Dance / Run - Photo
Although I posted the photorealistic paintings in first Dance / Run set, you can use the same technique to produce images in any style. For example, these are in fact renders that I made originally, before deciding to go with a more painterly style. The images in this set all used the same rendering technique, which to use these components:
- text prompt
- Control Net: Reference Only
- Control Net: Shuffle
The image I used for the two control nets were generated with Midjourney v5.2, but you can use any source images you like — including images that you already have, or made one from scratch with Stable Diffusion. I used Midjourney because it is the quickest way to create quick stylistic iterations.
To show that these are not cherry-picked results, I have included a grid showing 16 renders from a batch as the last image in the Instagram set, or see the following image:
Tech: Text prompts with Midjourney (for Control Nets) + Text prompts with Stable Diffusion: 50 steps, DPM++ SDE (not Karras), CFG 5, 512x512, Virile Reality v2, Denoising 0.5. CN0 Reference Only. CN1 Shuffle. ADetailer mediapipe_face_full. Hires 2x (1024x1024), 30 steps, 4x_NMKD-Siax_200k. Post: Topaz Gigapixel HQ 4x (4096x4096), Adobe Lightroom color correction.
Images
Note: some of these images came from a prompt with keywords like motion blur, double exposure, etc so that’s why some of them would have floating limbs — they’re intentional, though obviously the results are not perfect.
