Andrew Stevovich Web Development (2023)

The following is an abbreviated scope of work. A full itemized scope is listed at the conclusion of this article.

The goal of this redesign was to take on a genuinely comprehensive presentation of Andrew Stevovich's career, updating the technology, features, and ambition of the prior generation to a modern standard while building something capable of lasting and growing for as long as the work continues to exist. The site needed to hold an enormous amount of information and hold it well, presenting it with the visual clarity and breathing room appropriate to serious painting without the data ever becoming a burden to the viewer.

The design language is deliberately museum-esque. Clean, spacious layouts prioritize the paintings at practical large sizes. Images are dynamically sized to always present optimally based on orientation. The work is given room to exist on the page without competition from the interface around it.

The data architecture was one of the most substantial undertakings of the project. Designing a schema capable of holding the full complexity of a painter's career at this scale, across media types, periods, provenance chains, exhibition histories, publication records, and relational connections between works, required careful structural thinking. The schema was designed as part of a larger initiative into catalogue raisonné data architecture that has since become its own dedicated project, exploring the storage of artist corpus data free of any specific layout, in long-living, transformable structured data formats.

For this site the corpus was built from a combination of manual transcription, data processed from existing Google Sheets and Excel records through custom tooling, and automation strategies developed to handle the ongoing maintenance and growth of the archive. Python tooling was developed for corpus management, image processing, and color analysis. The data architecture and the website are treated as separate concerns: the corpus is the authoritative record, and the website is one layout expression of it.

Each artwork record is comprehensive. Fields include medium, glaze, varnish, signature location, dates, provenance, color data, perceptual color analysis, thematic categorization, and more. The site cross-references this data relationally: artworks display which reviews discussed them, which exhibitions featured them, which publications included them. Drawings show the paintings they were studies for. Reviews reference the works they address. Everything is connected.

A dedicated build step was developed to analyze artwork imagery for color data. This process extracts dominant color, average color, and color theme information from each work, combining direct mathematical color analysis with perceptual color interpretation leveraging machine learning to handle the nuance of how color is experienced rather than simply measured. This color data is incorporated into the site's records and used to provide color context and enable color-based sorting and exploration of the archive.

The site features elegant side menu navigation designed to function consistently across desktop and mobile. Settings accessible from the main menu include light and dark mode toggle, metric and imperial unit toggle for caption dimension displays, and an archival mode which surfaces substantially more detailed data fields for each artwork, suited to researchers and serious collectors.

The home page features a screen-size-responsive slideshow that gracefully introduces and transitions through paintings in random sequential order, selecting images appropriate to the viewer's screen ratio. The effect is immediate and considered, establishing the quality and range of the work before the visitor has navigated anywhere.

Artwork is presented through two complementary views. Gallery pages present work as beautiful, explorable grid displays that fill the screen with images, each selectable for individual viewing. List view pages present the same work as plainly captioned entries for methodical browsing. Both views serve different modes of engagement with the archive and are available across the major categories of work.

The custom image solution dynamically sizes images based on orientation to ensure optimal presentation across the large grid flows that characterize the gallery displays. Artwork detail pages are rich with relational data, contextual information, and the full depth of the catalogue record.

The site contains thousands of pages and was architected around O(1) scalability requirements consistent with a lifetime archive. Adding content carries no escalating build cost and no performance degradation. Updates target specific articles or content rather than requiring full rebuilds of the archive.

Pages are rendered at runtime through Lydio, my custom HTML rendering framework, operating over I/O rather than loading from disk. For large indexed content types such as artwork listings, reviews, and publications, Lydio templates render against Marle JSON documents at runtime, producing individual pages from single templates without any HTML build step. The site hot reloads from a data seed, assembling all relational links whenever the data file changes.

The server is built on Fastify, configured primarily as a static server, with a custom image router designed to serve optimally sized images for the site's large grid displays. Deep Frame provides authentication, error handling, security, and performance infrastructure.

A practical concern in designing a long-lived archive is longevity beyond the active maintenance period of its authors. In a time when both my father and I are gone I don't want the archive or presentation to be lost.

The site, despite being runtime rendered, was designed with an escape hatch: the server is capable of outputting itself as a purely static HTML site. In this form it can be self-hosted, opened as a local artifact, or hosted on free and persistent platforms such as GitHub Pages for as long as those services exist. Only minimal degradation of dynamic features occurs in this mode. The paintings, the data, the relationships, the writing, the record of a life's work: all of it remains intact and accessible.

This was a design requirement.

A fully responsive layout was implemented across all content types, display formats, and navigation systems to ensure consistent presentation across screen sizes and devices.

Structured metadata including Schema.org markup was implemented across all page and content types, with dedicated schema for individual artworks supporting rich indexing of the catalogue. Semantic HTML, descriptive metadata, and logical pagination support deep search engine crawlability across a large and varied archive.

The site was developed with attention to accessibility standards throughout. Clear semantic structure, descriptive image metadata, and consistent navigation patterns were prioritized across all content types and display formats.

All artwork displayed on the website is the work of Andrew Stevovich and remains his copyright. Reproduction rights are reserved.