Graphics, Image Processing and Animation Design
What this domain tests
Image model choice, file format properties, the classification of image-processing operations, and animation technique. Items ask what an operation does to pixels, which format suits a requirement, or which animation method produces a motion.
Core rules and concepts
- Raster versus vector. Raster images are pixel matrices, resolution-dependent, with file size proportional to area; vector graphics are geometric descriptions, resolution-independent, suited to logos, diagrams and type.
- Formats. BMP is uncompressed; TIFF is lossless and print-oriented, supporting CMYK and layers; JPEG is lossy DCT with no alpha; GIF is 8-bit with LZW, 1-bit transparency and animation; PNG is lossless with full alpha and no baseline animation; SVG is XML vector; WebP/AVIF add modern compression with alpha.
- Point operations act on each pixel independently: brightness/contrast, gamma correction, thresholding, histogram equalisation (global contrast redistribution). Neighbourhood operations use surrounding pixels: mean or Gaussian smoothing, median filtering (best for salt-and-pepper noise), Laplacian sharpening, Sobel/Canny edge detection.
- Geometric operations are affine transforms (translate, rotate, scale, shear) requiring interpolation - nearest neighbour (blocky but crisp), bilinear (smooth, slightly soft), bicubic (sharper, more ringing).
- Animation. Frame-by-frame versus tweened animation, keyframes plus interpolation and easing; onion skinning; principles such as squash-and-stretch, anticipation and follow-through. The 3D pipeline is modelling, UV unwrapping and texturing, rigging, animation, lighting, then rendering (rasterisation for speed, ray tracing for accuracy).
Worked example
Enlarging a 200×150 screenshot by four times. Nearest neighbour copies each source pixel into a 4×4 block, preserving crisp text edges but producing visible stair-steps. Bilinear averages the four nearest pixels, smoothing the staircase but blurring thin strokes. For screenshots and pixel art the correct choice is nearest neighbour; for photographs, bicubic is usually preferred. Stating why - the content's structure and the tolerance for blur - is what earns the marks.
Traps that cost marks
- Expecting JPEG to keep transparency; it has no alpha channel.
- Confusing dpi with pixel dimensions when computing print size.
- Describing histogram equalisation as a smoothing filter; it is a global point operation.
- Using a mean filter on salt-and-pepper noise, where the median filter is the right tool.