Two parameters in stable diffusion webui are denoising (similarity to source) and cfg scale (adherence to prompt). img2img does as it sounds, and inpainting allows masked modifications to a base image with a great deal of control and variability.

I'd recommend giving it a shot if you have an Nvidia GPU with ≥4GB VRAM.

Edit: There are also training and hypernetworks, but they require a body of source material, keywording, and significantly more time and compute resources, so I haven't attempted either.