ComfyPilotComfyPilot
模型库 / LoRA / Neon Wireframe (Krea 2)
Neon Wireframe (Krea 2)

Neon Wireframe (Krea 2)

3061810

LoRA 23 次下载 更新于 2026-11-23

底模:Krea 2

synthwavestylewireframe

下载模型 在源站打开

触发词

neonwireframe

文件清单 线路:原始(huggingface.co) · 切换

下载后放进对应的 ComfyUI 目录。models/ 指 ComfyUI 的模型根目录。

权重 Neon Wireframe Krea2 V1.safetensors 224 MB 下载
权重 NOTICE.txt 0 KB 下载
权重 Neon_Wireframe_V1_training_data.zip 85 MB 下载
权重 Neon_Wireframe_V1_provenance_proofs.zip 14 KB 下载

说明

Neon Wireframe (Krea 2)

作者:ReikanYsora

Glowing 80s neon wireframes, on anything

If you like my work, feel free to send me a little Buzz tip! 💙

Neon Wireframe turns your subject into a glowing synthwave mesh — or wraps it in one. Two modes, chosen in the prompt:

  • made of wireframe: the whole subject becomes a luminous holographic wireframe model (cars, creatures, skulls, cities…);
  • wrapped in wireframe: a neon mesh hugs the surface of a real subject like an energy field, following every curve.

You choose the color in the prompt (neon cyan, hot pink, neon purple, golden yellow, ice white…) or a pair (neon pink and cyan). Works on people, animals, vehicles, mechs, objects and landscapes, in photo, 3D render, anime and painting.

How to prompt neonwireframe, neon cyan neon wireframe, made of wireframe. 3d render, a tyrannosaurus. Background: dark.neonwireframe, hot pink neon wireframe, wrapped in wireframe. night photo, a full body shot of a woman in a fitted flight suit. Background: dark.- Start with neonwireframe, <color> neon wireframe, made of wireframe. or … wrapped in wireframe., then the style and the subject.

  • End with Background: dark. for the full glow.

Recommended settings

  • Strength 0.8 – 1.0 (0.5 – 1.2 usable)
  • Steps 8 · CFG 1 · euler / simple (Krea 2 Turbo); follow your checkpoint's own settings otherwise
  • Around 1 MP · Text encoder Qwen 3VL 4B · VAE Qwen Image VAE

Training Every training image was generated with the official Krea 2 Turbo model only: no other LoRA, no third-party model, no real person, no photograph. 70 hand-checked images (80 generated, 10 rejected in two control passes: real brand designs, iconic film vehicles, see-through clothing), adults aged 25 to 45 fully clothed, plus vehicles, animals, creatures, objects and cities, 14 colors, 5 styles. Rank 16, 1500 steps.

Full provenance The training data is attached: every image, its caption and a manifest with the prompt, seed and model of each image, so anyone can regenerate it with official Krea 2 Turbo. The dataset is timestamped with OpenTimestamps .ots files).

About the showcase Some showcase images use UNLEASHED ANIME (a Krea 2 Turbo fine-tune by Silbrio) to show compatibility with fine-tuned checkpoints.

License This LoRA is a derivative of Krea 2 and is distributed under the Krea 2 Community License Agreement (copy included, see NOTICE). Commercial use is allowed only below USD 1M annual revenue. The Acceptable Use Policy applies, with content filtering and AI disclosure where required.

Additional restrictions: no depiction of minors; no images of real, identifiable persons without their consent; no illegal content.

示例图