Thanks for the break down. Super helpful. I did something like this on the second part of my Doge story where I imported a midjourney image into Procreate and added lines and shading to really make it more cartoon looking. I think using AI art as a starting point is a great use case for a lot of artistic reasons.
Ok. Have you considered Stable Diffusion instead of Midjourney?
Stable Diffusion is free. Eventually Midjourney charges you.
Stable Diffusion can be installed and run locally - now, you do need a beefy machine for this. Like an Nvidia GPU with 6GB or more VRAM, but my current laptop has a 12GB Nvidia 3080ti (mobile), so I should be OK (as I type I'm waiting for my machine to download the libraries. You also need about 100GB of storage space... Well, my laptop as 3TB of SSD on two drives. I'm also installing the GUI version to try... I might end up using the CLI version.
Now, even if you're staying with Midjourney, here's the real point of this comment. Image to Image. Instead of a text prompt, you start with Simon-scribbles and let Stable Diffusion glam it out for you. I think this will give you more consistent results if you start doing character art. And, as shown in that Corridor video I sent you, if you have an insanely good GPU (you don't... What, was it 64 GB of VRAM to train?), you can train Stable Diffusion. Maybe you create Clarke in Makehuman, take him into Blender, animate a turnaround with some moving lights and render a PNG sequence to train? Boom, Stable Diffusion knows Clarke. With Blender and Mixamo and Hitfilm you can import a rigged and animated Clarke and export poses to feed into your starting image. Use GIMP to quickly comp stock images from Pexels or other sites.
Thanks for the break down. Super helpful. I did something like this on the second part of my Doge story where I imported a midjourney image into Procreate and added lines and shading to really make it more cartoon looking. I think using AI art as a starting point is a great use case for a lot of artistic reasons.
Ok. Have you considered Stable Diffusion instead of Midjourney?
Stable Diffusion is free. Eventually Midjourney charges you.
Stable Diffusion can be installed and run locally - now, you do need a beefy machine for this. Like an Nvidia GPU with 6GB or more VRAM, but my current laptop has a 12GB Nvidia 3080ti (mobile), so I should be OK (as I type I'm waiting for my machine to download the libraries. You also need about 100GB of storage space... Well, my laptop as 3TB of SSD on two drives. I'm also installing the GUI version to try... I might end up using the CLI version.
Now, even if you're staying with Midjourney, here's the real point of this comment. Image to Image. Instead of a text prompt, you start with Simon-scribbles and let Stable Diffusion glam it out for you. I think this will give you more consistent results if you start doing character art. And, as shown in that Corridor video I sent you, if you have an insanely good GPU (you don't... What, was it 64 GB of VRAM to train?), you can train Stable Diffusion. Maybe you create Clarke in Makehuman, take him into Blender, animate a turnaround with some moving lights and render a PNG sequence to train? Boom, Stable Diffusion knows Clarke. With Blender and Mixamo and Hitfilm you can import a rigged and animated Clarke and export poses to feed into your starting image. Use GIMP to quickly comp stock images from Pexels or other sites.
Just a thought.