Animation and performance

2D/3D animation, rigs, retargeting, motion capture, facial animation and lip sync

• articleIn Netflix's two dubbing premix/mix articles, how is performance and delivery intent (effort, breath, emotion, direction) specified for the dubbing studio, and is there any direction-note or emotion field?
  Neither article defines a direction-note, emotion or performance-note field, and neither mentions the dubbing director. Performance intent is specified only by reference to the original version (OV): efforts, breaths and screams must match it, dialogue level, projection and tonality must match it, and reactions are recorded only where the original has them. The two numeric items are a level example (-14 dB regular, -6 dB loud, unit unstated) and a warning not to make an over-upfront dub. Where Netflix states directing intent and lip-sync priorities (Key Moments, PLDL/Show Guide) is in other pages already covered by C2-B10-G2-05; the DAPT/ptulsconv comparison (C3-B212-UO-AM02-01/-02/-03) shows formats that at least offer a free-text note, Reason or EFF slot, which these two Netflix articles do not.
• articleDoes the W3C DAPT dubbing/audio-description script format (a public standard) have a performance-intent field, and how does it keep that apart from timing and from mix/M&E handling?
  At W3C DAPT Candidate Recommendation Draft 26 June 2026, timing, free-text script-event annotation, character/text, and mixing instructions are four separate data-model areas. There is no dedicated emotion, effort or breath field: the closest slots are the free-text Script Event Description (ttm:desc), its small provisional descType registry (pronunciationNote, scene, plotSignificance) and x- user-defined types, plus Rate/Pitch for synthesized speech. No M&E or stem concept appears in the document; mixing is gain/pan. So an agent can use DAPT to keep three kinds of data apart, but must define its own performance-intent convention.
• articleUnity Manual (R1): IK works backwards from a goal so the end effector reaches it, useful "if you want a character to grab an object or stand on an uneven" surface; API: SetIKPositionWeight, SetIKRotationWeight, SetIKPosition, SetIKRotation, SetLookAtPosition, bodyPosition, bodyRotation; setup:...
  Unity Manual (R1): IK works backwards from a goal so the end effector reaches it, useful "if you want a character to grab an object or stand on an uneven" surface; API: SetIKPositionWeight, SetIKRotationWeight, SetIKPosition, SetIKRotation, SetLookAtPosition, bodyPosition, bodyRotation; setup: valid humanoid Avatar, Animator Controller, enable IK Pass on the layer (which sends OnAnimatorIK), then a script sets goals and weights each frame. R2: animation states with humanoid clips expose "Foot IK" ("Enable to respect Foot IK for this state"). Compared with Unreal's documented plant/unplant thresholds and pelvis limits (C3-B567-01), Unity's built-in layer provides the IK solve and weights but leaves contact detection (raycasts), lock/release hysteresis and pelvis lowering to user code or add-on packages (inference from the API surface; Animation Rigging package not read).
• articleAnswers "how do procedural systems keep feet in contact and the body supported?" from engine documentation. Plant (contact) settings (R1): speed_threshold - "Bone is considered planted below this speed" (speed from an FK speed curve or graph root motion); unalignment_speed_threshold - speed at...
  Answers "how do procedural systems keep feet in contact and the body supported?" from engine documentation. Plant (contact) settings (R1): speed_threshold - "Bone is considered planted below this speed" (speed from an FK speed curve or graph root motion); unalignment_speed_threshold - speed at which the foot becomes fully unplanted, the range between the two approximating the foot's roll phase (the docs themselves note this "feels innaccurate most of the time"); distance_to_ground - within this distance of the plant plane the bone is planted and fully aligned; unplant_radius and unplant_angle - linear and angular deviation that release the lock; replant_radius_ratio/replant_angle_ratio - below these fractions of the unplant limits the foot replants (hysteresis); min/max_extension_ratio bound how far the leg chain may extend; separating_distance keeps the feet apart; ankle_twist_reduction limits slope alignment twist; adjust_heel_before_planting; reconstruct_world_plant_from_velocity...
• articleWhat documented methods lock feet / enforce contact during motion retargeting, and what are their limits?
  Partially verified: the curator re-read S44, S46; S45, S47 were not re-read by the curator and rest on the producer's pinned reading. Foot contact is enforced as a separate constraint stage, not by motion plausibility: the reference footskate solver takes explicit plant intervals and exposes a threshold trading knee pops against leg stretching; Auto-Rig Pro Remap documents IK feet as avoiding the sliding seen with FK feet; learned retargeting papers still add IK contact cleanup (producer reading). All methods need contact information, and wrong contact labels create wrong locks.
