Sauna
Sauna, a Finnish tradition. I decided to make a series using sauna as the subject matter, to see how different AI engines perform.
Btw, you can click on any images and see the full-size images.
Midjourney - Text Prompt
Midjourney is always the one with the most poetic voice. Using text prompts only, but combined with NijiJourney, an anime-leaning variation of Midjourney, I was able to create men that are muscular but still have that poetic nature that’s often seen in Midjourney AI art. Always use lots of photographic keywords so that NijiJourney won’t turn them into 2D comics.
Images
Images
Images
Midjourney - Image Prompt
The sauna that the men are in are somewhat “standard” looking, so I have added an image prompt to help with the architecture. Here are the source images:
By combining those images with a low image weight --iw 0.5, I created the images below, with a strong meditation / zen influence.
Images
Images
Images
Stable Diffusion with Virile Fusion
As I have previous written, Virile Fusion is a Stable Diffusion model that you can download from CivitAI with a strong male focus, aka gay art.
Using settings recommended by the author, I created the men below. Full technical specs are after the images.
Images
Images
Images
Technical Specs
- Steps:40
- Sampler: Euler a
- CFG scale: 6
- Size: 512x512
- Model hash: a936095729
- Model: virileFusion_v10
- VAE: vae-ft-mse-840000-ema-pruned.vae
- Denoising strength: 0.5
- Clip skip: 1
- Version: 5c51633
- Parser: Full parser
- ADetailer model: face_yolov8n.pt
- ADetailer confidence: 0.3
- ADetailer dilate/erode: 32
- ADetailer mask blur: 4
- ADetailer denoising strength: 0.4
- ADetailer inpaint only masked: True
- ADetailer inpaint padding: 32
- ADetailer model 2nd: hand_yolov8n.pt
- ADetailer confidence 2nd: 0.3
- ADetailer dilate/erode 2nd: 32
- ADetailer mask blur 2nd: 4
- ADetailer denoising strength 2nd: 0.4
- ADetailer inpaint only masked 2nd: True
- ADetailer inpaint padding 2nd: 32
- ADetailer version: 23.6.2
- Hires upscale: 2
- Hires steps: 20
- Hires upscaler: R-ESRGAN 4x+
- Post Upscaler: Topaz Gigapixel HQ 2x
- Post Processing: Photoshop and Lightroom
Stable Diffusion with Virile Reality
Virile Reality is a new Stable Diffusion checkpoint model that’s created by the same author of Virile Fusion, but with a focus on creating photorealistic images, instead of the near-photorealistic illustrations that Virile Fusion does very well. For this model, Scratchproof recommends using DPM++ SDE at 22 steps, and which is what I used to create the images below.
You can download Virile Reality from CivitAI — as above, just note that these pages contain NSFW images. You have been warned!
Images
Technical Specs
- Steps:22
- Sampler: DPM++ SDE
- CFG scale: 6
- Size: 512x512
- Model hash: d9911d91a2
- Model: virileReality_v10
- VAE: vae-ft-mse-840000-ema-pruned.vae
- Denoising strength: 0.5
- Clip skip: 1
- Version: 5c51633
- Parser: Full parser
- ADetailer model: face_yolov8n.pt
- ADetailer confidence: 0.3
- ADetailer dilate/erode: 32
- ADetailer mask blur: 4
- ADetailer denoising strength: 0.4
- ADetailer inpaint only masked: True
- ADetailer inpaint padding: 32
- ADetailer model 2nd: hand_yolov8n.pt
- ADetailer confidence 2nd: 0.3
- ADetailer dilate/erode 2nd: 32
- ADetailer mask blur 2nd: 4
- ADetailer denoising strength 2nd: 0.4
- ADetailer inpaint only masked 2nd: True
- ADetailer inpaint padding 2nd: 32
- ADetailer version: 23.6.2
- Hires upscale: 1.7
- Hires steps: 10
- Hires upscaler: R-ESRGAN 4x+
- Post Upscaler: Topaz Gigapixel HQ 2x
- Post Processing: Photoshop and Lightroom