Resources

Frame rate is the number of image frames produced or presented each second, usually expressed as frames per second (FPS). In interactive AR, it helps describe how frequently the visual scene updates.

Tracking drift is a growing error in an estimated position or orientation that makes virtual content gradually lose alignment with its intended physical location.

In AR and XR, a pose describes where something is and how it is oriented relative to a reference coordinate system at a particular moment.

WebGPU is a web API for graphics rendering and general-purpose computation on supported GPUs, giving applications explicit tools for organising GPU resources and work.

Real-time rendering creates images from a scene quickly enough to respond during an interactive experience, updating the view as the camera, objects or other inputs change.

WebGL is a browser graphics API that lets web applications draw hardware-accelerated 2D and 3D graphics in a canvas without installing a browser plugin.

Three.js is an open-source JavaScript library for creating and displaying 3D graphics on the web. It provides tools for scenes, cameras, geometry, materials, lighting and animation.

In AR tracking, jitter is a rapid, unwanted fluctuation in the estimated position or orientation of content, making an overlay appear to tremble against its physical reference.

Yaw is a change in heading: a left-or-right turn around the up axis of a chosen reference frame. Which coordinate axis represents that turn depends on the convention being used.

A joint is a transform in a digital skeleton that helps define a pose. Moving or rotating joints can carry connected parts and, through skinning, deform a character’s surface.

Mesh compression encodes geometry data more compactly for storage or transmission, with decoding needed before the geometry is used.

A draw call asks the graphics system to render a batch of geometry using a particular set of rendering state.

Z-fighting is a rendering artefact in which overlapping surfaces have depth values too similar to be resolved consistently.

Render order is the sequence in which a renderer submits objects or drawing operations to build an image.

A depth buffer stores depth values for rendered samples so the renderer can compare incoming surfaces with what is already there.

Alpha blending combines a surface’s colour with the colour already behind it, using alpha to control their contributions.

Alpha masking compares alpha with a threshold, discarding samples below it and keeping those that pass.

Backface culling skips polygons classified as facing away from the camera, rather than drawing both sides.

A normal map stores directions in a texture to change a surface’s lighting detail without changing its geometry.

Image-based lighting uses an image of an environment to supply lighting information for rendered surfaces.

An unlit material displays its base colour without calculating a response to scene lighting.

3D asset optimization reduces the cost of loading and displaying an asset while preserving the detail and behaviour its experience needs.

Basis Universal is a texture-compression system whose encoded textures can be transcoded into formats supported by different graphics devices.

KTX2 is a texture container format that stores image data and information needed to deliver textures to graphics applications.

Meshopt compression encodes 3D buffer data compactly, with a design that supports fast decoding in real-time applications.

Draco is a geometry-compression technology used to reduce the encoded size of 3D meshes and point clouds.

An animation clip is a named or grouped sequence of animated property changes that can be played over a defined duration.

Polygon count is the number of faces in a mesh; for real-time delivery, the exported triangle count is often the more useful comparison.

Texture compression reduces the storage or graphics-memory cost of texture data, with trade-offs in quality and compatibility.

Surface normals are directions associated with a surface that help the renderer determine how it should be shaded.

Physically based rendering is a family of shading methods that approximate how light interacts with materials using physically motivated rules.

Skinning binds a mesh to a set of joints so that moving the joints deforms the surface.

A pivot point is the reference point around which a rotation or scaling operation takes place.

Level of detail uses different representations of an object to match the amount of detail needed in a particular view.

Retopology rebuilds the face layout of a mesh to make its structure more suitable for animation, editing or efficient rendering.

Parenting makes one object or node a child of another so their transforms can be combined hierarchically.

Spatial audio uses information about sound sources and a listener to create the impression that sound comes from positions or directions in space.

Spatial mapping builds a digital representation of the surrounding physical environment for use by a spatial application.

Raycasting sends a mathematical ray from an origin in a direction to find intersections with geometry.

