The biggest misconception about AI is that we lack control.
With tools such as Midjourney, DALL-E 2 and Generative FIll by Adobe - it is no surprise. Although they can create stunning images, their design focuses on simplicity of use, which consequently imposes limitations.
From perspective of control, Stable Diffusion is the ultimate solution. Although the learning curve is steep, I gained full control over my design. With a flexible process, the possibilities are truly endless.
Here Is What's Possible
Logo Remix
Starting with an existing logo, we are able to make a redesign that closely resembles the original.
Style Remix
We can closely follow the fine details of an artwork when generating varieties. This way we only affect the style. Please note these are 512 X 512px.
High Detail Remix
Exporting at 1024x1024 gives us plenty of detail for potential upscales. While the ones above are are for restyling ideas, the examples below are much more polished.
Upscale
Supported up to 8K resolution.
Stylized QR Codes
Currently works at small scales or if we back up a good amount from the QR code.
Sketch + SVG Support
A rough sketch allows us to get variations of logos with similar compositions. We can configure to follow the sketch extremely close, or keep it vague. Exporting as SVG saves us a lot of time in post.
Image to Video
By creating a depth map of an image - we are able to use AI to turn it into a 3D model for animating. This approach is very limited right now and only allows for small movements.