visually clean results; at 4K H.265, 15–25 Mbps suffices. But here's the trap: most consumer cameras advertise 4K capability while recording at 100–150 Mbps H.264 or 60–100 Mbps H.265. On paper that seems sufficient, but real-world complexity — foliage, water, confetti, rapid handheld motion — can overwhelm the encoder's bit budget. The result: momentary quality drops during the exact scenes that need the most detail. Professional cameras address this with higher bitrate intra-frame formats. The Sony FX6 records XAVC S-I at 600 Mbps for 4K 60fps — that's 10 megabytes per second of data, ensuring every frame has enough bits for clean shadows and smooth gradients. The ARRI Alexa Mini LF writes ARRIRAW at approximately 3.2 Gbps (400 MB/s), which is why it captures latitude that no compressed format can match. For delivery, the target bitrate depends on platform and resolution. YouTube re-encodes everything regardless, so uploading ProRes 4444 vs H.265 at 50 Mbps both end up at YouTube's internal bitrate (roughly 12–20 Mbps for 4K). Netflix recommends x265 encoding at 15–18 Mbps for 4K HDR content with a MaxCLL of 1000 nits. The golden rule: never judge quality by bitrate alone — judge by the codec architecture, bit depth, chroma subsampling, and the actual content being encoded.