Resolution and aspect ratio — what 1080p, 4K, and 16:9 actually mean

Resolution and aspect ratio are the two foundational geometric properties of video. Resolution is how many pixels make up each frame. Aspect ratio is the shape of the rectangle those pixels form. Both shape encoder decisions, ABR ladder design, scaling pipelines, and the audience reach of your content. They sound trivial but accumulate hidden complexity — pixel aspect ratio vs display aspect ratio, 4K UHD vs DCI 4K, anamorphic vs non-anamorphic, square vs non-square pixels. This page is the engineering reference.

#What resolution is

Resolution is the spatial pixel count of a video frame, expressed as width × height. A 1920×1080 frame has 1920 columns × 1080 rows = 2,073,600 pixels.

Common video resolutions and their conventional names:

The "p" in 1080p stands for "progressive" (vs interlaced, denoted with "i"). Modern streaming is essentially all progressive; interlaced ("1080i") survives in some broadcast contribution feeds.

#4K UHD vs DCI 4K

"4K" is ambiguous. Two different specs claim the name:

• 4K UHD (Ultra High Definition) — 3840×2160. Consumer streaming, TV broadcast. Exactly 2× 1080p in each dimension. Aspect ratio 16:9.
• DCI 4K — 4096×2160. Digital Cinema Initiatives standard. Cinema projection. Aspect ratio 1.896:1 (slightly wider than 16:9).

Consumer "4K" essentially always means UHD (3840×2160). Cinema "4K" essentially always means DCI 4K. When ingesting content from mixed sources (cinema masters and streaming masters), verify which "4K" you're handling.

For pipeline use:

• Cinema sources may need cropping or scaling to fit consumer 16:9.
• Cinema content delivered to streaming typically letterboxes or pillarboxes to preserve the aspect ratio in 16:9 frame.

#What aspect ratio is

Aspect ratio is the rectangle shape of the video frame, expressed as width:height ratio.

Common aspect ratios:

• 4:3 (1.33:1) — historical standard television. Square-ish frame. Pre-2000s TV, retro computer monitors.
• 16:9 (1.78:1) — modern standard. HD/4K consumer video, modern TV, web video.
• 1.85:1 — common cinema "flat" widescreen. Slightly wider than 16:9.
• 2.39:1 / 2.40:1 — cinema "scope" / anamorphic widescreen. Significantly wider than 16:9.
• 21:9 (2.33:1) — ultrawide consumer monitors and some TVs.
• 1:1 — square. Instagram-era mobile content.
• 9:16 (vertical) — TikTok, Instagram Reels, YouTube Shorts. Mobile vertical video.
• 2.76:1 — Ultra Panavision 70 (rare cinema).

For streaming, 16:9 is the production default for landscape content. Cinema content (1.85, 2.39) typically gets letterboxed within 16:9 frames for delivery. Vertical content (9:16) increasingly has its own delivery paths (mobile-first platforms).

#Pixel aspect ratio vs display aspect ratio

Pixels can be square (typical digital) or non-square (legacy SD video). The distinction:

• Pixel Aspect Ratio (PAR) — the shape of an individual pixel. Square pixels have PAR 1:1. Non-square pixels have other ratios.
• Display Aspect Ratio (DAR) — the shape of the displayed frame. Always what the viewer sees.
• Storage / Sample Aspect Ratio (SAR) — same as PAR; the term used in some specs.

Math:

DAR = (storage_width × PAR) / storage_height

For square pixels (PAR = 1:1), DAR = storage_width / storage_height directly.

Where non-square pixels appear:

• PAL DVD — 720×576 storage with 16:11 PAR for 16:9 DAR (anamorphic widescreen).
• NTSC DVD — 720×480 storage with 32:27 PAR for 16:9 DAR.
• DV / DVCAM — 720×480 NTSC or 720×576 PAL with non-square pixels.
• Anamorphic cinema source — captured with anamorphic lenses; expanded horizontally for display.
• HDV / 1440×1080 — 1440 wide stored, displayed as 1920×1080 (4:3 PAR).

For modern digital production (any new content captured to file), PAR = 1:1 (square pixels). Non-square pixels are a legacy concern from broadcast and DVD-era content.

When pipelines ingest non-square-pixel content, the conversion to square-pixel output requires explicit scaling:

ffmpeg -i pal_dvd.vob -vf "scale=iw*sar:ih,setsar=1" -c:v libx264 output.mp4

The iw*sar expression scales width to compensate for the source's non-square pixel aspect ratio; setsar=1 declares output as square-pixel. The result is a properly-rectangled 16:9 frame.

