Loading candidate content…Fetching catalogue and voting results
Gaussian Splat Coding› CfP preparation› Candidate content & expert poll
← CfP portalUse cases & test categories

GSC CfP Candidate Content — Expert Poll

Transparent expert preferences for candidate test sequences · company-balanced aggregation

Informal expert preference poll. This tool supports CfP content selection and does not replace MPEG consensus. Votes are public by name and affiliation; the email address is used only to prevent duplicate ballots and is never displayed.
Candidate sequences
Two content sets are shown. Call for Content candidates are new material proposed for the CfP. Preliminary CTC contains the sequences already listed in WG5 N420; they are shown in a separate section so experts can also assess the existing test material.

candidates shown. Scores: 1 Reject · 2 Probably reject · 3 Moderate · 4 Probably accept · 5 Accept.

Ranking score = company-balanced average. We first average the scores from experts within each company/affiliation, then average those company means. The expert average is shown separately.
SequenceUse caseCompany-balancedExpert avg.CompaniesExpertsScores 4–5Readiness
Detailed public votes
Use-case coverage view. All 22 use cases from the CfP use-case model are shown, including those with no candidate content. Candidate sequences are ranked by company-balanced score. The coverage badge is a visual heuristic only: an average score of 4 or 5 corresponds to “Probably accept / Accept”; the minimum-company control determines which scores are considered for the badge.
Coverage by use case
Candidate content not yet mapped to a numbered use case
These items are visible for completeness but voting is intentionally disabled until the material is delivered or a sequence shortlist is defined.

Vote scale

1 Reject
2 Probably reject
3 Moderate / neutral
4 Probably accept
5 Accept
No opinion is allowed and is not included in averages.

Identity and duplicate prevention

Each expert registers once with name, company/affiliation and MPEG email. The email is converted to a server-side keyed hash and is not exposed. A private browser token lets the expert update the same ballot later.

Aggregation

The primary score is company-balanced. Experts from the same company are averaged first, so a company with several participants does not automatically receive more aggregate weight than a company with one participant.

Readiness

Technical readiness is kept separate from expert support. The badges reproduce the status information available in the updated JEE6.1 report: upload, license, camera/COLMAP information, pose traces, anchor availability and the new quality-validation field. A question mark means that the report does not yet provide a validated status.

Complexity labels: A/B/C are shown as Low/Medium/High content complexity. Other values such as D or M are retained exactly as reported because the report does not provide an agreed semantic definition for them.
Source: candidate metadata is synchronized with the updated October 2026 JEE6.1 data-preparation report (m78644 v2). The report now also tracks a Quality validation field; it is currently unpopulated for the listed I-3DGS candidates. Where the report marks information as incomplete or ambiguous, the page keeps that limitation visible rather than silently resolving it.
Preliminary CTC source: the separate CTC section is derived from approved WG5 N420, Common test conditions for Gaussian splat coding. CTC-specific 1F/NF status, start frame, view count, frame size, source contribution and available pose-trace material are kept separate from Call-for-Content metadata.