ADR-0459: vmaf-tune panel/display-aware recommendation workstream¶
- Status: Proposed
- Date: 2026-05-15
- Deciders: Lusoris, Claude (Anthropic)
- Tags: vmaf-tune, ai, hdr, training, panel, display, fork-local
Context¶
The 2026-05-15 HDR/UGC dataset license audit (Research-0136) surfaced a dataset class the fork has no machinery to consume: HDRSDR-VQA (Chen et al. 2025) ships 22 000+ pairwise JOD scores across 6 distinct HDR display panels (OLED + QLED + LCD variants). That is the only public dataset that cleanly separates panel-induced quality variance from encode-induced quality variance.
Today vmaf-tune emits a single CRF / preset recommendation per (target VMAF, codec) tuple and assumes a generic display. The HDRSDR-VQA pairwise data lets us learn panel-conditioned recommendations — i.e. the optimal CRF for "85 VMAF on a 2024 OLED" can differ measurably from "85 VMAF on a 2020 QLED" for the same source.
Decision¶
Stand up a panel-aware recommendation extension for vmaf-tune as a dedicated workstream tracked under this ADR. Concrete scope:
- Ingest the 31 open-sourced HDRSDR-VQA clips + JOD scores via a new adapter
ai/scripts/hdrsdr_vqa_to_corpus_jsonl.py. - Add a
panel_classenum column to the corpus schema covering the 6 HDRSDR-VQA display panels (OLED-2024, OLED-2022, QLED-2024, QLED-2022, LCD-2024, LCD-2022 — exact labels TBD per the paper). - Train a sibling
vmaftune_panel_predictor_v1ONNX that maps (canonical-6 features + panel_class one-hot + codec ENCODER_VOCAB) → CRF delta vs. the base panel. - Surface
vmaf-tune --panel oled-2024 ...CLI flag that loads the panel predictor and applies the per-panel CRF delta to the base recommendation. - Document the limit: 31 open-sourced clips is small for a robust model; the head ships as
Status: Proposeduntil the gate clears on a held-out subset.
Alternatives considered¶
| Option | Pros | Cons | Why not chosen |
|---|---|---|---|
| Stand up the workstream (this ADR) | Closes the only public dataset class the fork can't currently consume; differentiates the fork on a real-world axis | Small training corpus (31 clips); gate threshold is uncertain | Chosen — the differentiation value is high, and the dataset is the only public source for this signal |
| Skip panel-awareness; rely on display-agnostic VMAF | Smallest implementation surface | Loses the only dataset axis we have for panel variance; defers to client-side calibration which we have no control over | Rejected — leaves a measurable accuracy gap on real-world consumer displays |
| Fold panel-class into the existing fr_regressor_v3 head | No new ONNX | fr_regressor_v3 predicts VMAF, not CRF deltas; conflating the two would muddy both heads | Rejected — separate concerns, separate heads |
| Wait for a larger panel-pairwise corpus | More robust gate | No public dataset of comparable scale exists; YouTube SFV+HDR is opaque on access | Rejected — 31 clips is enough to scaffold the pipeline; expansion is a follow-up if/when more data arrives |
Consequences¶
Positive¶
- vmaf-tune gains a panel-aware recommendation surface that no open-source competitor offers.
- Establishes the data-ingestion pattern for future panel-pairwise corpora (LIVE HDR's ambient-condition data, future YouTube SFV+HDR data when access lands).
Negative¶
- Adds a 7th model to the registry (
vmaftune_panel_predictor_v1) with attendant maintenance burden. - 31-clip training set is small; gate may not clear immediately. Status stays
Proposeduntil it does.
Neutral¶
- HDRSDR-VQA ingestion is its own adapter, kept separate from the KonViD-150k / CHUG paths so the panel-class column doesn't pollute the existing corpora.
References¶
- Research-0136 — HDR/UGC dataset license audit (this ADR's parent research).
- HDRSDR-VQA — Chen et al. 2025 (https://live.ece.utexas.edu/research/Bowen_SDRHDR/sdr-hdr-bowen.html).
- ADR-0325 — KonViD-150k corpus ingestion (the existing separate-adapter pattern this ADR mirrors).
- ADR-0336 — KonViD MOS head (the model-registration pattern this ADR mirrors).