• article1916 practice and a proposal: what scene-writing check can an agent run on a draft using only Sargent's Ch. XXXIII, and which steps are his versus the worker's?
  Agent check labelled 1916 practice and a proposal. Steps 1-7 follow Sargent's claims (picture, one fact, strike detail, suggestion-only business, by-play test, close-up). Steps 8-9 are the worker's proposal: they measure brevity and test whether independent readers stage the same story fact, because Sargent states the idea (para 10) but gives no test or threshold. All of it is untested.
• articleWhat is 'by-play' for Sargent (1916), when should a scenario include it, how much, and what counts as misuse?
  1916 practice. Sargent defines by-play as acted action that is not required for the direct plot recital but shows side lights on character and adds value. He says it must still bear on the plot and look natural and required. Misuse: introducing a person or action only to heighten the moment's effect. He gives no amount or ratio for by-play and labels no example 'by-play' in this chapter; the chapter's one enriched scene (Lucy appears) is called allusion, not by-play (see -04).
• articleWhat does Sargent (1916) say a scenario writer should leave to the director and the players, and when may the writer specify detailed business?
  1916 practice. Sargent says the writer gives story facts and leaves minor detail, locations and exact movements to the director, whose version may suit the actors better. He allows detailed business only when it is likely to be unusually effective, and says authors seldom beat the director. He offers one detailed garden scene as a 'suggestion' and says experience will show when to detail. He also names a duty that stays with the writer: a close-up when expression carries the meaning. He gives no rule beyond experience for the boundary.
• article[Proposal] Using Lubbock's 1921 chapters IX-XII as a base, what steps can an agent follow to render a character's inner life in action and scene, and what must change for a screen adaptation?
  LABEL: this is a proposal by this worker built on 1921 criticism, not Lubbock's checklist and not a tested method. Nine steps: test whether the subject lies in the mind; fix one seeing eye; list the interior facts and when the reader should hold each; choose surface carriers (look, pause, movement, the image that occupies the mind); never state the fact; place an early glimpse in an unrelated moment; let talk be allusive; pace disclosure; keep to economy. Each step has a Lubbock anchor except the ledger and the recoverability check, which are mine. SCREEN NOTE (proposal): Lubbock's chapters never mention film (0 range hits for film, cinema, screen, photoplay; the one whole-text hit for 'screen' is a verb at L897), so a screen version is an extension: the surface carriers become performance, framing, editing and sound; voice-over is the 'report' position Lubbock ranks below showing; a whole-film subjective camera is not equivalent to dramatizing a mind (see C3-B002-S2-03, C3-B171-S36...
• articleHow does Lutz (1920) keep an animated character the same across many drawings and many workers, and does he say anything about a character changing?
  Lutz's consistency rule is about drawing, not personality: the chief animator first decides the portraiture of his characters, the skill 'still more needed' is keeping a character's portraiture through any series of drawings, helpers must keep a uniformity of portraiture, and since everyone, even the originator, tends to depart from the first-planned face, the approved remedy is a set of sketches drawn on special sheets for all workers to trace from (in a busy studio, engraved plates and printed copies). On a character changing in will, heart or conduct he is silent in the passages and terms searched: the plot he describes is a succession of mishaps growing in violence. The S81 concordance had him as 'no_record' on character change; this is a bounded silence plus a visual-consistency rule.
• articleFrom Lutz's 1920 chapter XI, what check can an agent run to decide whether an explainer segment should be animated and how to sequence it (1920 practice, offered as a proposal)?
  Proposal (synthesis, not Lutz's own list): a six-step check that restates his scattered statements. 1) Can the camera show the thing? If yes and footage exists, photograph it. 2) If not, or if a static plan is the clearer picture, draw it as a moving diagram. 3) Is a textbook schematic available to copy, and a colour convention? 4) Can a draughtsman draw the successive states within the production means? 5) Write the scenario and the titles first; order parts whole-to-parts; add a pointer only where a fault or detail needs naming. 6) End on the complete or corrected state. Evidence level: 1920 practitioner statements, mostly conditional; no source supplies a comprehension test, so the check is structural. AI use is a hypothesis only.
• articleWhat limits or failure modes of animated educational film does Lutz (1920, ch. XI) name, and which obvious ones does he not name?
