In order to be (somewhat) fair and to avoid cherry-picking, I have decided to have a solid comparison between models. On the old wiki, I cherry-picked generations.
I am aware that models have different prompting styles or recommended prompts or settings.
Common settings
- Negative Prompt:
(worst quality, low quality:1.4), nsfw - Sampler:
Euler - Size:
512x768 - Steps:
20 - CFG scale:
7 - Hires: none
- Clip skip:
2 - ENSD:
0(default) - Seed:
3252012142("cytlnuroisea" to numbers and I only picked the first 10 numbers) - Batch Size:
3
Prompts for Fiyu
1girl, blonde hair, blue eyes, short hair, wavy hair, hair ribbon, blue ribbon, detached sleeves, blue dress, portrait Negative prompt: (worst quality, low quality:1.4), nsfw Steps: 20, Sampler: Euler, CFG scale: 7, Seed: 3252012142, Size: 512x768, Clip skip: 2, ENSD: 0, Batch size: 3
Prompts for Aoi
1girl, white hair, green eyes, long hair, messy hair, choker, white jacket, open jacket, green shirt Negative prompt: (worst quality, low quality:1.4), nsfw Steps: 20, Sampler: Euler, CFG scale: 7, Seed: 3252012142, Size: 512x768, Clip skip: 2, ENSD: 0, Batch size: 3
Prompts for landscapes
Using a different resolution:768x512
scenery, mountains, grass fields, flowers Negative prompt: (worst quality, low quality:1.4), nsfw Steps: 20, Sampler: Euler, CFG scale: 7, Seed: 3252012142, Size: 768x512, Clip skip: 2, ENSD: 0, Batch size: 3