Voyage: review pages
XR Room Engine, the Voyage show. Pages for the owner's review.
Visual statusWhere each act of the show stands, with frames. Scene-loop trialHow each step of the scene-loop skill contributed, round by round. How far off is too farExample: the dog's artist skeleton made worse step by step; mark where it stops being good enough to animate. Grade the treesTwenty numbered trees, blind: is each one right, nearly right or wrong? Your grades check the score. Is the tree rightEach animal's reference (the model or the AI picture) beside the artist's skeleton and our tree, from the same camera. Joint skeleton checkPlan v56 step B: the fixed joint trees on the three AI animals, and the image model's joint marks measured against artist skeletons on ten rigged animals (pass line fails; 2D diagnostic), with a 3D viewer. Joint treesPlan v56: trees of balls whose points come from the image model's joint marks, next to today's shape-only tree, for the flamingo, agama and scorpion, with a 3D viewer. Joint marksPlan v56, first look: an image model marks the joints of three AI animals (flamingo, agama, scorpion) with red dots, twice each, plus named joints in numbers. AI animalsPlan v55: eleven animals drawn by an image model, each built from its simplified sheet as a tree of balls with sheets for wings and fins and a dome for a shell, checked against an unseen view. Fitter resultsThe plan v2 tests in one report: Blender tools, the tree of 37 models, the tool choice, trees from pictures, hard pieces, simplified pictures. Fitter plan v2Learning from creature artists: the reading, a first tree on eight true models, and the updated plan. Fitter roadmapModel from pictures: where we are, and the plan for the Codex reset. Views per partModel from pictures: more views, the views that agree picked part by part, special angles, and shading to choose the tool. Depth and smooth fitsModel from pictures: carving by depth (exact, and from surface direction), the fitted tubes without the steps, and AI pictures from more angles. Features stageModel from pictures: cut-in-half views (true models), and each part's fit refined to match the pictures.