Pixel Aspect Ratio Converter

What a pixel aspect ratio actually is

A digital frame is stored as a grid of samples — width × height — but the shape of each sample is not always square. Pixel aspect ratio (PAR) is the width-to-height ratio of a single sample. When PAR is 1:1, the stored raster and the displayed picture are the same shape. When PAR is not 1:1, the stored frame and the displayed frame are different rectangles, and the conversion is one multiplication: displayed width = stored width × PAR (height never changes — PAR describes the shape of a pixel, not a count of lines).

Standard-definition formats inherited non-square pixels from analog broadcast digitizing. D1 and DV sample the active picture line at 13.5 MHz, the BT.601 luma sampling rate chosen so a single clock could serve both the 525-line/59.94 Hz NTSC system and the 625-line/50 Hz PAL system. That shared 13.5 MHz clock produces 720 luma samples per active line in both systems — but NTSC's line duration and PAL's line duration are not the same, so those 720 samples cover different real-world widths in each system. Squeezed into an identical 720-sample-wide raster, the samples end up non-square: roughly 0.91:1 for NTSC 4:3, roughly 1.09:1 for PAL 4:3.

The math, worked twice

Formula: displayed_width = stored_width × (PAR numerator ÷ PAR denominator). Stored height carries straight through unchanged.

HDV / XDCAM 1080 (this tool's default state): 1440×1080 stored, PAR 4/3. 1440 × 4 ÷ 3 = 1920. Displayed size is 1920×1080 — exactly 16:9, matching a native square-pixel 1080p frame in extent, even though HDV only actually samples 1440 real columns and stretches them to fill it.

NTSC DV/D1 16:9: 720×480 stored, PAR 40/33. 720 × 40 ÷ 33 = 872.73. Displayed size is ≈873×480, aspect 1.818:1 — noticeably wider than a clean 16:9 (1.778:1), because the full 720-sample width includes horizontal blanking padding around the active picture, not just the active picture itself.

The 704-active nuance

Every 720-wide SD frame has an "active picture" region of 704 samples; the remaining 16 samples on each line are edge blanking left over from the analog signal's timing reference, not picture content. Run the same PAR math against 704 instead of 720 and the MPEG-4/H.264-family PARs land exactly on-ratio:

• 704 × 10/11 = 640 → 640×480 = exactly 4:3
• 704 × 12/11 = 768 → 768×576 = exactly 4:3
• (704 × 40/33) ÷ 480 = 16/9 exactly (displayed width 853.33, non-integer, but the ratio is exact)
• 704 × 16/11 = 1024 → 1024×576 = 16/9 exactly

That is not a coincidence — those four PARs (10/11, 40/33, 12/11, 16/11) were defined by the MPEG-4/H.264 spec specifically so the 704-pixel active region maps to a clean 4:3 or 16:9. Run the conversion against the full 720-sample width instead — which includes blanking pixels never meant to represent picture — and the result lands near but not on the clean ratio, which is exactly the "why is my math 1.818:1 instead of 1.778:1" surprise above.

Two PAR families, and why files disagree

More than one "correct" PAR exists for a given PAL or NTSC format, because more than one standards body defined one:

On top of those two rational families, older NLEs stored their own rounded decimals: NTSC 4:3 is commonly seen as a flat 0.9, PAL 4:3 as 1.0666 in some legacy exports, and early Adobe Premiere wrote PAL D1/DV as 1.0940 (4:3) and 1.4587 (16:9) in its own project metadata. None of these decimals are "wrong" exactly — they are rounded approximations of one of the two rational families above — and the rounding is why the same PAL DV master can report three slightly different aspect ratios depending on which application wrote its metadata.

Practical guidance

If a clip reads "720×480" or "720×576" but plays back stretched or squeezed in a tool that ignores PAR, that tool is showing the raw stored raster, not the displayed picture — apply the format's PAR (or use this converter) before judging framing. When upscaling archive DV or D1 footage onto a modern square-pixel timeline, convert to square pixels first, then upscale: resize a PAL 4:3 DV master to 768×576 (the active-704 square-pixel size) before running it through any upscaler, rather than upscaling the stretched 720×576 raster and baking the distortion into the enlargement. Modern cameras and delivery formats are square-pixel (PAR 1:1) almost without exception; the one PAR that survives in current production is the anamorphic lens desqueeze — 1.33×, 1.5× or 2× — which this tool's "Anamorphic 2x" preset models the same way as any legacy SD PAR.

1. Enter the stored resolution the way the file reports it — 720x480, 720x576, 1440x1080 and so on.

2. Pick a PAR preset for NTSC or PAL DV/D1, HDV/XDCAM or DVCPRO HD, or type a custom squeeze.

3. Read the displayed width and resulting ratio off the to-scale diagram, then size the delivery frame in the Aspect Ratio Calculator.