CYBERPUNK ANIMAX

LORA
Original


Updated:

CYBERPUNK ANIMAX is a LoRA trained to inject glossy anime/cyberpunk aesthetics into your prompts — think neon-lit cityscapes, reflective fabrics, glowing circuitry woven through hair and eyes, and highly detailed anime faces with dramatic cinematic lighting. It’s optimized to produce clean linework, rich surface detail (metallics, gloss, holographic cloth), and strong lighting effects (neon rim-lights, backlight, lens-flares) while keeping anime/manga proportions and facial landmark fidelity.

Suited styles & strengths

  • Anime / Manga style: crisp cell-shading or soft painterly anime — keeps correct facial anatomy and expressive eyes.

  • Digital art / High detail: amplifies micro-detail — fabric folds, seams, glowing filaments, and reflective surfaces.

  • Cyberpunk / Futuristic: excellent at neon glows, holographic UI elements, tech backgrounds and glowing runes.

  • Character-focused composition: works best for solo characters (1girl / 1boy / solo) with medium-to-close framing but also supports full-body with detailed outfits.

  • Works well with: reflective materials, glowing hair/eyes, light particles, and technological props (e.g., headgear, energy blade).

  • Not ideal for: photoreal human realism (it’s anime-first) or highly abstract painterly styles without explicit instruction.

Recommended prompt structure (quick)

Main subject + style tags + lighting/effects + outfit/accessories + pose + background + detail boosters + LoRA name/version

Example:

(1girl, solo, long white hair with glowing cyan streaks, blue eyes, pale skin), cyberpunk anime style, glossy reflective jacket, pleated miniskirt, thigh-high boots, glowing circuitry tattoo, standing, looking at viewer, dynamic pose, neon rim light, bokeh, city at night, high detail, masterpiece, CYBERPUNK_ANIMAX:1.2

Recommended LoRA / Sampling Settings (best-effort presets)

Use these as starting points — adjust to taste. I give two presets: Detailed Portrait and Full-Body Scene.

Preset A — Detailed Portrait (close-up / head & shoulders)

Sampling Method: DPM++ 2M Karras (or Euler a if unavailable)

Sampling Steps: 28–40 (start 30)

CFG Scale: 6.5–9 (try 7.5 for natural creativity vs. strict adherence)

Clip Skip: 2 (improves stylized faces for many anime models)

Upscaler: RealESRGAN x4 (or ESRGAN_4x / Lanczos for minor upscaling)

Hires Steps: 20 (when using highres fix / hires. steps param)

Denoising Strength: 0.55–0.7 (0.6 is a good middle ground for hires fix)

Resolution Size: 1024×1536 or 1024×1024 (portrait-oriented 2:3 or square; use 1536 height if you need more vertical detail)

Preset B — Full-Body / Environment (scene-heavy)

Sampling Method: DPM++ 2M Karras

Sampling Steps: 35–50 (start 40 for complex scenes)

CFG Scale: 7–11 (higher CFG if you want strict adherence to prompt)

Clip Skip: 2

Upscaler: RealESRGAN x4 (if heavy crop/upscale required)

Hires Steps: 25–30

Denoising Strength: 0.5–0.65

Resolution Size: 1536×1024 (landscape) or 2048×1152 for ultra-detailed scenes (watch VRAM)

Practical Tips & Tweaks

LoRA weight: start ~0.6–1.0 for visible effect; reduce to 0.3–0.5 if you want subtle influence.

Combine with base model prompts: pair CYBERPUNK ANIMAX with base models that already favor anime detail (e.g., SDXL-anime variants) for best results.

Negative prompt: include common negatives to reduce artifacts: lowres, deformed, bad anatomy, extra limbs, mutated hands, text, watermark, poorly drawn

Clip Skip testing: try clip-skip 2 vs 1 — many anime LoRAs perform better with 2.

Lighting control: if you want stronger neon/glow, add repeated tags like (neon, neon lighting, glowing, light particles, rim light) — repetition often emphasizes effect.

Outfit detail: use material tags: reflective material, shiny clothes, metallic plating, holographic to get that polished cyber look.

Hires-fix workflow: generate at half/res then use hires/denoise at ~0.6–0.7 to retain stylized linework and gloss.

Avoid overfitting: if you fine-tune LoRA further, keep a diverse dataset across lighting, poses, outfits to prevent the LoRA from forcing a single pose/style.

Example Prompts

Portrait — intense neon:

masterpiece, best quality, (1girl), close-up, long aqua hair with glowing highlights, glowing cyan eyes, pale skin, anime style, glossy reflective jacket, neon rim light, light particles, bokeh, cyberpunk city background, detailed hair strands, intricate circuitry tattoos, CYBERPUNK_ANIMAX:0.9

Full body — action pose:

(1girl, solo), dynamic pose, wielding energy blade, black bodysuit with glowing runes, pleated miniskirt, thigh-high boots, leg straps, crown headgear, glowing hand, neon trails, cyberpunk background with holographic billboards, masterpiece, high detail, CYBERPUNK_ANIMAX:0.85

Version Detail

Trained by Tensor
FLUX.1
1500
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