Live · real API calls · each step opens when the previous one finishes
Run a survey live
Six stations, top to bottom: clip, measure, labels, draft, program, survey. Every number below is printed by the API's code from this session's calls; nothing is stored on the server.
Step 1: Clip
Try it without a code
Run the one live step anyone can, with no clip, no code and no cost: re-run the survey program Muse Wonder 1.0 wrote against the booklet's 22 printed conditions, in the sandbox (no model call). There is no one-click clip: our real-hull test clips are third-party footage we may not re-host. The Meta model calls themselves are recorded: audit them by response id on /evidence.
Then the real-hull measurement: recorded SAM 3.1 and Spark 1.3 output, every number re-derivable from the run record.
Open the recorded real-hull run →Optional: bring your own prepared clip (judge code)
Supported marks: metric ship draft marks (white numerals, whole metres with “M”), or the 1:10 barge strip.
Camera: side-on to the hull, held still, the mark column and the waterline both in frame, cropped close to the marks. On iPhone, export or share the clip as MP4 first (Most Compatible). Prepared MP4 only: at most 10.5 s, 8 fps and 3 MB.
Step 2: Measure
Opens after step 1.
Step 3: Labels
Opens after step 2.
Step 4: Draft
Opens after step 3.
Step 5: Program
Opens after step 4.
Step 6: Survey + certificate
Opens after step 5.