Driven by inexpensive data acquisition, storage, and compute cycles, commercial plant breeding is in the midst of a revolution in phenotyping. We will present 3 cases of video and images for plant phenotyping. (1) Maize yield is the product of component traits, such as weight per kernel, kernels per row, rows per cob, and ear and kernel shape. Tedious manual processes can be replaced by processing images of maize ears/cobs/kernels at a resolution of 1200 dots per inch. Core methods used to extract complex maize traits from images include windowed Fourier transform, counting, and Bayesian-eigen. Concordance of manual and automated measurements was high (R² of 0.76–0.99), and these high-throughput phenotypes have been used successfully for GWAS studies. Publicly available software for automated phenotyping has been used to analyze >250K images of ears/cobs and 5 million kernels collected over 4 years at multiple research institutions across North America. (2) Cellular-level phenotypes may prove to be a novel and rich source of information for early screening in plant breeding programs. Machine vision approaches are being studied to extract cellular level traits, including expansion and signaling, from digital microscope imagery. As storage costs continue to drop and microscope automation advances, machine vision phenotypes extracted at this level likely will become part of the breeding/screening process. (3) In contrast to high capital outlay systems, low-cost high-throughput maize imaging platforms, which acquire and stream images to a cloud provider are being developed. For one low-cost system, images contain 3 plants, a barcode to encode metadata, and a resolution standard. Software extracts traits such as leaf color, number of leaves, plant width and height, stem diameter, digital biomass, and leaf curvature. These complex traits are stored in JSON/XML format for downstream processing. In comparison to fully automated systems high cost systems, this phenotyping station costs less than $5K, is mobile, scalable, and can image ~100 plants per hour. Cost effective phenotyping to capture new and more complex traits is driving developments in automation, new analysis techniques, and high-throughput computing.