Fire - Part 3 - Get Creative with Stable Diffusion
This is part 3 of a series. Start in part 1 if you haven’t read it.
Get Creative
The results from Part 2 were pretty good, but why stop there?! It’s time to get creative!
Once you’ve gotten a basic look and feel that you think that you can work with, then you should get creative and create the art that you want to create. Explore using different keywords in the prompt to influence the images. Add new effects to enhance it.
Model
The checkpoint model I used for these Stable Diffusion renders is Airfucks Wild Mix, which besides being a Stable Diffusion with a male focus (i.e. great for gay themed works), is an excellent model to create a unique painterly style. I talked the creator Airfucks on Discord and told him that I would play with it a bit more, and the results don’t disappoint.
Images
I won’t be sharing full prompts here because all the images are different and it’s really painful to try to grab all of them one by one. Most of the keywords are the same as those used in Part 2: Stable Diffusion, with some minor differences, which I will highlight.
Van Gogh
I learned from Unstable Diffusion that AWM is very good at paintings, so I tried some options with paintings. Because of how SD interpret tokens, sometimes it will insert famous paintings by those artists into the mix. In this case, it inserted the self portraits of Van Gogh into the image, which is really funny and that’s why the guy has blond hair.
Van Gogh in Latex
I really liked this one, and just how fluffy the “trail” is, but some of my friends have gotten into trouble showing happy trails before so I didn’t want to risk it on Instagram.
Everything else
Technical Parameters
I do tweak the control net balances parameters from image to image, but here are some starting values.
- Steps: 25
- Sampler: DPM++ SDE Karras
- CFG scale: 6
- Size: 512x512
- Model hash: 70525c199b
- Model: airfucksWildMix_v10
- VAE: vae-ft-mse-840000-ema-pruned.vae
- Denoising strength: 0.5
- Version: a844a83
- Token merging ratio: 0.5
- Token merging ratio hr: 0.5
- Parser: Full parser
- ControlNet 0:
- preprocessor: reference_only
- model: None
- weight: 1
- starting/ending: (0, 1)
- resize mode: Crop and Resize
- pixel perfect: False
- control mode: Balanced
- preprocessor params: (512m 0.5, 64)
- ControlNet 1:
- preprocessor: shuffle
- model: control_v11e_sd15_shuffle [526bfdae]
- weight: 1
- starting/ending: (0, 1)
- resize mode: Crop and Resize
- pixel perfect: False
- control mode: Balanced
- preprocessor params: (1024, 64, 64)
- 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 version: 23.7.6
- Hires upscale: 1.5
- Hires steps: 10
- Hires upscaler: 4x_NMKD-Siax_200k
- Post:
- Topaz Gigapixel HQ 4x
- Adobe Lightroom color correction