Gym Selfie, Flex Friday
Selfie Series
On Instagram, Kevin Winston (IG: @kevinwinston) tagged me and asked that I participate in “Flex Friday” by posting a picture.
I know Kevin from Yale, and he knows that I only make AI works from my GymDreams8 account, so I was surprised by the tag. He clarified that he wanted me to share the men I create with AI — so of course, I agreed.
Interestingly, Kevin seemed to assume that I had a collection of gym selfies stashed in my archive. Nope — but I made these specifically for you!
To accomplish this, I began by taking Kevin’s photo from his Instagram story and utilized the generative fill feature in Photoshop to create an image measuring 1024x1024 pixels. This image served as the basis for driving an OpenPose Control Net, allowing me to set the pose for all the figures in the composition.
Next, I proceeded to render a series of images using two different checkpoint models: Virile Reality v3 Beta 2, known for its photographic style, and Airfuck’s Brute Mix, which produces a photorealistic style. By using the OpenPose control net, I ensured that each render maintained the exact same pose as the original photo.
By employing Dynamic Prompt, I was able to batch these renders while incorporating variations in the color of the figures’ shorts and sneakers. Additionally, I applied some limb fixes using Photoshop AI to correct issues typically seen in AI generative works.
Images
Virile Reality
Images
Airfuck’s Brute Mix
Images
Tech
Text prompts in Stable Diffusion txt2img, with OpenPose control net and dynamic prompts.
- Models:
- Virile Reality v3 Beta 2, 30 steps, Hires 15 steps
- Airfuck’s Brute Mix 1, 20 steps, Hires 10 steps
- Sampler: Euler a
- CFG scale: 5
- Size: 512x512
- VAE hash: 235745af8d
- VAE: vae-ft-mse-840000-ema-pruned.vae.ckpt
- Denoising strength: 0.4
- ADetailer model: face_yolov8n.pt
- ADetailer ControlNet model: control_v11p_sd15_openpose (cab727d4)
- ADetailer version: 23.9.1
- ControlNet 0:
- Module: dw_openpose_full
- Model: control_v11p_sd15_openpose (cab727d4)
- Weight: 1
- Processor Res: 1024
- Guidance Start: 0
- Guidance End: 1
- Pixel Perfect: False
- Control Mode: Balanced
- Hires upscale: 2, 1024x1024
- Hires upscaler: 8x_NMKD-Superscale_150000_G
- Automatic1111 Version: v1.6.0
- Post upscale: Topaz Gigapixel AI HQ 4x, 4096x4096
- Post Process: Adobe Lightroom
Full Tech
- Steps: 20
- Sampler: Euler a
- CFG scale: 5
- Seed: 3768898257
- Face restoration: CodeFormer
- Size: 512x512
- Model hash: 8ff8347bb0
- Model: airfucksBruteMix_v10
- VAE hash: 235745af8d
- VAE: vae-ft-mse-840000-ema-pruned.vae.ckpt
- Denoising strength: 0.4
- Style Selector Enabled: True
- Style Selector Randomize: False
- Style Selector Style: base
- ADetailer model: face_yolov8n.pt
- ADetailer prompt: handsome {french|italian|dutch|german|british} face
- ADetailer confidence: 0.3
- ADetailer dilate/erode: 4
- ADetailer mask blur: 4
- ADetailer denoising strength: 0.4
- ADetailer inpaint only masked: True
- ADetailer inpaint padding: 32
- ADetailer ControlNet model: control_v11p_sd15_openpose (cab727d4)
- ADetailer version: 23.9.1
- ControlNet 0:
- Module: dw_openpose_full
- Model: control_v11p_sd15_openpose (cab727d4)
- Weight: 1
- Resize Mode: Crop and Resize
- Low Vram: False
- Processor Res: 1024
- Guidance Start: 0
- Guidance End: 1
- Pixel Perfect: False
- Control Mode: Balanced
- Hires upscale: 2
- Hires steps: 10
- Hires upscaler: 8x_NMKD-Superscale_150000_G
- Lora hashes:
- bulgerk-dickprint: b1934ecdac9a
- Automatic1111 Version: v1.6.0