Keep the ratio with the raster
A stored raster and its displayed shape can differ when samples are not square. Before making an owned archive derivative, record width, height and sample aspect ratio. The dimension pair alone is insufficient to decide whether the intended picture is narrow or wide.
Take an invented 720 by 576 raster with a sample ratio of 16:15. Width divided by height, multiplied by that ratio, gives 4:3. This is a worked geometry case, not a claim that every archive with those dimensions uses that interpretation.
Look for reliable source records and visual references. A circle or known logo can support a review, but a distorted graphic baked into the old source can mislead. Preserve uncertainty where provenance is weak.
Deliver the intended shape
If the destination requires square pixels, choose a raster and resampling treatment that expresses the intended display shape. If it can preserve the correct sample ratio, a different handoff may be suitable. Check its actual requirements rather than forcing one universal archive format.
Review faces, circles and the active picture edges after export. Confirm that padding or a player fit mode has not introduced an additional stretch. A file reporting square pixels can still contain a picture that was stretched incorrectly earlier.
Keep a note of the input ratio, chosen output raster and any crop. This makes a later reviewer able to understand why the dimensions changed. The acceptance criterion is the intended composition in the destination, not a preference for familiar numeric dimensions.
Questions answered
Can 720 by 576 alone determine the intended display shape?
No. Include sample aspect ratio and active-picture context.
Does square-pixel metadata guarantee an unstretched picture?
No. Review the actual resampling history and displayed composition.
Sources & further reading
Official references for the subjects discussed. Service terms and device software can change.