  Named limits are about means, not about viewers: the medium shows only in monochrome or simple outlines; colouring by hand is tedious and costly; existing colour processes lack some natural colours (from his own observation); refined stories need better drawing and figure skill; and some subject matter cannot be photographed, so diagrams must be substituted. He also states the premise 'all brains function in a similar way' as an assumption behind his case. He does not name comprehension failures (overload, speed, misread titles), cost figures for educational films, audience testing or teacher uptake: in Lutz's text of chapter XI the stems fail, mistake, confus*, overload, test, audience, attention, memory and retain have 0 hits (bounded scan).
• articleWhat does Lutz (1920, ch. XI) predict for animated educational film and the medium, and what does each prediction depend on?
  Lutz predicts: children will 'very soon' learn from film as well as be entertained; the possibilities of educational, travel and scenic film are not exhausted; colour cartoons will come, depending on a process that reproduces nature's full range of hues ('awaiting the inspired inventor'); with colour, animated paintings and gallery exhibitions of art works with movement; museums with projecting rooms and fireproof film libraries. He calls it 'fantastic dreaming' and argues that formerly visionary things became commonplace. He also states what the medium lacks now: it 'needs, for one thing, color'. This candidate only records what he predicted; comparison with later history was not done (zero-fetch batch).
• articleWhich statements in Lutz's chapter XI report existing practice or history, and which are conditional proposals, claims or predictions?
  Chapter XI is mainly a proposal in the conditional mood. In Lutz's text of the chapter (lines 4097-4371) 'would' occurs 32 times, 'could' 21 and 'will' 9, and there are 0 footnote anchors. The statements that report existing practice or history are few: Demeny's lip-reading apparatus (a note of 1892 appears only in the illustration credit), Desvignes' 1860 steam-engine pictures ('is recorded', no citation), animated maps 'frequently' used with news reels, educational/travel/scenic films shown in theatres (live photography), stencil tinting of photographic film, and a caption on drawings used in making a gasoline-engine film. None of these reports an animated educational film's audience, learning result or cost. An agent should therefore cite the proposals as Lutz's views, and the few reports as unreferenced assertions.
• articleHow does Lutz (1920, ch. XI) say an animated educational film should be planned and sequenced: scenario, order of parts, titles, schematic style, pointers?
  Lutz gives planning rules by example rather than as a rule list. The film "must be planned well" and a scenario would practically be written; titles and wording are thought out beforehand for educational value; for the muscle film the series of drawings is the first thing to prepare. His sequences go from the whole to the parts and from a fault to its correction: a sectional heart, then flow through the cavities, then a body-wide schematic with a colour code; a whole evolutionary tree, then separate drawings of branches; a poorly composed picture built on screen, faults pointed out with a cut-out pointer, then the components moved into the good arrangement. The chapter gives no durations, no number of titles and no simplification rule beyond 'schematic diagram like those in the books' (bounded scan).
• articleWhich subjects does Lutz (1920, ch. XI) say suit animated drawings rather than live photography, and what reason does he give for each?
  Lutz proposes animated drawings for subjects where the camera cannot show the thing: the heart and circulation, muscle action, the skeleton's levers, evolution as a tree; for machinery in operation, where a drawing replaces a static plan; and for change over time, such as changing maps. His reasons are (a) the camera cannot photograph some details, (b) the eye grasps a moving drawing in moments that paragraphs would take to explain, (c) a trained draughtsman can draw it, and (d) for maps and evolution, the subject is itself a change of a diagram. He keeps live photography for experimental apparatus and for physiological subjects where scientific moving pictures exist, and his art-method films film a demonstrator rather than animate. He does not name invisible, slow or microscopic phenomena as such (bounded scan of chapter XI).
• articleIn E. G. Lutz's 1920 Animated Cartoons, chapter XI, what does he count as an 'educational' animated film, and how does he separate instructional films from stories for children?
  Lutz gives two senses of 'educational film'. The strict sense covers only instructional films. A broader sense, which he credits to unnamed 'enlightened opinion', also admits any theme or story for children, even with a touch of humour; his example of that second type is Lewis Carroll's Alice tales as drawn by Tenniel, and he adds that such a story needs correspondingly better drawing than the ordinary comic cartoon. The examples after the opening paragraphs are instructional (art methods, physics, machinery, history, physiology, evolution). The classification is Lutz's own 1920 usage; he gives no test of learning for either kind.
• articleHow do Lutz's (1920) cartoon-plot and gag rules line up with the photoplay-manual comedy rules already in the batch (C3-B036-S10-05, -06), and what checklist can an agent use to plan a short animated gag sequence?
