Technopat Sosyal

@walter wick Kullandığınız model, SDXL modeli gibi duruyor. SDXL'in anime modelleri şu anlık çok kötü, Anime için SD 1.5 daha iyi.
SDXL'i eğitmesi de çok zor, yakın zamanda bu durum değişecek gibi de durmuyor :/

Ben kendi SD 1.5 modelimi kullanıyorum:

Disty0/SoteMix · Hugging Face

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

huggingface.co

SoteMix - v3.0 | Stable Diffusion Checkpoint | Civitai

SoteMix V3 is trained at 1024x1536 for high resolution image generations. Also available at Hugging Face: OpenVINO and ONNX are supported too. http...

civitai.com

SoteMix - v1.0 | Stable Diffusion Model - CHECKPOINT

2.3K runs, 28 stars, 11 downloads. Also available at Hugging Face:https://huggingface.co/Disty0/sotemixLCM Version:https://huggingface.co/Disty0/lcm_sotemixM...

tensor.art

Örnek bir resim

Kod:

Prompt: (masterpiece, best quality, highres), 1girl, solo, pov, full body,
BREAK
desert, sand, collapsed buildings, post-apocalypse,
lens flare, light scatter, depth of field, lens refraction,
BREAK
dark red hair, (long hair:0.9), blue eyes, straight hair, cat ears, blush,
(medium breasts:0.75), mature female, white sweater,

Negative: (worst quality, low quality, lowres), zombie, blurry,
interlocked fingers, loli, kid, large breasts,

Parameters: Steps: 40| Seed: 811563238| Sampler: Euler a| CFG scale: 7| Size-1: 768| Size-2: 1280| Parser: Full parser| Model: SoteMix| VAE: blessed2_vae| Backend: Diffusers| App: SD.Next| Version: 327d15a| Operations: hires; txt2img; upscale| Second pass: True| Hires force: True| Hires steps: 10| Hires upscaler: ESRGAN 4x-AnimeSharp| Hires upscale: 2| Hires resize-1: 0| Hires resize-2: 0| Hires size-1: 1536| Hires size-2: 2560| Denoising strength: 0.4| Latent sampler: Euler| Image CFG scale: 11| CFG rescale: 0.7| Hypertile UNet: 256| Pipeline: StableDiffusionPipeline| Full quality: True