
After a lot of trial and error I got python to make a sprite sheet out of MNIST pixel array data.

After a lot of trial and error I got python to make a sprite sheet out of MNIST pixel array data.

Desperately trying to figure out why the background of my 9's are not transparent.

Buster Benson's "Life in Weeks"
From: https://buster.wiki/me/ via https://twitter.com/buster/status/1080530097868701701

The printed versions of my generated snowflake card are here. My first printed project saved directly from a canvas element. Slowly making my way towards a photoshop free workflow.
From: https://beta.observablehq.com/@grantcuster/generative-snowflake-card-explorations

Lots of great stuff – especially for what I've been thinking about lately – in this interview with Folkert Gorter and Jon-Kyle Mohr.
From: http://v-e-n-u-e.com/Life-Online-An-Interview-with-Folkert-Gorter-and-Jon-Kyle-Mohr via https://twitter.com/jakedowsmith/status/1076210819157975045

I'm thinking about Homecoming's use of aspect ratio and how it could possibly translate to web design.

We got the printed copies of the federated learning report today.

The clothes on the character on Lynn Fisher's site change with the screen size. I'm very interested in designs where the screen size (and the layout differences that flow from it) are made more visible to the viewer.

I worked out the scroll so it automatically centers the text walk.
From: https://beta.observablehq.com/@grantcuster/text-walk-scroller

This holiday card should be a good way to see if my ideas about how to go from canvas element to print will actually work.

I made some snowflakes. I think I want to put them in a hexagonal grid next.

Something is off in my math here, but symmetry makes everything look intentional.

I'm researching procedurally-generated snowflakes.
From: https://mathematica.stackexchange.com/questions/39361/how-to-generate-a-random-snowflake

I finished the prototype demo video player for the prototypes we've made at Fast Forward Labs. We wanted something we could project at events/booths. It's a good overview of the machine learning prototype work we've been doing for over 4 years now!

Making a video player and figuring out some sizing. I'm continuing to explore doing almost all the layout as math in the render function.

So justified photo layouts are done by turning layout into a graph-traversal problem, and that approach is based on a dynamic line-breaking algorithm Donald Knuth and made for laying out text in TeX.
From: https://medium.com/google-design/google-photos-45b714dfbed1

Moved part of my blog over to a new, grid-based layout which shows the browser-size at the top. Lots to clean-up but enjoying the move so far.

Word Hy-phen-a-tion by Com-put-er by Franklin Mark Liang. (The text layout research is going great.)