I’ve had the domain luminome.com registered since about 2000. It’s long served as a platform for sharing work and exploring design approaches. In the downtime of setting up a new fine-art studio here in France, I thought it would be good to see if I could unearth and communicate the last decade or so of my computer graphic design projects. Included are some that were for clients and others that are in-house development work — genesis content for analog artwork. I work to stay current with all things internet development: I see it as another important language to learn.
The latest revisions to luminome.com (deployed about a week ago) include a view of my computer design work dating back to about 2008 with about 400 hi-res images! Go take a peek and come back for explanation.
Here’s how it came about
I decided not to use Python for this. Node seemed like a better fix because I needed to code for the whole stack. The steps involved were as follows:
Create a local Postgres Database with a minimal schema (filepath, date-created, 2 {…JSON} columns).
Traverse 2 archive volumes using the linux “find” command. Hilariously simple-minded approach. Basically find anything with a “.ai” extension (adobe Illustrator) ‘find’, [`${src}`, ‘-iname’, ‘\*.ai’]. And pipe those results directly into the database.
Create a frontend viewer that allows browsing the results (about 3000 in my case) by generating quick PDF previews of the work. This allows deciding which files are germane and/or relevant. I got the collection down to about 400 files.
Discover a nice utility called GhostScript which I find really reliable for getting good raster images from PDFs (not to mention optimizing PDFs as well, it’s super impressive). Use GhostScript to rip down all the selected PDFs to PNG raster images at a couple of different resolutions.
To deploy the project with minimal overhead, the images are hosted here at my namesake domain while the app logic is hosted at railway. The app fetches a static JSON version of all the relevant DB entries and displays them. I’m fond of this approach because it’s ostensibly serverless in-situ and simply leverages an existing server for the content.
Results.
Roundtrip was about 2 weeks of production time. For me the most important is getting this content in a place that it can be shared. Thank you for your time and all the best. —SAC