About & licences
Parallax is a projection-geometry teaching lab using simplified models. Not for clinical use. The models are not anatomically validated and the lessons have not been tested with learners.
The rig does not model machine travel or collisions. One view, or a pair that looks reassuring, is not a general test of containment.
Responses reset when you refresh. Only your colour-theme preference is saved. There are no accounts, uploads or analytics.
How the X-ray images are made
Each image is calculated through the 3D model, not taken as a screenshot of it. For every pixel, Parallax traces a straight line from the X-ray source to the detector and measures how far it passes through bone or wire. A longer path through material, or material that blocks more of the beam, makes that pixel brighter. Structures at different depths can overlap in one view and separate in another.
The archived whole-femur explorer’s outer shell is set to block more of the beam than its inner core; its Uniform option treats the whole bone as one material. The separate CT-informed proximal-femur lesson uses a heuristic shell and interior envelope with weak, smoothly varying derived density. Its wire adds stronger opacity. All views use the same brightness scale, without automatically adjusting the contrast of each image.
Both current lessons calculate new projections in your browser when you move the C-arm. The fracture lesson starts with a fixed AP reference. The archived projection-fundamentals lesson uses images calculated in advance. Looking around the 3D scene only changes your viewpoint; it does not change an acquired image.
The images leave out soft tissue, scatter, the range of X-ray energies, blur and detector noise. Pixel edges can shimmer. These simplifications help show the geometry, but do not reproduce all the features of a clinical X-ray.
Femoral wire position
This lesson uses a CT-informed proximal-femur model. It is a different model from the HFValid-derived shaft used in Guidewire across a fracture. The archived whole-femur explorer is separate from both. The 105 mm-long, 1 mm-radius wire represents a planned screw, with its tip standing in for the screw tip. Its 130° guide angle is relative to a measured local shaft direction, not clinically registered anatomy. Fixed wire positions and model-defined AP and lateral views still need expert review; the lateral view makes the neck look shorter.
The tip-to-marker distance is not clinical TAD. It measures to the same gold surface marker in both images and adjusts for magnification using the known depth of each endpoint. Dimensions are model millimetres, not independently validated anatomical measurements. The 33 mm head envelope used to check framing is not an anatomical head radius. The 25 mm combined-distance limit and 8 mm centring limit are rules for this exercise, not clinical grading rules.
Baumgaertner et al., 1995 studied lag-screw tip–apex distance and cutout in sliding hip-screw fixation. Its 25 mm finding is a teaching reference, not a guarantee of safe fixation. The indexed abstract was inspected, not the complete paper. This lesson does not assess fracture reduction, implant selection, stability or overall fixation suitability.
Guidewire across a fracture
Compare AP with a saved additional angle before deciding whether the guidewire enters the distal canal. The model uses an approximate central-shaft canal with artificial end caps, a 16 mm fracture gap and a 10 mm posterior shift of the distal fragment. The fixed straight wire and olive represent only a local segment, not an entire instrument or entry route.
The starting AP is calculated in advance; new C-arm views use an illustrative finite-source worker renderer. A saved non-AP view unlocks an answer, but does not certify sufficient evidence. This is not clinical fluoroscopy. Anatomical validity and learner benefit remain unestablished.
The outer surface and derived canal come from HFValid collection v3, Aldieri, La Mattina, Szyszko, Baruffaldi and Viceconti (2025), under CC BY 4.0, not MIT. Parallax derived a shaft canal, transformed coordinates and authored the fracture and wire presets. Raw CT is not bundled; no source-author endorsement is implied. See attribution and changes and provenance.
Original work
Original code, documentation and procedural teaching assets use the MIT licence. The imported bone models and their derived cores, envelope and density are excluded.
Three.js
The 3D scenes use Three.js 0.186.0: MIT licence · version record.
Estate theme & fonts
The shared Sangeev.me theme is @sangeev/estate-ui 2.0.0-alpha.7, under its MIT licence. Literata and Atkinson Hyperlegible Next are bundled under their SIL Open Font Licences. All runtime assets are served locally; no font service or CDN is used.
Archived experiments
The earlier cylinder lessons and whole-femur explorer remain available in the archive. They are not part of the two-lesson sequence. Projection fundamentals retains its diagonal-breach exercise; it has not been replaced by a claim that two orthogonal views always resolve uncertainty.
The cylinder and wire are mathematical shapes. The free Femur explorer uses the whole BodyParts3D left-femur surface with an authored inner core, not a recovered tissue segmentation. Its two image models use the same view and fixed intensity mapping. It remains experimental.
BodyParts3D, © The Database Center for Life Science licensed under CC Attribution 4.0 International. Redistribution relies on the current publisher grant, not MIT. The original OBJ comments retain the older CC Attribution-Share Alike 2.1 Japan notice: see attribution, changes and licence discrepancy and model provenance.
CT-informed proximal femur
The femur lesson derives from the Norm_02 subset of Dual Fluoroscopy In Vivo Pelvic and Hip Joint Kinematics in Patients With Cam Femoroacetabular Impingement Syndrome, by Andrew E. Anderson, Penny R. Atkins, Niccolo Fiorentino, Rich Lisonbee, Jared Zitnay and Lindsay Schuring. Published 2025-09-30 by The Hive: University of Utah Research Data Repository. The derived outer surface, interior envelope and density retain Creative Commons Attribution 3.0 Unported (CC BY 3.0), not MIT. No endorsement by the source authors or repository is implied.
Parallax converted coordinates and rasterized the surface, reconstructed a restrained interior envelope, and masked, smoothed and quantized CT-derived density. Source CT resolution is approximately 0.685547 × 0.685547 × 1 mm; resampling to 1 mm adds no acquisition resolution. Raw CT is not bundled. See attribution and changes and source identifiers, hashes and methods.
Anatomical status remains partial. The imposed 1.8 mm inset and heuristic envelope are not measured cortical thickness or an accurate tissue segmentation. The density is not resolved trabecular anatomy or calibrated attenuation. Artificial end caps and source truncation remain. Sampled entry-rim checks do not certify the continuous entry disk; the continuous interior-course bound excludes the first 3 mm entry collar. Numerical checks do not establish clinical safety or learner benefit.
The source package reproduces projections from the supplied meshes and keeps the earlier procedural model’s numerical evidence separately. None of these models has anatomical or clinical approval.