RedCraft-红潮-METAFILM

基于 artificialanalysis.ai -Text to Image Arena 擂台赛精选作品集

元影智能工作室 by AiARTiST

境内已发布在 Modelscope 魔搭社区，感受极速下载！

RedCraft | 红潮 CADS Commercial & Advertising Design System · 模型库

全网唯一支持反蒸馏FLUX模型在线生成的平台（社群免费）：

AIGC 专区 – 图片生成 · 魔搭社区 Model: qijitech/RedCraft-12b-10steps-FP16-AIGC

The only platform (community-free) that supports the online generation of De-distillation FLUX models.

RedCraft | 红潮 CADS Commercial & Advertising Design System插图

Also on Huggingface.co soon

The models in this link are in a parallel relationship, not version upgrades

本链接内的模型是并列的平行关系，并非全部是版本推进

The differents can be found in the \’About this version\’ section on the right

不同版本的说明在右侧的 ‘About this version’ 清单内

以下是模型列表 list of models ：

Reveal NSFW

是FLUX.1 DEV规格的FP8 FT模型，主打男女爱情动作与人体艺术：

↳ RedCraft | 红潮 CADS – Reveal NSFW

FLUX. 1 DEV FP8 FT model, featuring romantic actions and body art

Reveal3 uncensored

是FLUX.1 DEV与反蒸馏技术结合实现的高画质版本迭代：

↳ RedCraft | 红潮 CADS – Reveal3 uncensored

Reveal‘s high-definition update by De-Re-Distillation Quality Optimality

Relustion IL NSFW

是基于SDXL规格的全量训练模型 Illustrious XL 的写实化FT版本：

↳ RedCraft | 红潮 CADS – Relustion IL NSFW

Realistic FT of Illustrious XL，full optimization based on SDXL

Relustion ULTRA

是在 Relustion IL 的基础上，进一步加强写实化的高清版本：

↳ RedCraft | 红潮 CADS – Relustion ULTRA

a high-definition version that enhances realism on Reliability IL

Relustion XL

是在 SDXL CADS3 的基础上，结合NSFW训练集制作的高清量化版本：

↳ RedCraft | 红潮 CADS – Relustion XL

It is a high-definition quantized version based CADS3 and combined

with the NSFW training set，Used for HD refine of FLUX and IL models

RASCH.1 / 2

是两个不同 Schnell 反蒸馏FT模型上，结合RED.1风格的高速模型：

↳ RedCraft | 红潮 CADS – RASCH.2

↳ RedCraft | 红潮 CADS – RASCH.1 Forge

ReFLEX NSFW

是Schnell NF4版本的二次元绘画模型，主打结构稳定与提示词还原准确：

↳ RedCraft | 红潮 CADS – REFLEX NSFW

high-speed model mixed RED.1 on different Schnell De-distillation FT models

在保证提示词准确的基础上，De-Re-Distilled(DRD) Schnell模型兼顾了速度与质量

而且蒸馏模型具有天然的肢体稳定性与画风稳定性，即便是4bit量化版本效果也很好

6~10GB显存占用，4-8步出图，速度飞快（特别适合于画风训练与建筑装修模型）

the De-Re-Distilled (DRD) Schnell model balances speed and quality

Moreover, the distillation model has natural stability , even the 4-bit quantified

6-10GB of video memory usage, 4-8 steps of image output, fast speed

(Especially suitable for artistic style creation and architectural decoration)

以下是RedCraft系列，基础美学模型Red.1的简介

RedCraft RED.1

BF16 CADS Commercial & Advertising Design System

可能是目前快速出图(10步以内)版 BF16 模型中，出图质量较好、细节较丰富的基础模型。

Fine Quality Steps 10-20 Model, In some details, it surpasses the Flux series models and approaches the 20B Parameters models.

Based on METAFILM AI – Commercial & Advertising Design System, Merge of flux-dev-de-distill, finetuned by ComfyUI, Block_Patcher_ComfyUI, ComfyUI_essentials and other tools. Recommended 10-20 steps. Greatly improved quality compared to other 12B models.

基于

De-Distill & CADS商业素材 FP16

支持在线生成 ComfyUI WebUI

10-20 STEPS Euler / DPM++2M | beta / SGM_Uniform

CFG 3-3.5

Real CFG needs to be set (ignore guidance, or set to 0)

——————————————————————————

未蒸馏PRO版本将在测试完成后公开

——————————————————————————

同样推荐选择大杨老师的 8bit finetune 版本：

Flux1-DedistilledMixTuned-V1 – v1.0 fp8 | Flux Checkpoint | Civitai

可能是目前能找到最符合官方基准风格并且出图效果最好的加速模型

RedCraft | 红潮 CADS Commercial & Advertising Design System插图1

Recommend:

UNET versions (Model only) need Text Encoders and VAE, I recommend use below CLIP and Text Encoder model, will get better prompt guidance:

Long CLIP: https://huggingface.co/zer0int/CLIP-GmP-ViT-L-14/blob/main/ViT-L-14-TEXT-detail-improved-hiT-GmP-TE-only-HF.safetensors
Text Encoders: https://huggingface.co/silveroxides/CLIP-Collection/blob/main/t5xxl_flan_latest-fp8_e4m3fn.safetensors
VAE: https://huggingface.co/black-forest-labs/FLUX.1-schnell/tree/main/vae
GGUF Version: you need install GGUF model support nodes, https://github.com/city96/ComfyUI-GGUF

Simple workflow: a very simple workflow as below, needn\’t any other comfy custom nodes(For GGUF version, please use UNET Loader(GGUF) node of city96\’s):

RedCraft | 红潮 CADS Commercial & Advertising Design System插图3

Thanks for:

https://huggingface.co/wikeeyang, Wikee Yang for carefully finetune the 8-bit model and providing the model informations，You can find it here：

wikeeyang/Flux.1-Dedistilled-Mix-Tuned-fp8 · Hugging Face

https://huggingface.co/Anibaaal, Flux-Fusion is a very good mix and tuned model.

https://huggingface.co/nyanko7, Flux-dev-de-distill is a great experimental project! thanks for the inference.py scripts.

https://huggingface.co/MonsterMMORPG, Furkan share a lot of Flux.1 model testing and tuning courses, some special test for the de-distill model.

https://github.com/cubiq/Block_Patcher_ComfyUI, cubiq\’s Flux blocks patcher sampler let me do a lot of test to know how the Flux.1 block parameter value change the image gerentrating. His ComfyUI_essentials have a FluxBlocksBuster node, let me can adjust the blocks value easy. that is a great work!

https://huggingface.co/twodgirl, Share the model quantization script and the test dataset.

https://huggingface.co/John6666, Share the model convert script and the model collections.

https://github.com/city96/ComfyUI-GGUF, Native support GGUF Quantization Model.

https://github.com/leejet/stable-diffusion.cpp, Provider pure C/C++ GGUF model convert scripts.

————————————————————————————————————————

测试问题请留言，业务合作看个人首页 +V Zyuan980

————————————————————————————————————————