🚀 Sodex (v1.0) - High-Performance Fine-Tuned Model

Sodex Banner

Sodex, Stable Diffusion XL 1.0 (SDXL) tabanı üzerine inşa edilmiş, 1024x1024 yüksek çözünürlüklü görseller üretmek için seçkin görsellerle A100 GPU üzerinde özel olarak eğitilmiş (fine-tuned) çok yönlü bir modeldir.

⚡ Kullanım

Bu model diffusers kütüphanesi ile tam uyumludur. Gerekli kütüphaneleri yükleyin:

pip install torch diffusers transformers accelerate

1️⃣ Text-to-Image (Metinden Görsele)

import torch
from diffusers import AutoPipelineForText2Image

pipe = AutoPipelineForText2Image.from_pretrained(
    "fasdgdsagasdg/sodex",
    torch_dtype=torch.float16,
    use_safetensors=True
).to("cuda")

prompt = "A hyper-realistic cinematic portrait of a cyberpunk samurai, neon lights, 8k resolution"
negative_prompt = "blurry, low quality, deformed"

image = pipe(prompt=prompt, negative_prompt=negative_prompt, num_inference_steps=30, guidance_scale=7.5).images[0]
image.save("sodex_t2i.png")

2️⃣ Image-to-Image (Görselden Görsele)

import torch
from diffusers import AutoPipelineForImage2Image
from PIL import Image

init_image = Image.open("foto.jpg").convert("RGB").resize((1024, 1024))

pipe = AutoPipelineForImage2Image.from_pretrained(
    "fasdgdsagasdg/sodex",
    torch_dtype=torch.float16,
    use_safetensors=True
).to("cuda")

prompt = "Turn this person into an anime character, studio ghibli style, 8k"
image = pipe(prompt=prompt, image=init_image, strength=0.60, guidance_scale=7.5, num_inference_steps=30).images[0]
image.save("sodex_i2i.png")

⚙️ Tavsiye Edilen Ayarlar

  • Çözünürlük: 1024x1024
  • Adım Sayısı (Steps): 25 - 40
  • Guidance Scale (CFG): 7.0 - 8.5
  • Image-to-Image Strength: 0.55 - 0.65
Downloads last month
17
Safetensors
Model size
3B params
Tensor type
F16
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for fasdgdsagasdg/sodex

Finetuned
(1225)
this model