Disty0/SoteMixV2.1

SoteMix V2.2 is available here:
https://huggingface.co/Disty0/SoteMixV2.2

SoteMix V2 is trained at 1024x1536 for high resolution image generations.

Positive Prompts:

masterpiece, best quality, highres, 1girl,

Negative Prompts:

worst quality, low quality, lowres, zombie, interlocked fingers,

Do not use any negative embeddings.

Sampler: Euler a
Steps: 30-40
Clip Skip: 1
CFG: 4-7
Base Resolution: 512x / 768x / 1024x / 768x1280 / 960x1280 / 1024x1536 / 1920x1080
512x can be blurrier than SoteMix V1.
Model can still be chaotic at 1024x1536 and 1920x1080.

Second Pass / Hires:
Sampler: Euler / Euler a
Steps: 10 with Euler / 20 with Euler a
Upscaler: RealESRGAN 4x+ Anime6B / ESRGAN 4x-AnimeSharp with 0.2-0.3 denoise strength.
CFG: 6-9
Resolution: 2x of the base resolution.

Training:
My GPU couldn't handle full model training at these resolutions so i trained it as a 512 layer Lora with SoteMix V1 as the base.
Used highres as the trigger word. Also used raifu trigger word with my OC character.

Resolution: 1024x1536 with Bucketing
Batch Size: 1
Steps: 12000
GPU: Intel ARC A770 16GB

Bucket:

bucket 0: resolution (832, 1664), count: 49
bucket 1: resolution (896, 1536), count: 1
bucket 2: resolution (960, 1536), count: 12
bucket 3: resolution (960, 1600), count: 492
bucket 4: resolution (1024, 1344), count: 112
bucket 5: resolution (1024, 1408), count: 31
bucket 6: resolution (1024, 1536), count: 768
bucket 7: resolution (1152, 1344), count: 6
bucket 8: resolution (1216, 1216), count: 25
bucket 9: resolution (1344, 1024), count: 16
bucket 10: resolution (1344, 1152), count: 2
bucket 11: resolution (1536, 896), count: 59
bucket 12: resolution (1536, 1024), count: 7
bucket 13: resolution (1600, 960), count: 82
bucket 14: resolution (1664, 832), count: 21
bucket 15: resolution (2304, 576), count: 1

Merge:
Merged SoteMix V1 with Lunar Radiance Light and then merged the Hires Lora i trained on top of it.
Merge ratio: (0.6 SoteMix V1 + 0.4 Lunar Radiance Light) + 0.7 Hires Lora