  Compared by candidate ID only: Lutz's cumulative chain and his remark that the early string of incidents no longer suffices match the photoplay manuals' failure mode of loosely connected incident chains (S10-05); his chase around an object as a 'never-failing' gag sits beside, rather than against, that warning if the chase is made to escalate. His anticipation-then-hit pattern (stone and hat) is compatible with Sargent's anticipation then surprise (S10-06). The comparison is limited to the prior rows' summaries and quoted phrases. The closing agent check restates Lutz's practice (labelled 1920 practice) and adds clearly marked proposals; none of it is tested.
• articleWhich failure modes of animated-cartoon comedy and plotting does Lutz (1920, ch. X) name or imply, and how good is his evidence for each?
  Lutz names four failure modes: story additions by an artist without the dramatic idea (confuse the story), too much business (retards the plot), lettered dialogue inside the drawings (breaks continuity), and a bare string of incidents or a single calamity (not enough). A fifth is implied by his pause argument: continuous laughter fatigues the audience. All are assertions by one author; the laughter-wave support is his recollection of a live-action film and of the hill-climbing gag, with no audience data.
• articleWhat does Lutz (1920, ch. X) say about the relation between how easy a gag is to draw and what the story needs, and does he advise choosing cheap-to-animate gags?
  Chapter X does not say 'choose cheap gags'. What Lutz states is narrower: because a good draftsman can do anything with his figures, the danger is overdoing business and slowing the plot; the spin effect is 'easy to produce'; dialogue balloons are an easy way to run up footage but are forgiven only for a good jest; and words belong on separate titles (also for export). The gags he describes (one drawing turned round, a cycle of drawings repeated, blur, a figure vanishing behind a house or hill) are, as a text-based inference, ones that need few drawings. He gives their reason as laughter and pacing, not effort. The mechanics of cycles and holds are in C3-B229-G70-01 and -03 and are not repeated here.
• articleWhat does Lutz (1920) say an animated-cartoon plot should be: how long, how simple, how many characters, and whether characters should recur?
  Lutz says a scenario (a skeleton, not a full stage-style script) is written first, and that by 1920 an animated cartoon must have a real plot: orderly parts leading to a main point or climax and an untangling. A comic scenario with two characters is easiest to tell; their clash is carried by a cumulative chain of mishaps growing in violence, ending in a calamity and an after-climax. The best cartoon is told by pantomime alone. Chapter X gives no film length and says nothing about recurring series characters; those are recorded as bounded negatives.
• articleWhat comic devices does E. G. Lutz (1920, Animated Cartoons, ch. X) say an animated cartoon should be built on, and which of his statements are claims rather than examples?
  Lutz's stated aim is 'a laugh in every foot of film', and he files the cartoon with extravagant farce. The devices he actually names in chapter X are mostly motion devices: reeling or spinning (blow victim, pinwheel fight, windmill, merry-go-round), freedom from the body's physical limits, regular repetition (a stream of figures, a parade), planned pauses (chase behind a house, hill-climbing) and anticipation completed by consummation (stone and top hat). He gives mechanical/periodic motion as the reason such gags work and claims they 'never fail'; no audience evidence is offered. In the part of the text searched, he does not use the words exaggeration or metamorphosis.
• articleWho does what in the 1920 animated-cartoon process Lutz describes: chief animator, assistant animators, tracers, photographers?
  Lutz names four roles: the chief animator, assistant animators, tracers and photographers. The chief animator designs the characters (front, profile, back, three-quarter), lays out the scenes, hands out scenes to assistant animators and keeps the most important one. The animator plans and draws what needs artistic judgement; the tracer repeats unchanging details; cels and pegs cut repeated tracing. The text does not say assistants draw in-betweens specifically, and this is one author's description of a small-studio method.
• articleHow does Lutz build a walk, and how does he use cycles and repeats of a few drawings (effects, galloping backgrounds)?
  Walk: draw the two outstretched extremes (A, B) one step apart, the middle passing position (C), then in-betweens (D, E); the second step reuses C, D, E as outlines with limbs swapped, so a walk is two alternating sets of drawings. Cycles: effects (smoke, water, stars) and animal gallops repeat 3-5 drawings in order; a misordered cycle can read as moving backwards; backgrounds move 1/16-1/8 inch per exposure to give speed and depth. The text and an illustration caption disagree on drawing counts and the figures were not inspected.
• articleWhat does Lutz (1920) say a key or extreme drawing is, what an in-between is, and how many in-betweens a movement needs?