#Anamorphic content

Anamorphic content is captured or stored with horizontally-compressed pixels that should be expanded for display. Two scenarios:

Anamorphic capture (cinema) — anamorphic lenses optically squeeze a wide cinema image onto standard 4:3 or Super 35 film/sensor. The image looks horizontally squashed in source; the projector expands it horizontally for display.

Anamorphic storage (DVD) — DVDs store widescreen content as 720×576 (PAL) or 720×480 (NTSC) with non-square PAR. Decoders expand horizontally for 16:9 output.

For pipelines processing anamorphic source, the expansion is part of the ingest stage. The output is square-pixel content at the correct DAR.

#Letterbox vs pillarbox

When content aspect ratio doesn't match container aspect ratio, the difference is filled with black bars:

• Letterbox — wider content in narrower container. Black bars on top and bottom. E.g., 2.39:1 cinema content in 16:9 streaming frame.
• Pillarbox — narrower content in wider container. Black bars on left and right. E.g., 4:3 content in 16:9 streaming frame.

The black bars are part of the encoded frame. They cost bitrate (small amounts; black areas compress very well) and reduce the active picture area. Some players auto-crop black bars on display ("zoom mode") but this is consumer-controlled rather than encoder-controlled.

For streaming, letterbox/pillarbox is operational reality — content from multiple aspect ratios coexists in 16:9 streaming infrastructure. Premium streaming services occasionally ship aspect-ratio-correct variants (2.39:1 native delivery for compatible players) but this is rare.

#Resolution and aspect ratio in ABR ladders

ABR ladder design picks (resolution, codec, bitrate) per rung. Resolution scaling considerations:

Aspect ratio preservation — all rungs should share the same DAR. Don't ship 16:9 top tier and 4:3 floor tier; this confuses player adaptation.

Resolution stepping — typical ladder steps in roughly 2× pixel-count progression: 4K (3840×2160) → 1440p (2560×1440) → 1080p (1920×1080) → 720p (1280×720) → 540p (960×540) → 360p (640×360) → 240p (426×240).

Even-multiple resolutions — encoders prefer width and height divisible by codec macroblock size (16 for H.264; 8-64 for HEVC depending on settings). Common resolutions are designed to be encoder-friendly.

Display-fit logic — players pick variants based on bandwidth + display capability. 4K rungs are wasted on 1080p displays; players usually pick the closest-matching variant.

For typical 2026 premium streaming ladders, 6-8 rungs from 360p to 4K covers the audience well. Below 360p is rare (cellular bandwidth has improved). Above 4K is rare (8K install base is small).

#Resolution scaling

Scaling between resolutions requires interpolation algorithms:

• Nearest-neighbor — pick the closest source pixel. Fast; produces pixelated output. Used for pixel art preservation.
• Bilinear — average 2×2 source pixels. Fast; produces soft output.
• Bicubic — weighted 4×4 source pixel kernel. Better quality than bilinear; modest cost.
• Lanczos — sinc-windowed kernel (typically Lanczos-3 or Lanczos-4). High quality; common production default.
• Spline — spline interpolation. High quality; similar to Lanczos.
• AI / ML upscaling — neural network-based scaling. State of the art for upscaling (e.g., 1080p to 4K). Tools: Real-ESRGAN, NVIDIA RTX Video Super Resolution, Topaz Video AI.

For ffmpeg scaling, the scale filter exposes the algorithm:

ffmpeg -i input.mp4 -vf "scale=1280:720:flags=lanczos" output.mp4

For most streaming pipelines, Lanczos is the production default — high enough quality for downscaling (the common direction); modest compute cost.

For upscaling (less common in streaming but used in some pipelines for 4K from 1080p source), AI/ML upscaling produces dramatically better output but at significantly higher compute cost.

#Resolution and bitrate

Bitrate scales roughly with pixel count, but not linearly. The general guidance:

• 4× pixel count (e.g., 1080p → 4K) ≈ 2-3× bitrate at equivalent quality.
• 2× pixel count (e.g., 720p → 1080p) ≈ 1.5-2× bitrate.

The sub-linear scaling reflects the encoder's ability to compress related pixels efficiently — adjacent pixels at higher resolution are more correlated, allowing more efficient prediction.

For ABR ladder calibration, this scaling is a starting point. Per-content tuning (using VMAF measurement on actual content) refines the bitrate per rung based on content complexity.

