Anatomy Movements Approximate body shape
SOURCE EXTERIOR / REPRESENTATIVE SHAPE
Explore an approximate body shape. External outline only · no fitted internal anatomy
Fit body view
BODY PARTS 3D / MALE REFERENCE
Explore the body. Layer by layer. Static male reference · 713 selected source objects
AB09 / SOURCE MODEL + RECORDING
Study the recorded squat. Source bone display · first observed squat sequence
Recording Observed squat High-knee toe-jog Fit current pose Reset
19 source display geometries · 2 facets are boundary references, with no filled patch
AI-SEGMENTED INDIVIDUAL REFERENCE / 001
Explore the torso-fat surface. A separate donor · source label 35, not the anatomy body
Fit reference Reset Opacity 100%
Source Torso-fat · individual AI-derived reference, not a personal fat estimate
MOTION CAPTURE / ACTUAL RECORDING
Watch the recordedwalking landmarks. 24 skin markers · illustrative connections, not bones
Loading source anatomy…
Reset view Focus selected Isolate
Layers ↗ Reset landmark view Follow motion
Actual source translation retained · skin landmarks, not anatomical mesh motion
Geometry stays still. The timeline changes recorded signals, not anatomical pose.
INSPECT THE ANATOMY
Find a structure Isolate reveals a selection beneath other layers.
Select a source structure.
Body-wide partial static reference: 713 selected source objects/parts, with many unresolved structures. Female internal anatomy, complete tissue systems, same-body fat and anatomical motion remain unfinished.
RECORDED SQUAT / AB09
One source model. Both recordings use the same AB09 source model and bone display geometry. Each moves it with that trial’s stored body poses and own recording clock. These are separate from the static anatomy and walking records.
The first observed squat sequence runs from 0 to 3.250 seconds, with standing-like context before the descent and settling after the rise. It is not established as a unique complete repetition. External load is unknown.
Source model and motion limits A source-scaled demonstration model: coarse feet, no patella and a combined torso. The source model also has limitations at high knee flexion. Numerical replay fidelity does not establish clinical pose accuracy. Two source polygons retain only their original boundary edges; their interiors are not invented.
There are 651 stored filtered pose frames. Playback selects the nearest source frame; slow playback changes viewing time. No blended loop, muscle deformation, simulated activation or atlas registration.
Inspect the two boundary references Unresolved interiors remain unfilled: skull source polygon 1160 and spine source polygon 433.
Source boundary Skull · polygon 1160 · 10 edges Spine · polygon 433 · 6 edges Locate boundary Boundary reference unavailable until the source loads.
Locate isolates the existing boundary and changes only the view. Fit current pose restores all source surfaces.
Actual scene numerical readback Current rendered body matrices and point 0 of each loaded geometry, in metres. This is technical source checking, not clinical validation.
Scene unavailable Model, clip and rights ledger Reload selected recording SOURCE-NATIVE POSE / SAME-RECORDING RAW EMG
Recorded time 0.000 s Source row 0 ▶ Play source poses ← Reverse Playback 0.25× 0.5× 1×
Recorded sensor pair Tibialis anterior Rectus femoris Biceps femoris group Gluteus medius Medial gastrocnemius Vastus lateralis Gracilis Gluteus maximus Left Right Source recording loading…
What these signals mean Unfiltered raw voltages share the pose recording’s timestamps. Acquisition at 1,259 Hz was published after upsampling to 2,000 Hz. Biceps femoris sensor heads are unresolved; medial gastrocnemius is distinct from the walking record’s lateral head. Amplitude is not a comparison of effort across muscles. Clipping cannot be established without hardware limits. All 6,501 stored samples are retained and drawn ; cursor values use the exact raw sample at the source pose time.
DONOR 001 / SOURCE LABEL 35
Torso-fat, in its own frame. This is the source’s Torso-fat label, a separate individual reference. It is not automatically visceral fat, skin or a layer registered to the anatomy body.
