Introduction for NEO GENESIS

NEO GENESIS is envisioned as a high-quality, high-resolution anime-inspired character LoRA designed to create elegant, realistic yet stylized school-themed portraits. It focuses on a single young woman with a curvaceous figure, bob haircut, and brown eyes/hair, dressed in a dark blue school uniform. The background — chalkboard, desks, and bookshelves — evokes a soft, nostalgic classroom vibe. This makes NEOGENESIS ideal for anime-style official art, visual novel cutscenes, or illustrations with a realistic yet vibrant aesthetic.

The LoRA’s goal:

Deliver masterpiece-level detail (4K+ output).

Keep natural anatomy and soft lighting while maintaining anime crispness.

Create full-body renders with sharp focus and immersive backgrounds.

Recommended Style

Because your tags blend “official art,” “masterpiece,” “sharp focus,” and “soft lighting,” the most suitable styles would be:

Anime Realism / Semi-Realism (like Kyoto Animation or Makoto Shinkai style)

Illustration for light novels or game CGs

Modern high-quality anime art (Cygames or Azur Lane official art vibe)

This style uses clean line art, soft gradient shading, and lush, detailed environments without losing the character focus.

Suggested Best Settings

Sampling Method ➡️ DPM++ 2M Karras or DPM++ SDE Karras

These produce smooth, consistent high-detail images.

Sampling Steps ➡️ 28–32 steps

Enough for detail but still efficient.

CFG Scale ➡️ 6.5–7.5

Balanced prompt adherence without overbaking details.

Clip Skip ➡️ 2 (if using anime model like Anything v5 / Counterfeit v3)

Keeps faces and colors closer to anime style.

Upscaler ➡️ 4x-UltraSharp or R-ESRGAN 4x+ Anime6B

To upscale while retaining lineart and details.

Hires Steps ➡️ 12–15

Good for refining textures at high res.

Denoising Strength (Hires Fix) ➡️ 0.35–0.45

Preserves original composition but sharpens edges.

Resolution Size ➡️ 512×768 for base, upscale to 2048×3072 (portrait)

Perfect for full-body shots with rich backgrounds.

Additional Tips

For very official-art results, combine the LoRA with a base anime model such as Anything v5, Counterfeit, or AOM3.

Use a slightly lower weight (0.6–0.8) on the LoRA if backgrounds get overtrained; raise to 1.0–1.2 if the uniform details need boosting.

Add (official art, high quality) and (detailed background) at the start of the prompt to push style.

Version Detail

Trained by Tensor
FLUX.1
1200
1

Project Permissions

Reprinting is strictly prohibited

    Use Permissions

  • Use in TENSOR Online

  • As a online training base model on TENSOR

  • Use without crediting me

  • Share merges of this model

  • Use different permissions on merges

    Commercial Use

  • Sell generated contents

  • Use on generation services

  • Sell this model or merges

Related Posts