Week 6: March 8th, 2023

Working with: Runway ML AI Magic Tools, DALL-E

I worked with DALL-E before, as well as lightly with Runway ML’s background editor (via green screen video editing function). So, I wanted to try Runway’s Infinite Image tool and see how it compares with my experience with DALL-E.

This is my starting image:

At first glance, the image generation frames take in very minute details and blends the generations well with the existing image.

It also takes stylization preferences, as well as being able to blend different styles of images, such as by adding “surrealism,” “oil painting,” “animated,” “cartoon,” or “cute.”

I was also able to tailor size of object within the image generated to some degree, such as by specifying “small” when trying to generate a hummingbird or snowman.

I found Runway’s context matching to be more thorough and adaptable than in DALL-E. I was pleasantly surprised to be able to let Runway fill in most of the picture by being very liberal in my prompt. With DALL-E, I had abrupt changes in background instead of a smooth blend.