  Lutz's method: draw a base ("key") sketch of the figures, draw the two extreme positions of the action on separate pegged sheets, then draw the intermediate position over them. He says one drawing between extremes can do for a swift comic action, and a slower action should have at least three; the head-turn illustration repeats that scale. He allows rough drawing of some in-betweens when the action is quick. His terms are "key sketch", "extreme" and "intermediate/in-between"; "breakdown" does not occur anywhere in the text.
• articleHow do 3D productions get the limited-animation (stepped, "on twos") feel, and what rationale and settings do primary sources document?
  The Xrd developer account (2015) is the primary rationale found: they tried full animation early, it did not read as 2D, so they removed key interpolation (every frame is a key), used about 500 bones per character, avoided simulation, and deformed the mesh on every key to add imperfection. Blender documents a Stepped Interpolation F-curve modifier (step size, offset, start/end) that holds values every N frames. No inspected source documents "on twos" frame numbers for a film; secondary reporting about Spider-Verse exists but no primary source was read (see negative results).
• articleHow do anime-look 3D productions edit vertex normals to get clean face shading, which methods exist, and what do they break?
  The Xrd developers state they edited normals on every major feature and hand-crafted face normals, and did not use normal maps. A Japanese ASW lecture deck (published 2025-07-30; glossed here) names three control methods: place vertices so automatic normals come out right, transfer normals from another mesh, and set normals by hand (reserved for faces; whole-body is called impractical), and lists limits: normals rotate with bones, no subdivision surfaces, extra care with hull outlines. Blender documents Normal Edit (with Max Angle to avoid shading artefacts, described as useful for toon-like shading) and Data Transfer (custom normals).
• articleWhat goes wrong with inverted-hull outlines in production (width vs distance and FOV, hard edges, custom normals, floating lines, fabric) and what fixes did a studio document?
  The ASW English deck (published 2025-11-28; 79 slides) lists concrete failure modes and fixes for the hull: scale width with camera distance and FOV (tan(FOV/2) factor, FOV from the projection matrix), control width with a vertex-colour channel (default 0.5), push the back-face mesh away from the camera to keep only the outer rim, use a second normal set for the hull when hard edges or customised normals break it (stored in the tangent channel at import), and set the width to 0 in valleys to close gaps that make lines float. Unity Toon Shader exposes parallel controls (width map, Z offset, distance fade, baked normal map for outlines). The deck says it avoids code and is engine-independent; the second-normal-in-tangent approach needs an engine import change.
• articleWhich outline method should I start with (inverted hull, post-process edge detection, Freestyle, Line Art), and what does each trade away?
  A practitioner comparison from Arc System Works (speaker Motomura, deck published 2025-11-28) sets post-process outlines (constant line width, lines at mesh intersections, little per-part control) against the inverted hull (per-area control, no intersection lines, thickness upkeep). Xrd used the hull for per-vertex control and in-DCC preview. Unity Toon Shader implements two hull-style modes (extrude along normals, or scale). Blender offers Freestyle (render-time) and Line Art (stroke-based). Epic's post-process documentation shows a CustomDepth/stencil silhouette; it is a general doc with no toon guidance. These are statements by tool makers and one studio; no source compares the methods on equal footing.
• articleIn Blender, what are the line options for a hand-drawn look (Freestyle vs Line Art with Grease Pencil), what does each produce, and what limits does the manual state?
  At Blender 4.5 LTS (manual path observed 2026-10-22), Freestyle is a render-time line engine that outputs lines as a render pass to be composited with Alpha Over, with documented limits (memory, no face-intersection edges, no Z depth). The Line Art modifier generates Grease Pencil strokes from a camera view, can be baked over a frame range, and has option-level fixes (overscan, chaining) for edge continuity. Shader To RGB (EEVEE only) plus Color Ramp is the manual's toon-shading hint. The manual pages read never mention an inverted-hull outline.
• articleHow does toon/cel shading actually work (a stepped ramp on lighting), and what are the first controls to set in Blender, Unity or a custom shader?
  Three primary sources describe the same core: lighting is reduced to a hard step. Motomura (Arc System Works, 2015) describes the Xrd shader as a step function of a threshold, a light vector and a normal, with artists painting the threshold per vertex. Blender documents a Color Ramp (Constant interpolation) after Shader To RGB (EEVEE only, with stated PBR costs) as a toon shader. Unity Toon Shader documents base/1st/2nd shade colours with step and feather controls and shade colours independent of light colour. The sources do not give numeric defaults; they treat the look as an art-directed choice.
• articleI loaded a PC2 file in Blender's Mesh Cache modifier: which Time Mode should I use, and what do Frame Start and Frame Scale do?
