Ahli PixAI
PixAI
PixAI
Generate
Sign in
Laman Utama
Jana
Model
Studio
NEW
Kontes
Cipta dengan Mio.2
Kotak Alat
NEW
Guide

SD

🚧 DarkAngel-Mix [2D]

Cipta

  • Penjana Imej AI
  • Penjana Animasi AI
  • Kotak Alat
  • Penjana Tematik
  • Latih LoRA
  • Ejen Mio.2
  • Studio

Temui

  • Tag Popular
  • Ranking
  • Pasaran Model
  • Kontes
  • Berita

Tentang

  • Guide
  • Cara Menggunakan PixAI
  • Tsubaki.2
  • Kenali Mio
  • Peraturan Kandungan

Harga & Bantuan

  • Keanggotaan
  • Pek Kredit
  • Kontak

Aplikasi Mudah Alih

  • Dapatkan aplikasi PixAI
  • App Store
  • Google Play
© 2026 PixAI
  • Syarat Perkhidmatan
  • Dasar Privasi
  • Dasar Hak Cipta
  • Attributions
  • Undang-undang Transaksi Komersial Tertentu
🚧 DarkAngel-Mix [2D] - AI Model cover image
Dimuat naik pada
9 Jul 2024, 5:48 PTG
Kegunaan
3.9k
Ulasan
Cemerlang  (1)
Langkah Pensampelan

26

Kaedah Pensampelan

Euler a

Skala CFG

6

Negatif

(bad-artist-anime, by bad-artist:1.4), (worst quality, low quality:2), (bad quality:1.9), normal quality, (EasyNegativeV2, easynegative:1), (bad eye, unclear eyes, undetailed eyes, missing limbs, fused limbs, broken limbs, fewer digits, extra digits, mutation, deformity:1.4), (interlocked fingers:1.3), (bad anatomy:1.1), monochrome, watermark, ng_deepnegative_v1_75t, verybadimagenegative_v1.3, (white background, simple background:1.3), (NSFW, cleavage:1.3)

Kebenaran
  • Allow image generation & sharing
  • Benarkan pengguna memuat turun model anda
  • Kegunaan komersial

Keterangan

Recommended: 20 steps for closeups, +26 steps for medium shot, +30 steps for full body. Face fix for 26+ steps. 1024x res with Euler a sampler. If you don't do it that way: With a lower resolution use highRes fix or enhance for the best quality possible.

Remove N-SFW negative prompts if you want to make N-SFW or background negatives if you want another simpler backgrounds.

My intention with this model was to create a fast model that worked great on higher resolutions (you still can use lower ones but it will be better with higher). V2 has a lot of improvements like better eye/face, hands, backgrounds... It will be a very experimental model. It seems to work very good with 1024x resolution and low steps. I made it merging some models I like. You can use lower steps if you increase resolution. Or you can also enhance the image. Not tested enough. v0 may have more issues but from v1 it should be fine enough with a good face, N-SFW and decent hands. v2 should have better backgrounds and eye/face, maybe hands too. This is just an experiment I'm making while I'm trying to improve my Everything-Mix model.

Issues: Sometimes bad anatomy. Increasing steps may fix it.

Komen