Source-reported donor: male, 78 years · 1.7018 m · 70.7604 kg. The scanned extent is not the donor’s measured height.
AI segmentation supplies the boundaries. This view preserves the source-derived surface and every fragment; it is not a clinical ground-truth boundary or body-composition estimate.
Source and surface method Verified Version 3 segmentation (23 September 2026), with the clinical workbook linked by that version. Source filenames were corrected in Version 3; this individual join was checked against that release.
Full-resolution case 001 / label 35. Source spacing: 0.9765625 × 0.9765625 × 2.344 mm. The source scan posture and physical frame are retained. The surface has 1,808,184 triangles and 72 retained digital-grid contact edges: a derived non-manifold reference, not a regular solid or a physical tissue-junction claim.
Level 0.5 Lewiner extraction, step 1, no smoothing, filtering or decimation. Helper vertices follow that algorithm’s construction. Source affine and winding are audited; no second mirror or anatomy registration.
Actual loaded material readback Loaded materials unavailable Material alpha is a display setting. Dense overlapping fragments do not provide calibrated whole-image optical transmission.
Source and conversion ledger · Primary collection ↗ Reload fat reference RECORDED WALKING LANDMARKS
One shared raw clock. 24 skin-marker trajectories and eight bilateral raw EMG channels use source timestamps. This view plays the actual recording in seconds.
Points are recorded landmarks. Thin connections help read their positions; they are not bones or tendon paths.
Raw EMG is signed voltage, not %MVC or force. It does not colour the anatomy.
The static anatomy reference and processed right-stride study use a separate clock. Their motion registration remains unvalidated.
Coordinates & source Source millimetres → metres, with the fixed display remap (X,Z,−Y). Ground translation remains intact; the default camera follows the recorded position without changing the landmark coordinates. Switch Follow motion off to fit the whole trajectory. No guessed heel-strike alignment.
Raw array metadata and hashes WALKING / RAW SIMULTANEOUS RECORDING
Landmarks + signed EMG 0.000 s Participant1 / V4 / T2 ▶ Play landmarks ‹ › Playback ¼× ½× 1× Reverse raw
Raw channel signed V · simultaneous raw source time
0 s 3.02 s Source200Hz markers /2000Hz EMG · shared timestamps · playback stops at the boundary, no endpoint smoothing
WALKING / ONE MEASURED STRIDE
Right-leg surface EMG Stride phase 0% 4 km/h · reference participant0.000 / 1.040 s ▶ Play recording ‹ › Playback ¼× ½× 1× Reverse
Tibialis anterior Surface EMG envelope · %MVC
0 100 0% 100% Moreira 2021 · Participant1 / V4 / T2 / right St1 · source-normalised %MVC
This hosted copy includes the active browser study and its licences. Editable source and review evidence are retained in the original archive.
× SOURCE NOTES
Source references, An earlier donor torso-fat study is retained in the original archive. It is not part of this hosted three-mode copy.
These are authentic BodyParts3D meshes and separate recorded datasets. Educational Movement poses the same partial source anatomy with an authored visual rig; it is not registered to the recorded walking or AB09 models and does not show measured activation.
Anatomy & rights BodyParts3D release 4.0, official ISA obj_99 source: 99% polygon reduced. Raw source units are millimetres; the editable Blender source and GLB explicitly convert to metres. Smooth shading and study colours are derived. No surfaces were subdivided or fabricated.
BodyParts3D, © The Database Center for Life Science, licensed under CC Attribution 4.0 International. Original OBJ comments retain the historical BY-SA2.1Japan notice; the current 2025 archive grant and original source bytes are preserved.
Current official asset licence ↗ · Source/member/hash ledger · Editable Blender source Selected source coverage 198 bone objects, 360 muscle objects or parts, one skin surface, 34 cartilage/disc objects, 17 other-connective objects, 30 nerve/ganglion references, 71 artery references and two lacrimal gland references. These 713 source-object/part counts are not normal human structure totals. The prior 67 unresolved objects, seven first-tissue exclusions and 97 further held records are not rendered. Every added tissue category remains partial.
