I am able to get similar quality results out of some custom Stable Diffusion checkpoints like CyberRealism, but prompt engineering takes a lot more time and care.
I often use Midjourney as more of a creative tool as it typically gives high quality creative outputs from relatively basic prompts, which I then will take back to my SD models to reverse engineer and fine tune prompt to get exactly what I’m looking for.
I did, gen’d with Midjourney
Prompt: Gollum using an iPhone inside of an upscale restaurant, hyperrealistic fujifilm xf f/1.8 iso200
Midjourney has been holding on to the lead for a long time now. Congrats to the MJ devs
I am able to get similar quality results out of some custom Stable Diffusion checkpoints like CyberRealism, but prompt engineering takes a lot more time and care.
I often use Midjourney as more of a creative tool as it typically gives high quality creative outputs from relatively basic prompts, which I then will take back to my SD models to reverse engineer and fine tune prompt to get exactly what I’m looking for.