Monitoring aspect ratio conversions in legacy video with ReCAP

Legacy television collections in Australia often contain PAL-era material digitised at 720×576 for 4:3 display. Modern delivery is built around 16:9 screens, and archived programmes now pass through editing, transcoding and online systems that assume wide frames. A production can therefore acquire black borders, lose image edges or change proportions without anyone making a deliberate editorial decision. The damage is easy to miss in a crowded playlist, yet it affects playback, preservation and rights management.

The EU-funded Real-time Content Analysis and Processing initiative addresses related problems in broadcast-quality video analysis. As described on the ReCAP project site, the work combines metadata extraction, video-quality monitoring, face and logo recognition, and duplicated-content detection. A ReCAP-based workflow can connect those capabilities to the monitoring and reporting of aspect ratio conversions in legacy content, linking each visual change to evidence about the asset, its source and its processing history.

Why frame geometry needs a preservation record

A frame’s dimensions do not always tell the whole story. A 720×576 PAL file may carry flags specifying a 4:3 display, while a 1920×1080 HD file normally uses square pixels for 16:9. Applying square-pixel assumptions to incorrectly flagged material can make it appear stretched. Records should therefore distinguish coded dimensions, pixel aspect ratio and intended display aspect ratio rather than storing only a width and height.

The available conversion methods carry different consequences. Pillarboxing a 4:3 programme inside 16:9 preserves the complete image but leaves black areas at the sides. Letterboxing a wide master inside a 4:3 frame adds bars above and below. Cropping removes material from the edges, while stretching distorts faces, lettering and logos. Each method changes the viewing experience, and each may cause a different problem for an archive trying to preserve an authentic presentation.

Those consequences are particularly relevant to programmes containing lower thirds, scoreboards, sponsor graphics or captions. ReCAP’s content analysis can support checks that important on-screen elements remain present after processing, while its quality monitoring can identify related signs of aggressive transcoding. Geometry and content need to be assessed together: an intact frame can still contain visible artefacts, and a technically complete frame can still have been cropped incorrectly.

Detecting conversion changes across an archive

The first step is to capture the frame information supplied with each file. Coded width and height should be recorded alongside pixel and display aspect ratios, scan type, codec, container and transfer notes. Original and derivative metadata must remain separate; otherwise, a later transcode can overwrite the evidence needed to explain when a change occurred.

Conversion evidence can then be sought across the timeline rather than in one thumbnail. Automated analysis can measure masked areas, compare frame geometry and identify transitions between differently formatted sections. A dark scene may resemble pillarboxing, so consistency across moving shots, scene changes and title sequences provides stronger evidence than an isolated black region. Uncertain cases should remain marked as unresolved rather than being assigned a conversion method automatically.

Duplicated-content detection adds an important layer by grouping alternate versions of the same programme. That makes it possible to compare a digitised master, an edited broadcast copy and a streaming derivative as one body of evidence. The separate encoding delivery notes provide useful background when teams need to check whether frame-size metadata survives a protected or repackaged delivery path.

Turning technical signals into usable reports

A useful conversion report records the original format, the observed output format, the inferred method, the affected time range and the confidence of the finding. It should also retain supporting frame images, relevant metadata and any quality indicators that could explain the result. A single status without evidence is difficult to defend when the asset is used in a later restoration or rights dispute.

Plain-language findings can accompany the technical fields. One entry might state that a 4:3 master was delivered as 16:9 with pillarboxing and no detected horizontal crop. Another might identify a 16:9 source cropped to 4:3, with evidence of a clipped lower-third edge during a specific shot. The wording should reflect the evidence: an analyst should not claim that a crop removed a title if the title is no longer visible in the reference material.

Different users also need different levels of detail. Media operations teams can receive an exception list, while archivists can open the underlying measurements and sample frames. Face and logo recognition can add context about visible content, and duplicate analysis can reveal inconsistent derivatives within a collection. For archives adapting the workflow to their own catalogue, the ReCAP contact team is the relevant route for discussing how these outputs should be combined.

Working within Australian media realities

Australian archives contain a varied mixture of historical and contemporary material. News bulletins, children’s programmes, drama and advertising created for stations in Sydney, Melbourne and Brisbane may exist as PAL transfers, while other recordings arrive as square-pixel HD or web-native files. The existence of 16:9 material does not mean that every black border is defective: some black areas are intentional, and some legacy files may carry misleading or incomplete aspect-ratio flags.

The local market adds several distinct workflow demands. Public-service operations associated with ABC and SBS, commercial broadcasters, subscription services and regional operators manage different licensing, delivery and retention requirements. A master may serve an internal archive, a broadcast package, a streaming service and promotional clips, creating repeated opportunities for inconsistent conversion or cropping.

Production conversations in Australia often use a quick phrase such as “it looks right” to close a visual review. A ReCAP report supplies the measurements behind that judgement, including whether the 25-frame cadence was retained and whether the displayed proportions changed. It also helps operators notice when a crop cuts into subtitles or Australian caption safe areas, or removes a sponsor bug that carries contractual requirements. This connects an informal viewing impression with an auditable production record.

Building a repeatable conversion review

A reliable review begins with reference records for known preservation masters. For each sample, teams should capture the intended display aspect, coded dimensions, pixel aspect ratio, provenance and transfer history. These records provide a baseline for identifying later changes. The original format should not be inferred solely from visible borders, because a legitimate archival master may already contain bars or have been reformatted before it entered the collection.

Repeated analysis then makes the process sustainable. Checks can run after new digitisation, storage migration, codec replacement or delivery into a new asset-management platform. Duplicate detection can flag versions that disagree with their apparent originals, while video-quality measurements can distinguish compression artefacts from geometry problems. Reports should record changes over time so that an isolated finding can be compared with earlier runs and subsequent transfers.

Clear governance keeps automated findings useful. Thresholds can reflect each catalogue’s preservation policy, uncertain classifications can enter a human review queue, and approved reformatting decisions can be stored with their supporting evidence. Live workflows may also benefit from alerts when a newly received file has unexpected frame dimensions. The important distinction lies between deliberate conversion, properly documented and technically unsuitable conversion that alters the programme without an accountable decision.

The strongest report answers three practical questions: what changed, where did it happen and which content was affected. ReCAP contributes the measurements, visual analysis and version relationships needed to answer them, while human reviewers retain responsibility for ambiguous cases and editorial context.

The practical takeaway is to treat every aspect-ratio change as an inspectable event rather than an invisible storage detail. Preserve the original geometry, attach evidence to each finding and keep a record of every approved reformat. That combination gives producers a clear diagnosis, preservation teams an accountable history and automated workflows a dependable signal for acting on defective conversions.