Articles on: Use Cases

How do I make Pixel Art?

Currently, there's no easy or direct way to make pixel art using a text to image genAi program. This is because pixel art relies on the grid of pixels traditional digital images use as reference. However, Scenario uses a Latent Diffusion model. With Latent Diffusion, the training images are compressed into "latents", which can almost be thought of as the echo of the original image. Those are used as a way for the program to infer possible images to create.

We can generate a very similar pixel-esque style, which is similar in many ways to the avenues that we might take when we try to generate 3D art. However, a pathway to a usable and general pixel model has not been fully forged - but we're working on it!

Updated on: 22/01/2023

Was this article helpful?

Share your feedback


Thank you!