GSC CfP — Use Cases & Test Categories

V6.3: the former Simulation 1/2 are now treated as reference scenarios. Experts can create, save, export/import and compare their own scenarios before converging on a consensus candidate.

PROPOSED AVATAR EXTENSION UC30 — Conversational avatar with an animatable Gaussian splat appearance and the dedicated AV-AI-1F category are handled as a proposed extension to the 22-use-case m78621 scenario model. They are intentionally kept outside the editable C1/C2/CW derivation engine. Candidate content: Arago (m79037).

Test-sequence complexity: A = Low content complexity, B = Medium content complexity, C = High content complexity. The A/B/C codes remain in parentheses for traceability; quantitative thresholds remain to be agreed.

Consensus workflow

Use casescommon characterization

Reference 1extended coverage

Reference 2primary condition

Expert scenarioseditable C1/C2/CW choices

Comparisonassignments + resulting categories

Consensus candidateagreed CfP basis

Reference scenarios

Reference scenario 1 — Extended coverage

Relaxed mapping: secondary test conditions are allowed where useful to explore broader coverage.

Reference scenario 2 — Primary-condition / simplified mapping

One preferred test condition per use case. This is the current concise candidate for CfP test-category definition.

Scenario exchange

Local collaboration model: saved scenarios are stored in this browser only. Use JSON export/import to exchange proposals between experts or collect feedback from multiple organizations.

Proposed avatar use case extension

UC30 m78832 AV-AI-1F

Conversational avatar with an animatable Gaussian splat appearance

A conversational agent, for example a museum guide or virtual teacher, is represented by a photorealistic Gaussian-splat head or upper-body avatar. The static avatar asset is delivered once and stored by the receiver. During the conversation, speech audio and a compact animation-parameter stream are transmitted; the receiver deforms and renders the Gaussians.

1F · 6DoF object · Medium content complexity (B) · AI · quality evaluated after animation

GSC scope: static avatar asset, including canonical Gaussians, expression-dependent/aligned variant data and binding information. Separate scope: the animation-parameter stream belongs to Animation Compression.

Related use cases: UC23, UC24 and UC26. m78832 also proposes extending UC26 to per-primitive, per-blendshape attribute offsets.

Candidate content: Arago — m79037 · 71,066 SH3 Gaussians · neutral + 50 aligned blendshape states · binding + skeleton + trim data.

Search Frames Immersiveness Fidelity Config

rows shown

Reference scenario 1 — Extended coverage

Relaxed mapping: secondary test conditions are allowed where useful to explore broader coverage.

Reference scenario 2 — Primary-condition / simplified mapping

One preferred test condition per use case. This is the current concise candidate for CfP test-category definition.

Use-case assignments

UC	Use case	Fidelity Need	Reference 1	Reference 2	Difference

Working scenario: create an expert proposal by changing C1/C2/CW assignments. Changes are automatically retained in this browser.

Scenario name Expert / organization Notes

Reset from Import JSON

UC	Use case	Fidelity Need	Ref. 1	Ref. 2	Working C1/C2/CW	vs Ref. 2

Resulting test categories

AV-AI-1F The avatar category is a proposed special category and is therefore not generated or edited by the baseline 22-use-case C1/C2/CW scenario rules above.

Saved expert scenarios

Scenario A Scenario B show only assignment differences

Use-case differences

UC	Use case	Scenario A	Scenario B	Change

Resulting category differences

AVATAR CONTENT Arago — m79037

71,066 SH3 Gaussians · 51 aligned GS states (neutral + 50 blendshape variants) · per-Gaussian binding · skeleton · trim data · static 1F coded asset.

Associated evaluation motion: 907-frame tracked performance at 25 fps plus speech-driven sequences. The animation stream is evaluation material and is not counted as GSC-coded content.

Open candidate content & voting

Generated automatically from detailed Content rows.

Content requirement	Frames	Immersiveness	Test-sequence complexity	Use cases	Variants	Production

Proposed avatar-specific category

SPECIAL CATEGORY AV-AI-1F — animatable avatar, static asset. Proposed in m78832; bitstream code TBD. It is separate from the scenario-derived categories shown below.

Category	CfP code	Input	Config	View	Coding conditions	Evaluation	Rate points	Candidate
AV-AI-1F	TBD	Static animatable avatar asset (1F)	AI	6DoF object	C1 and C2; binding indices / identifiers / masks lossless; cardinality and order preserved or mapped	After animation: apply identical reference motion to original and decoded assets, render common viewpoints, compute RGB/YUV-PSNR and optionally SSIM/LPIPS	Proposed relative grid: 1/2, 1/4, 1/8, 1/16 of float16 asset size	Arago (m79037)

Scenario

Test category	Condition	Config	Frames	Use cases	Required content characteristics	Subjective test

Automated consistency checking. The derivation workflow checks coherence between use-case assignments, content requirements and derived test categories. The previous static QA-issue count has been removed; any future issue list should be regenerated from the current source spreadsheet rather than carried over from an earlier revision.

Consistency-checking role

Expert assignments remain the input. Automated derivation propagates those assignments to the Content Summary and Category Summary and is used to detect inconsistencies between the detailed use-case model and the derived CfP structure.