From Masters to Game Assets: Encoding, Resolution, Bitrate, and File Naming

A master is not a game asset. A production pipeline should retain a high-quality archival master, then generate delivery versions with standardized encoding for each target platform; every file enters the manifest with an asset ID, version, technical parameters, and checksum. Do not let the application link directly to “final_really_final2.mp4”.

An archival master preserves the best picture and sound as a source for future re-encoding; editing intermediates serve post-production; runtime files serve specific platforms; proxies serve review and remote collaboration. These four have different purposes. Low-bitrate review files must not be used as the source for release packages.

Record the master’s resolution, frame rate, color space, bit depth, audio tracks, and basis for loudness measurement. Keep subtitles as separate files or data wherever possible, unless a specific platform requires burned-in subtitles. Generate all conversions automatically or reproducibly from the approved master.

Higher resolution is not automatically better. Choose it based on viewing distance, image detail, device decoding capabilities, and storage and download targets. Bitrate depends on motion, noise, dark areas, and the encoder; the same average bitrate produces different results for static dialogue and a chase on a rainy night.

Build a set of representative clips: dark scenes, fast motion, fine strands of hair, gradients, subtitles, and the first frame after a clip switch. Run blind tests for each encoding tier and record visual quality, time to first frame, dropped frames, temperature, memory use, and file size. Final presets come from evidence, not universal numbers found online.

Use consistent containers, codecs, resolution, frame rate, color settings, and audio formats across branching clips wherever possible, and place suitable keyframes near entry points. Parameter changes may force the player to rebuild its pipeline, causing black screens or volume jumps. Automatically check duration, black frames at the beginning and end, silence, keyframes, and audiovisual synchronization after every encode.

Variable-frame-rate footage is prone to timecode problems across different players, so production pipelines usually first normalize it to the frame rate agreed for the project. Validate the specific choice on every target platform.

A recommended name consists of the asset ID, purpose, language or platform, and version, for example VID_C03_S02_N010_MAIN_zhCN_PC_v007.mp4. Do not use spaces, ambiguous abbreviations, author names, or “final”. Store dates and approval status in the asset manifest or versioning system to prevent file names from growing indefinitely.

Increment version numbers and never overwrite released files. Nodes reference logical asset IDs, which the build manifest then resolves to platform files, so changing the encoding does not alter narrative data.

The manifest records the asset ID, source master, output path, version, duration, resolution, codec, bitrate, audio tracks, subtitles, checksum, node references, approver, and status. Before building, verify that files exist and their checksums match; automatically report orphaned files and duplicate IDs.

Review approval applies to a specific checksum, not a “file with the same name” in a folder. Every replacement creates a new version and triggers regression checks on reference paths.

Standardize sample rates, channel layouts, and loudness policies, and check dialogue, peaks, and switching boundaries. Test multichannel-to-stereo downmixes on target devices to avoid dialogue cancellation. Record the time base, character set, styling, and safe area for subtitles in each language.

For multilingual dubbing, choose separate audio tracks or separate files, but assess package size and download strategy. Either way, logical nodes must not miss choice points because a language change alters the duration.

Batch-transcode with fixed configurations and produce reports to avoid inconsistencies caused by editors manually exporting files one by one. Automated gates check naming, parameters, checksums, black frames, and missing tracks; human gates check compression artifacts, color, subtitles, sound, and narrative continuity.

Back up archival masters, project files, fonts, and subtitle sources as well, and verify that runtime assets can be rebuilt from scratch. Backing up only the final compressed files removes the source material needed for future fixes.

Video files are huge. If a small editing change alters an entire file, a patch may require players to download several GB again. Split packages by chapter and stable clip boundaries to avoid a single subtitle correction triggering updates to all media; excessive fragmentation, however, also increases file requests and management overhead. Run a rehearsal using the actual platform’s differential update mechanism, recording initial installation size, patch size, and peak disk usage.

Before release, perform a complete download from an empty cache and verify it, then upgrade from the previous production release. Both paths must produce the same asset manifest. For optional HD and dubbing packs, test installation, uninstallation, and language switching so that old files do not keep occupying space or remain incorrectly referenced.

Assets can be marked approved only after technical checks, audiovisual review, and narrative checks. If a new version has problems on target devices, the manifest can point back to the previous approved version; rollback should not require changes to node data. Any temporary manual replacement must be caught by the checksum report after the build.

Next step: select five representative clips that are the hardest to encode, generate two or three candidate tiers, and blind-test image quality and performance on the lowest-spec target device; after locking the presets, batch-produce the full work using the manifest and checksums.