#High-resolution content (8K and beyond)

8K UHD (7680×4320) is 4× the pixel count of 4K UHD. In 2026:

• 8K consumer hardware is shipping (Samsung, LG flagship TVs; some streaming devices).
• 8K content is rare. YouTube has some, NHK broadcasts some, premium streaming services occasionally ship.
• Bandwidth requirements are significant (~30-80 Mbps for 8K HEVC at acceptable quality).
• Most viewers don't have displays large enough to perceive 8K detail at typical viewing distances.

For pipeline operations, 8K is currently a niche output. Most consumer streaming caps at 4K UHD; 8K is reserved for specific platforms and customer requirements.

5K (5120×2880) is the resolution of Apple Studio Display and iMac. Mostly relevant for desktop video editing and color-grading workflows; rarely used for streaming delivery.

#Vertical and square video

Mobile platforms (TikTok, Instagram Reels, YouTube Shorts) drove the rise of vertical (9:16) video. Square (1:1) video had a moment with Instagram's original square-only constraint.

For streaming infrastructure, vertical content adds operational considerations:

• Manifest signaling — players need to know orientation to layout correctly.
• Aspect-correct ABR ladder — vertical ladder rungs should be 9:16 throughout.
• Cross-platform delivery — same content for both vertical and landscape platforms requires parallel encodes.

Most consumer streaming infrastructure handles 16:9 primarily. Vertical content delivery is increasingly its own track in the streaming-infrastructure ecosystem.

#Operational considerations

Things that matter for resolution and aspect ratio in production:

• Even-numbered dimensions — codec macroblock requirements often require width and height divisible by 2, 4, or 16. Verify after scaling.
• Aspect ratio preservation through scaling — don't accidentally crop or stretch when scaling. Use scale=W:-1 (preserve aspect by computing height) or scale=-1:H (preserve aspect by computing width).
• Color space and bit depth interaction — scaling can produce out-of-range colors at scale boundaries. Use format=yuv420p after scaling to clip to valid range.
• Source resolution detection — pipelines that auto-pick ABR ladders based on source resolution need accurate detection at ingest.
• Display-aspect signaling — setsar=1 produces square-pixel output that signals correctly. Without it, downstream tools may misinterpret.
• Watermark / logo positioning — hardcoded coordinates break when ladder rungs have different resolutions. Use percentage-based positioning.

#What MpegFlow does with resolution and aspect ratio

MpegFlow's DAG runtime detects source resolution and PAR/DAR through an FfprobeExecutor stage upstream of encode. Cross-stage data flow wires the probe output into the parallel FfmpegExecutor rendition stages so resolution targeting and PAR/DAR normalization derive from the real source rather than assumed defaults. The partitioner persists each rendition stage to job_stages with explicit dependency tracking; per-stage retry handles transient failures.

ABR ladder configuration in workflow YAML specifies resolutions per rung; each rendition is its own stage on FfmpegExecutor. Default ladder templates use the standard pixel-count progression (4K → 1440p → 1080p → 720p → 540p → 360p) at 16:9; customers configure different ladders for non-16:9 content (vertical, ultrawide, cinema aspect ratios) by overriding per-rung values.

For source content that requires aspect-ratio handling (cinema 2.39:1 in 16:9 streaming), the letterbox/pillarbox padding runs through filter parameters at the FfmpegExecutor stage with configurable color (typically black, sometimes other colors for branded delivery).

For 4K content specifically, the pipeline distinguishes UHD (3840×2160) from DCI 4K (4096×2160) at the probe stage and the rendition stages apply appropriate scaling/cropping when output is to streaming UHD from cinema-source DCI 4K.

The strict-broker security model handles resolution/aspect-ratio work like any pipeline payload — workers carry no ambient credentials; content access flows through short-lived presigned URLs scoped per stage; access is disposed on completion.

For customers configuring their first ABR ladder, the conversation typically focuses on audience device mix (informs the resolution tiers worth shipping), content nature (16:9 vs cinema vs vertical), and bandwidth budget (informs how deep the floor tier should reach). Resolution and aspect ratio decisions are foundational and worth explicit attention rather than relying on defaults.

The general guidance: detect resolution and aspect ratio explicitly at ingest, normalize PAR/DAR to square-pixel output, preserve aspect ratio across ABR rungs, and design the ladder to match audience device mix. Resolution and aspect ratio handling is one of the parts of pipeline engineering that's invisible when it works correctly and visible (broken thumbnails, stretched images, missing video areas) when it doesn't.