Excluded coverage includes extensor carpi ulnaris, flexor digitorum superficialis, digastric and hyoid, plus intercostal/pelvic representations, hand sets and foot sesamoids. Their part/group identities or overlapping representations need further review. No fat layer or verified female body is supplied.
Added cartilage/disc, other-connective, neural, artery and lacrimal reference surfaces use the same BP3D static male source frame. The nerve/ganglion and artery layers do not map every peripheral path or vessel, and two lacrimal objects do not make an organ atlas. No new layer carries recorded activation, flow or secretion. In Educational Movement these same selected source objects use approximate source-derived rig weights. They remain separate from the donor-fat, exterior and recorded movement models. One original smooth-normal corner on the right stylohyoid ligament is retained as a source-lighting limitation; the 103 new objects have no such exception. Original topology is retained; source positions are converted to metres and float32.
103 selected objects and 97 held records · 103-source fidelity check · Editable 103-object source · First added-tissue selection and seven exclusions · Independent source fidelity check · Editable added tissue source Excluded source names and IDs Complete selection and exclusion ledger · Source-to-export coordinate audit · Whole-body editable Blender source Educational Movement source and limits Fourteen authored movements, including illustrative walking and running cycles, pose the same 713 partial BP3D source objects and eight anatomy layers. Walking and running loop at an authored seam; the other twelve examples stop at their endpoint. Rig pivots and visual weights are reconstructed for study, not measured attachment footprints or clinical mechanics. Warm role colours represent reviewed qualitative exercise roles; cool colours mark passive stretch targets. Unhighlighted muscles may also participate. This phase is not recorded time, and no colour gives EMG magnitude, force, strain or personal activation. The BP3D poses are not registered to the Moreira or AB09 recordings.
The source-rest GLB geometry and FJ/FMA IDs are retained. The editable source includes the visual armature and actions; arbitrary fractional Blender action frames can differ slightly from browser quaternion interpolation. A complete clean-room rebuild has not been certified.
Current rig and visual-weight ledger · Fourteen authored movements and source roles · 713-object visual mapping · Matching editable Blender source · Construction and rebuild limits The recording Source-plane angle values loading…
Participant1, male,25 years,1.81m,66.2kg; walking4km/h. Selected right St1 is1.04s, with1001 paired angle/envelope samples. A300ms RMS envelope and MVC normalisation were applied by the source study. The eight columns in the processed right file represent four muscles across two strides, not eight simultaneous bilateral channels.
Left same-phase channels remain unavailable. BF is a surface recording group; no independent short-head signal is shown. Playback speed changes viewing time, not the physical walking speed. The source’s unequal start/end values are preserved; playback stops at the end.
Primary walking study ↗ · CC0 dataset record ↗ Local software & next stages Application code is MIT; Three.js r180 is MIT. All runtime files are local. The earlier recorded-landmark, donor-fat, and recorded-model views remain preserved in the prior portable package and source files; they are not loaded as main study modes here. The separate Moreira stride signal panel remains in Anatomy and is not a measurement of these educational poses. Entered measurements are processed locally. No analytics, accounts or uploads are used. This hosted copy includes browser assets and attribution; editable and raw sources remain in the preserved archive.
Complete anatomy, sex-specific internal layers, complete fat/measurement fitting and validated recorded anatomical deformation remain unfinished. Movements uses authored reconstructed poses and qualitative muscle roles; its clip phase is not measured time. Illustrative walking and running do not show recorded ground-contact timing, speed, or activation. The static anatomy skin is one source reference. The separate Approximate body shape mode offers estimated exterior fitting; no internal anatomy is fitted. Source-backed educational relationships describe possibilities in study conditions, not a diagnosis or computed injury cascade.
Application licence · Asset/data/vendor notices · Exact candidate identity