From the course: Learning ComfyUI for Stable Diffusion

Unlock this course with a free trial

Join today to access over 24,900 courses taught by industry experts.

Upscaling to increase resolution

Upscaling to increase resolution - Stable Diffusion Tutorial

From the course: Learning ComfyUI for Stable Diffusion

Start my 1-month free trial Buy for my team

Upscaling to increase resolution

“

- [Instructor] Once you've generated an image that you like, you will probably want to upscale it. The current state of the art with AI image generation will only give us results of about one megapixel, 1024 x 1024. By comparison, a high definition video image is about 2 megapixels. An ultra high definition video image or so-called 4K image is about 8.3 megapixels. We will want to upscale things in order to make them bigger and look better in print and so on. We could upscale things in lots of ways in Stable Diffusion and in ComfyUI. The most basic way is just the same as you would do in an image editing program like Photoshop. Simply scale up using bicubic interpolation. But a more sophisticated method would allow us to achieve larger ratios of upscaling. There are particular pixel interpolation models that will give us very good results at two to four times scale. That's what we're going to see here right now. So I've got a graph already that's generating an image of a hummingbird…

Contents