  At main commit f0e5cc32 (observed 2026-10-22), Frame mode treats the evaluated value as a sample index: index = Frame Scale x scene frame - Frame Start, and ignores the file's fps and start frame (manual and RNA tooltip). Blender's own PC2 exporter manual defines 'Sampling' in frames (0.1 = 10 samples per frame), and the only tracker record about this (#41118) says a cache exported at sampling 0.1 needs Frame Scale 10 to play in real time. So for a cache with s frames per sample, Frame Scale = 1/s gives real-time playback; Time and Factor modes are not recommended for PC2 until checked (candidate 03). This is a code and document reading, not an execution.
• articleIs there a Blender bug report, commit message or regression test about the Mesh Cache modifier's PC2 Time-mode fps handling (frame = ((time / fps) - start) / sampling) or its Factor-mode scaling?
  Scoped negative result. In the sources searched on 2026-10-22 (path-filtered commit lists for six Mesh Cache files from 2013-01-21 to 2026-06-15, GitHub commit search, projects.blender.org issue and PR search, and the tests/ tree at main commit f0e5cc32) none addresses PC2 Time-mode fps handling or Factor-mode scaling. No commit message mentions fps for these files except a mechanical FPS-macro replacement (2025-08-07). No test file for the Mesh Cache modifier, PC2 or MDD was found. The nearest records are issue #41118 (PC2 sub-frames, resolved as an exporter bug in 2014, mentioning Frame Scale), #60824 (units list naming Mesh Cache frame_start) and #34607 (MDD interpolation).
• articleDo Maya, Unreal Engine and Houdini document ML-based deformation or simulation tools, and do they replace or complement the traditional rig/solver?
  Yes for deformation, and in every page read the ML model is trained on the output of the existing rig or simulation and complements it. Maya's ML Deformer approximates a complex deformer for posing and playback or crowds and keeps the original deformer for final rendering; Unreal's ML Deformer approximates deformations at runtime from data made in Maya or Houdini and recommends Neural Morph for flesh-like and Nearest Neighbor for cloth, and Epic says Fortnite used it for muscles and cloth; Houdini (docs labelled 22.0) has ML Deform, an ML Deformer recipe that approximates an expensive flesh simulation, and an ML Volume Upres recipe for pyro. No vendor page read documents ML replacing a cloth or fluid solver; Houdini ML cloth was not found in pages read.
• articleDo learned physics simulators stay stable over long rollouts and generalise beyond their training data, and what do the sources say breaks?
  Both papers report plausible rollouts far longer than the training trajectories (GNS: thousands of steps; MGN: 40,000 steps from a model trained on 400-step trajectories) after training on one-step targets, and GNS credits input-noise injection for this. Generalisation claims are specific: GNS to about 10x more particles and unseen shapes, MGN to a 20k-node windsock and wider angle/speed ranges with RMSE 11.5 to 12.4/13.1. Failures are also stated: GNS rigid block deformation in vigorous shaking and a stuck-goop bad seed; MGN cloth decoherence after about 50 steps; a baseline GNS without mesh-space features is unstable on cloth. A later PyTorch re-implementation README warns accuracy may degrade over time and extrapolation in time should be avoided.
• articleWhat is a 'learned physics simulator' such as DeepMind's GNS or MeshGraphNets, what does it take as input and predict, and what is it trained on?
  In the two DeepMind papers a learned simulator is a graph neural network trained, per scenario family, on trajectories produced by a conventional solver (MPM, SPH, PBD, ArcSim, COMSOL, SU2). It reads the current state (particles or mesh nodes with a short velocity history), predicts per-node acceleration, integrates it with an Euler step, and is applied repeatedly to roll out a trajectory. It is a fast surrogate of the solver it was trained on, not a model that discovers physics; the training supervision is one-step prediction.
• articleHow do primary motion-generation papers define foot skating, penetration, floating and jitter, with what thresholds and units?
  Definitions differ by paper. PhysDiff (v3): Penetrate and Float are distances from the ground to the lowest body mesh vertex with 5 mm tolerance, Skate is mean horizontal displacement of foot joints in contact in two adjacent frames, Phys-Err is their sum, all in mm. GMD (v3) and Yi et al. (v1) use a skating ratio: fraction of frames where a foot slides more than 2.5 cm while foot height is under 5 cm. Body-momentum (v1) measures jitter as the derivative of acceleration in world coordinates (10 m/s^3) and foot sliding in mm during contact. MDM v2 uses a foot-contact loss in training, not an evaluation metric.
• articleWhat time interval did Muybridge state for his walking plate, and what does it imply in modern frame terms?
