A family visits the Egypto-Dutch Empire Museum
From "The Illustrated Children's Book of the Egypto-Dutch Empire." Available in museum gift shops and better bookstores everywhere.
Stable Diffusion | Photoshop
=====
I think there is a lot of misunderstanding about how AI fits into an artist's workflow. This is too little space here to do into detail, but in summary.
Digital camera image => Stable Diffusion + text that describes or misdescribes the image with references to key historic master painters (which can shift as the process continues) => drilling down through widely diverse variations until finding an interesting one with good composition, a credible distribution of lights and darks, and pleasing color => generating a series of very close iterations and selecting the best (4 to 6 images) => Upscaling those iterations => in Photoshop compositing the best elements of the close iterations into a single image => color balancing, contrast adjustments, overlaying textures, cropping, sharpening and otherwise fixing flaws (which is a significant task). Some areas in complex compositions with multiple people I will zoom in on hands and especially heads, copy out those areas, enlarge them and re-render them in Stable Diffusion to improve detail then composite them back in. I will often repaint areas by hand to make corrections.
The process is similar to that of a Photoshop artist who creates composites. Same set of Photoshop skills, but the source material is generated differently.
A family visits the Egypto-Dutch Empire Museum
From "The Illustrated Children's Book of the Egypto-Dutch Empire." Available in museum gift shops and better bookstores everywhere.
Stable Diffusion | Photoshop
=====
I think there is a lot of misunderstanding about how AI fits into an artist's workflow. This is too little space here to do into detail, but in summary.
Digital camera image => Stable Diffusion + text that describes or misdescribes the image with references to key historic master painters (which can shift as the process continues) => drilling down through widely diverse variations until finding an interesting one with good composition, a credible distribution of lights and darks, and pleasing color => generating a series of very close iterations and selecting the best (4 to 6 images) => Upscaling those iterations => in Photoshop compositing the best elements of the close iterations into a single image => color balancing, contrast adjustments, overlaying textures, cropping, sharpening and otherwise fixing flaws (which is a significant task). Some areas in complex compositions with multiple people I will zoom in on hands and especially heads, copy out those areas, enlarge them and re-render them in Stable Diffusion to improve detail then composite them back in. I will often repaint areas by hand to make corrections.
The process is similar to that of a Photoshop artist who creates composites. Same set of Photoshop skills, but the source material is generated differently.