Hit testing checks where a ray or other query intersects a target, often to find a candidate location for AR content.

Plane detection estimates flat surfaces in a physical environment, such as a floor, tabletop or wall.

A keyframe stores the value of an animated property at a particular time.

Interpolation calculates values between known samples, such as the positions or rotations stored in animation keyframes.

Easing shapes how an animated value speeds up or slows down over the course of a transition.

Rigging creates the controls and relationships used to pose and animate a 3D object or character.

A 3D coordinate system describes positions using an origin and three axes, usually called X, Y and Z.

Local coordinates describe a position or direction relative to a particular object or parent, rather than the whole scene.

World coordinates describe positions and directions in the shared reference frame of a scene.

An object origin is the zero point of an object’s local coordinate system; it can sit inside, outside or at the edge of its geometry.

A reference grid is an editor guide made of regularly spaced lines that helps judge alignment, spacing and scale.

A reference plane is a flat geometric guide used to describe placement, alignment or movement in a 3D scene.

Perspective projection makes objects appear smaller as they move farther from the camera, creating familiar depth cues in a flat image.

Orthographic projection uses parallel viewing rays, so an object’s displayed size does not shrink simply because it is farther from the camera.

A vanishing point is the location in a perspective image toward which the projections of a set of parallel lines appear to converge.

A 3D transform describes how an object is positioned, oriented and scaled relative to a reference frame.

Rotation changes an object’s orientation around an axis and a pivot point.

Translation moves an object by an offset without, by itself, changing its orientation or shape.

Scale is a transform that multiplies an object’s size along one or more axes.

Real-world scale is the relationship between distances in a digital scene and measured distances in the physical world.

A material defines how a rendered surface looks and responds to light, using settings and, optionally, textures.

A texture is image or sampled data used by a material to vary a surface property across an object.

UV mapping assigns 2D texture coordinates to a 3D surface so the renderer knows which part of a texture belongs at each point.

Relocalization is the process of finding a device’s pose again within a previously established spatial map or reference.

A scene graph organises scene elements and their relationships, often as a hierarchy of parent and child nodes.

Prepare a reference image, a visual layer and optional audio for Painta. Learn current file choices, chroma basics and what to check before your first AR work.

Eligibility, entry steps, judging, prizes, payment, authorship and privacy for the Painta Instagram Art Challenge 2026.

Build your first AR artwork in Painta: upload a reference, add a video or image, align the layer, check playback and save a scene ready for a real phone test.

Improve AR image tracking with a sharper reference, consistent cropping and practical phone tests. Diagnose recognition failures, drift and alignment issues.

Publish your Painta artwork, review preflight checks, test the public viewer and download a QR code. Prepare a gallery label that visitors can actually scan.

Remove a plain video background with Chroma Key in Painta. Choose the right colour, adjust the threshold and check moving edges before testing the AR result.

Arrange transparent images at different depths in Painta. Align cut-outs, inspect overlaps and organise the scene while keeping the physical artwork readable.

Add narration and background audio in Painta. Set timing, balance volume, use the speaker controls and test how visitors start sound in the public AR viewer.

Animate a layer in Painta using three keyframes. Record starting, middle and final poses, check timing and understand how the playhead affects your changes.

Add a prepared GLB model to Painta, check textures and orientation, adjust scale and depth, and test the full model or animation against your physical artwork.

Create interactive text in Painta with a website tap action and an optional hotspot. Check readability, timing and the destination in the public phone viewer.

Improve mobile AR loading and playback by simplifying video, images and 3D models. Compare real phone tests and preserve the visual detail your artwork needs.

Reuse keyframes when updating a layer in Painta. Copy and paste selected attributes, keep your new media and follow the step-by-step MedusASCII walkthrough.

Build your AR profile and personal brand in Painta. Improve your bio, statement, links and portfolio with real examples from Bernie Torras’s public profile.