Generative AI

Whilst my professional work utilising stable diffusion has proven useful, the clients and projects are often limited in scope. Through these personal projects I have learnt how to properly engineer prompts, utilise control nets and LoRAs, upscale and work with inpainting and use multiple models.

Each one of these images were generated using models based on stable diffusion 1.5 or SDXL, rendered locally using the auto1111 interface. Generally speaking, each image took around 30 seconds to generate using an RTX3090 GPU.