By Figure31

Lyam is a low-fidelity generative NFT collection of 340 images

All images are made from a different scene frame of Alain Renais' 1961 Last Year at Marienbad. Each frame is autonomously reduced to 10 different inverted grayscale values on a 4 x 10-pixel grid. This grid is then scaled up 1000 times using a bilinear interpolation algorithm. The machine fills the void created by the initial loss of information.

Each artwork can be recreated using only its title found in the token metadata. The title follows this format: LYAM_serial number_grid format_grayscale values_scale_interpolation algorithm.

To recreate an artwork follow these steps

  1. Make a 10 x 4 pixels grid at 1 px/in using the grayscale values (0-255) found in the title
  2. Scale up the image 1:1000 using a standard image processing bilinear interpolation algorithm

Figuring out the right order of the grayscale values will be a trial and error process. It can potentially have an impact on how the artwork will look, just like how the characters in Last Year at Marienbad are confused about their memories.

The collection was made with banal tools in order to enter into dialogue with other artists and collectors on what makes art generative or not, and what is needed to make quality art. All NFTs of the collection are minted on OpenSea without any specific alteration to the smart contract as a gesture of contradiction with the prevalent idea that good art has to be made with a complex technical apparatus.

Lyam on OpenSea

Thread about Lyam

The smallest denominator of these images is their title. Art isn't defined by its tools. This "total work of art" is a statement of independence.