HOME | DD

Description
The final stage of the iterations.

gymdreams8.com/blog/mage-3/

I thought that I was on the right track when I used an image to influence the visuals, but felt that the influence created more of a “disc” in the background, and was not what I was going for.

So I tweaked the prompt to interpret how I’d like to utilize those shapes, in way that’s more like a hologram instead of a solid disc.

What might not be immediately apparent to people who started working with image prompts — whether it’s for Midjourney or with Control Nets in Stable Diffusion — is that the images you supply are more of an influence and not a source. It’s a bit different than img2img, where you start with an image to change it to something else. For control nets, your text prompt is still the king, and you’re simply applying the influence of external images to control how your text prompts will get realized.

Going back to the in-progress images, I felt that I was close. The particle effects added poetry. The man has the magic. I added a robe for the figures so that when combined with `breeze` and `wind`, I could break up the symmetry of the composition.

And that’s how everything eventually came together. The main reason that I switched to using sacred geometry was that I turned these mages into priest mages — ones that are no longer for battle, but functioning on a more spiritual side of magic. They are in fact walking inside a temple, so it would make more sense to use symbols that have a direct link, not to mention how much more pleasing it would be when forms are elemental themselves — yup, that’s Bauhaus applied to fantasy works!

21 images.

Tech: Text prompts with Stable Diffusion: 30 steps, DPM++ SDE Karras, CFG 5, 512x512, Airfuck’s Wild Mix 1.0, Denoising 0.5. ADetailer with prompts face_yolov8n.pt. CN1 Shuffle. CN2 Line Art Invert. Hires 1.5x (768x768), 15 steps, 4x_NMKD-Siax_200k. Post: Topaz Gigapixel HQ 4x (3172x3172), Adobe Lightroom color correction.

Description
The final stage of the iterations.

gymdreams8.com/blog/mage-3/

I thought that I was on the right track when I used an image to influence the visuals, but felt that the influence created more of a “disc” in the background, and was not what I was going for.

So I tweaked the prompt to interpret how I’d like to utilize those shapes, in way that’s more like a hologram instead of a solid disc.

What might not be immediately apparent to people who started working with image prompts — whether it’s for Midjourney or with Control Nets in Stable Diffusion — is that the images you supply are more of an influence and not a source. It’s a bit different than img2img, where you start with an image to change it to something else. For control nets, your text prompt is still the king, and you’re simply applying the influence of external images to control how your text prompts will get realized.

Going back to the in-progress images, I felt that I was close. The particle effects added poetry. The man has the magic. I added a robe for the figures so that when combined with `breeze` and `wind`, I could break up the symmetry of the composition.

And that’s how everything eventually came together. The main reason that I switched to using sacred geometry was that I turned these mages into priest mages — ones that are no longer for battle, but functioning on a more spiritual side of magic. They are in fact walking inside a temple, so it would make more sense to use symbols that have a direct link, not to mention how much more pleasing it would be when forms are elemental themselves — yup, that’s Bauhaus applied to fantasy works!

21 images.

Tech: Text prompts with Stable Diffusion: 30 steps, DPM++ SDE Karras, CFG 5, 512x512, Airfuck’s Wild Mix 1.0, Denoising 0.5. ADetailer with prompts face_yolov8n.pt. CN1 Shuffle. CN2 Line Art Invert. Hires 1.5x (768x768), 15 steps, 4x_NMKD-Siax_200k. Post: Topaz Gigapixel HQ 4x (3172x3172), Adobe Lightroom color correction.