  Muybridge's 1887 prospectus describes plate 14: a model walking one stride, twelve phases about 0.120 s apart by chronograph, whole movement about one and a half seconds. Intervals in the series ranged from 1/60 s to several seconds and were timed with a 100 Hz tuning fork. Twelve phases at 0.120 s span 1.32 s (11 intervals), a little shorter than his stated 'about one and a half seconds', so the figure is approximate; a stride of about 1.3-1.5 s is slower than the 1.09-1.15 s in modern samples. Per-plate interval columns in the catalogue table were too garbled in the OCR to read.
• articleWhat do open papers measure for monocular and few-camera markerless capture against reference systems, what is the evidence for commercial tools such as Move AI, and what does it imply for animation production use?
  Open results are research-method numbers, not commercial-tool numbers: on Human3.6M (marker-based) a monocular lifting model reports 46.8 mm MPJPE from detected 2D keypoints (37.2 mm from ground-truth 2D); on in-the-wild 3DPW WHAM reports 57.8 mm; on EMDB global trajectories WHAM reports 354.8 mm W-MPJPE100, 6.0 % root-translation error and 4.4 mm foot sliding. A 2022 clinical review gives 40-60 mm joint-centre differences and up to 30 degree ankle-angle differences for monocular methods, and OpenCap (two or more phones) reports 4.5 degree joint-angle MAE against skin-marker capture, with the authors noting that skin-marker error itself limits what can be shown. No open independent validation of Move AI or Flow Studio mocap against marker-based ground truth was found (bounded Europe PMC search), and Move's own accuracy page gives no numbers. Implication (synthesis): treat markerless output as animation reference or blocking requiring cleanup of root drift and foot contact, not as ma...
• articleWhat footage will Autodesk Flow Studio (Wonder Studio) accept for character replacement, clean plate and camera tracking, and what shooting rules does the documentation give?
  At help.wonderdynamics.com pages stamped "Updated on October 19, 2026", Flow Studio documents hard limits (.mp4/.mov, 2 minutes, 15 shots, hard cuts only), partial support for occlusion, no support for object or actor interactions, a standalone Clean Plate tool (humans only, up to four actors, hidden background must be visible at some point) and a standalone Camera Track tool (no variable zoom; best with wide shots, parallax, distinctive surfaces). No accuracy figures are published on these pages; the limits are vendor-documented constraints, while claims such as the Live Action Easy page wording ("let our advanced AI handle all the heavy lifting") are vendor claims (E16).
• articleWhat can I give Move.ai (Move One single camera, Genesis multi-camera), what do I get back, and what capture conditions do the Move docs require?
  At docs observed 2026-10-22 Move documents two products: Move One (single phone camera or an uploaded .mp4/.mov/.avi up to 1 GB, 30 s free / 60 s paid clips, credit-metered per person-second) and Genesis (6-12 calibrated cameras, on-premise). Outputs are FBX/USDC/USDZ/Blender (Move One) and FBX/BVH/GLB/USDZ/Blend/C3D/JSON plus preview video (Genesis). Capture rules are practical (full body visible, locked camera, contrast clothing, no mirrors) and the docs publish no accuracy numbers; capability tables are vendor specifications, not measurements.
• articleHow do kamishibai makers design around the nuki (pull-out) reveal, and what transition and pacing rules do documented performers use that an animation or audio storyteller could borrow?
  A 2013 Japanese college bulletin on training childcare students to make kamishibai documents design consequences of the pull-out (nuki) mechanic: the cards are pulled from right to left as seen from the audience so figures walking forward must face left; the script goes on the back of the previous scene; margins hidden by the stage must be measured; students exploit pulling for stepwise reveals (a rainbow gaining one colour per scene) and for participatory riddles (a coloured paper interposed between silhouette cut-out and picture is pulled after the audience guesses). A US performance handout adds timing guidance (vary the pull speed: sudden cuts for surprise, dissolves for suspense) and cautions that overdoing character voices can steal focus from the art. No open source read names or describes specific sub-techniques of nuki.
• articleWhat is kamishibai (paper theatre), how are the cards written and handled, and what are the first things a new performer needs to do?
  Kamishibai is a Japanese picture-card storytelling form. A narrator shows a stack of illustrated cards (one picture per scene) in a small stage frame (butai) while reading text printed on the back of the cards. The scripts for each picture are on the back of the previous card, so the card the performer reads is not the one facing the audience; as each card is read it is slid out and moved to the back. Street kamishibai (gaito) flourished from the 1930s with about 2,500 performers in Tokyo in 1933; educational kamishibai grew alongside it. Practical guides here are Western educators' adaptations, not Japanese master accounts.
