Download available at https://civitai.com/models/946205
Despite what Pixai says, it is a merge and NOT trained. The recipe is in the description below
KonpaEvo Mix is supposed to create a clean and vibrant base style. Artist and style tags should still work perfectly fine.
Use with CFG +-2.5 | range of 1.5-3.5 seems good (Comparison image CFG 1-6)
4S Lightning Lora @ 8 steps | 8S PCM Lora @ 16 steps
Positive prompt: masterpiece, best quality + artist/style tag(s)
Negative prompt I personally use:
worst quality, low quality, bad anatomy, bad proportions, extra digits, extra legs, extra arms, disfigured, twitter username, simple background, watermark, artist name, multiple views, text, blurry eyes, patreon username
Settings:
Sampler: Euler a
Steps: ~24
CFG: 1.5-3.5
I personally used a CFG of 5 so far since I just assumed the model would work best with that since its noobai based but @lizardon1024 tested with CFG 1 and got really nice results out of it. The sweet spot seems to be around 1.5-3.5.
Should you choose a speed boosting option and you are not sure whether to use 4S Lightning or CFG 1, please read my comment on CivitAI. TLDR: Use whatever you like, I think both options have their advantages.
Note about CFG 1 and samplers:
SDE samplers like DPM++ 2M SDE help quite well with the paleness. DPM++ SDE however is twice as slow as euler so I recommend against using that one. Remember to use Karras scheduler because normal doesn't seem to work
If you want to use Euler A with CFG 1: Use Luna XL VAE or Hi-Res Fix with Hi-Res Fix CFG at 4-8 can bring back vibrancy to the image, itll be slower than Hi-Res fixing at cfg 1 but the initial gen will be faster, seems worth it to me if you don't want to use SDE samplers.