• articleHow does a professional rakugo performer say pauses (ma), first lines, word repetition and breath create understanding when there are no sets, costumes or narration?
  Katsura Utanosuke, a professional Kamigata rakugoka, explains in a 2011 talk (printed in a university repository) that rakugo has no scenery, costume or movement, so the audience must imagine 'what is not there'; the only means is speech technique. He defines ma as the blank time where no line is spoken and gives concrete rules illustrated on a short beginners' piece, 'Kuchinashi': set the tone with the first line, show two characters by contrasting pitch, vary repeated words to avoid boredom, pause before a keyword and hold the breath, pause again after a strange phrase so the audience absorbs it, and repeat an identical line but deliver it differently.
• articleWhat is a rakugo performance made of (makura, main story, ochi), and what does one seated performer with a fan and hand towel actually do in it?
  Rakugo is a seated solo comic or dramatic story performance in which one performer in kimono plays every character with only a folding fan (sensu) and a hand towel (tenugui). An official Agency for Cultural Affairs introduction and a 2019 speech-synthesis paper agree on the arc: an audience-warming makura (preliminary talk, often improvised), the main story (almost all character dialogue), then a closing line, the ochi (also called sage). The paper adds that Edo-style rakugo can also have an optional opening greeting (maeoki) and a conclusion (musubi) used instead of ochi in some stories. These are Edo/Tokyo-centred descriptions; the Kamigata (Osaka-Kyoto) tradition adds a desk and clapper (see -02).
• articleWhat does the MPEG_animation_timing glTF extension define and what governs its use?
  MPEG_animation_timing is a vendor extension (not KHR) in the Khronos glTF registry. It carries only a required pointer to an accessor of BYTE/SCALAR timing samples whose format and synchronisation meaning are defined in ISO/IEC 23090-14 clause 7.6.3, so the README alone cannot be implemented. It also requires MPEG_accessor_timed, MPEG_buffer_circular and MPEG_media. The README was registered for the first edition in February 2023, still cites the FDIS text, and has not changed since March 2023. Use is under the ISO licence and IPR policy, not a Khronos licence.
• articleWhich format should I use for a 3D handoff: glTF, USD, FBX, Alembic or OBJ?
  Each 3D format says what it is not, which gives a choice rule by payload. glTF 2.0 is for finished runtime assets and says it is not an authoring format. USD is for assembling and layering many assets into sets, shots and worlds, though UsdSkel does not provide general rig execution. Alembic is for baked, computed geometry such as simulation and animation caches, and is not rig storage or a native-scene replacement. OBJ is static geometry and materials, with no animation, skeletons, lights or camera positions. FBX (via the Autodesk SDK or independent readers like ufbx) carries rigged characters with skinning, blend shapes, lights and cameras. The rule itself is a proposal; what the receiving application imports still decides.
• articleI baked a cloth simulation in Blender: is the bake itself something I can hand to another renderer, or do I export a different file?
  A Blender cloth bake is Blender's own cache (inside the saved .blend, or as .bphys files in blendcache_[filename] beside it), protected until cleared and replayed identically; the manual does not present it as interchange. To hand cloth to another renderer, export Alembic: it stores the baked vertex positions and transforms, not the simulation setup, with no particle-system support (hair as zero-width curves, particles as points) and material names only as face sets.
• articleWhat are a skeleton/rig, skinning weights and blendshapes (morph targets), and what do glTF, USD and FBX each store?
  A skeleton is a joint hierarchy, skinning weights say how much each joint moves each vertex, and blendshapes/morph targets are weighted vertex offsets: glTF stores linear blend skinning via skins, joints and JOINTS_n/WEIGHTS_n attributes and morph targets as deltas; UsdSkel stores skeletons, skinning and animation but explicitly not general rigging/execution (control rigs, IK), so formats carry the deformation result inputs, not the control rig.
• articleHow do VRM 1.0 humanoid, expressions and springBone define avatar rigging and secondary motion, and what changed from VRM 0.x?
  VRM 1.0 requires a core set of humanoid bones (hips, spine, head and limbs) with chest/upperChest/neck optional though required in 0.x, and forbids zero or negative bone scale; its expressions rename the 0.x presets (joy->happy, sorrow->sad, fun->relaxed), add surprised, and let expressions override mouth, blink and look-at; VRM 0.x specifies metres, right-handed Y-up, model facing -Z and T-pose, constraints the 1.0 files read do not restate.

More