RedCraft | 红潮 | ERNIE RedMIX

Finished uploading REDZimage15_ComfyUI_workflows_example.zip19.11 GB - AIO checkpoint 11.46 GB - none AIO bf16(single transformers) 5.73 GB - none AIO fp8 (single transformers) 670.08 MB - Lora<ul><li>You can download the LoRA version later on this page</li></ul>（⬆️Pruned Model fp16 (670.08 MB) is LoRA Safetensors⬆️）<ul><li>Single file transformers have been uploaded BF16/FP8 e4m3fn</li></ul>---RedZimage⚡️1.5 版本介绍Distilled from Nano Banana Pro big-data sampling from real user photos(thanks to 瓜哥@guahunyo),based on the open-source research by 小智@xiaozhijason, synthesized into the dataset ZImageTurboGen-3k.---The dataset is specifically designed for the following research scenarios:Data distribution shift analysis 数据分布偏移分析Reversibility / de-distillation research of distilled models蒸馏模型的可逆性/去蒸馏（de-distillation）研究Model diversity generation research 模型多样性生成(HF): <a target="_blank" rel="ugc" href="https://huggingface.co/datasets/lrzjason/ZImageTurboGen">https://huggingface.co/datasets/lrzjason/ZImageTurboGen</a>---Improved the style of the ZiTurbo version that was heavily tuned on community AI-generated web images.The REDZ training set comes from the ultimate essence of the community— 20+ year veterans' lightning-fast meme-battle handspeed.(Thanks to 佬杨@wikeeyang, @亮亮rayne, 十字鱼@Gluttony100,可乐@Colour, and everyone else)---RedZimage⚡️1.5 蒸馏了 nano banana pro 来自真实用户照片的大数据采样（感谢 @guahunyo ），基于@xiaozhijason开源的研究向合成数据集 ZImageTurboGen-3k数据集专为以下研究场景设计：1. 数据分布偏移分析⚡️2. 蒸馏模型的可逆性/去蒸馏（de-distillation）研究⚡️3. 模型多样性生成研究⚡️（HF）：https://huggingface.co/datasets/lrzjason/ZImageTurboGen⚡️改善了ZiTurbo版本来自社区AI合成网图训练集调教的风格，REDZ 训练集来自社区精华++20年佬同志斗图手速（感谢@wikeeyang @亮亮rayne @Gluttony100 @Colour 及所有人）---etc

redcraftERNIERedmix_redzit15AIO.safetensors

redcraftERNIERedmix_redzit15AIO_trainingData.zip

redcraftERNIERedmix_redzit15AIO.gguf

RedCraft | 红潮造相⚡️REDZimage

RedZiT⚡️1.5AIO

ERNIE♥RedMIX

ZIB Distilled⚡️FUNxAGILE

ZIB Distilled⚡️DPO Veris

ZIB Distilled⚡️DX3 Lucis

ZIB Distilled⚡️RedDX2

REDZIB⚡️DX1加速版

RED-红桃K🧡[King of Hearts]

RED-OMNI [Flux.1&DreamO]

RED-Edit (ICEdit)1.1🔥多任务

赩梦|REDiDream(NSFW i1)

RED-UNO（In-Context）

Reveal5[SFW]ULTRA

RealReveal5

Illust3Relustion🔥

PONY3Relustion PRO🔥

RED[创意] Epicus BIG Movie

F.1[修图] Fill.NSFW

🌹NewReveal[F.1]ULTRA🌹

RASCH.3 (RUSH·Reveal)🔥

New REVE[A]L 🏛🔞🔥

Relustion 1.5 [FP32]

光輝2Relustion[IL]

PONY2Relustion(FP32)🔥

RED.2 BF16 (ArtAug)

TURBO Reveal(HOTFix2.1)🔥

PONY Relustion🔥

Reveal ULTRA [v3.5]🔥

Relustion ULTRA🔥

Relustion XL [FP8]

REFLEX NSFW

RASCH.2

Relustion IL NSFW

Relustion-IL-Forge

RASCH.1 Forge

Reveal NSFW

Red.1 Q4 GGUF

Red.1 BF16

RedCraft-造相·红-RED⚡️Zimage做好工具人 服务艺术家Forever in memory of METAFILM Studio founder <a target="_blank" rel="ugc" href="https://zhuanlan.zhihu.com/p/16987553665">Mr. Yuan Bo</a>RED-Zimage 红潮造相1.5AIO 12/03/2025The currently uploaded version is A-I-O bf16 Checkpoint. Please place it in the Checkpoints directory of your inference environment当前已上传版本为 A-I-O bf16 Checkpoint，请放入推理环境的checkpoints目录The environment for generating the example image is <a target="_blank" rel="ugc" href="https://www.comfy.org/zh-cn/">ComfyUI</a> 0.3.73. Download this link for the complete example image, which includes the example workflow例图生成环境为<a target="_blank" rel="ugc" href="https://www.comfy.org/zh-cn/">ComfyUI</a>0.3.73，下载本链接完整例图中包含示例工作流REDZimage1.5_extracted_lora Rank128 正在导出中，稍后可在本页面下载LoRA版本（不另开链接）REDZimage 1.5-lora Rank128 is currently being exported, You can download the LoRA version later on this pageRedZimage 1.5AIO compare with RedZimage 1.5LoRA:⬅️ On the left is RedZimage 1.5AIO / RedZimage 1.5LoRA on the right ➡️⚡️RedZimage 版本介绍 1.5Version Introduction⚡️⚡️Distilled from Nano Banana Pro big-data sampling from real user photos (thanks to 瓜哥@guahunyo),⚡️based on the open-source research by 小智@xiaozhijason,synthesized into the dataset ZImageTurboGen-3k.⚡️Improved the style of the ZiTurbo version that was heavily tuned on community AI-generated web images.The dataset is specifically designed for the following research scenarios:Data distribution shift analysis 数据分布偏移分析Reversibility / de-distillation research of distilled models蒸馏模型的可逆性/去蒸馏（de-distillation）研究Model diversity generation research 模型多样性生成(HF): <a target="_blank" rel="noopener noreferrer nofollow" href="https://huggingface.co/datasets/lrzjason/ZImageTurboGen">https://huggingface.co/datasets/lrzjason/ZImageTurboGen</a>The REDZ training set comes from the ultimate essence of the community⚡️20+ year veterans' lightning-fast meme-battle handspeed.(Thanks to 佬杨⚡️@wikeeyang, 亮总⚡️@亮亮rayne, 十字鱼⚡️@Gluttony100,可乐⚡️@Colour, and everyone else⚡️⚡️⚡️⚡️⚡️⚡️)⚡️蒸馏了 nano banana pro 来自真实用户照片的大数据采样（感谢 @guahunyo ），⚡️基于@xiaozhijason开源的研究向合成数据集 ZImageTurboGen-3k数据集专为以下研究场景设计：1. 数据分布偏移分析2. 蒸馏模型的可逆性/去蒸馏（de-distillation）研究3. 模型多样性生成研究（HF）：<a target="_blank" rel="noopener noreferrer nofollow" href="https://huggingface.co/datasets/lrzjason/ZImageTurboGen">https://huggingface.co/datasets/lrzjason/ZImageTurboGen</a>⚡️改善了ZiTurbo版本来自社区AI合成网图训练集调教的风格， ⚡️REDZ 训练集来自社区精华++20年佬同志斗图手速（感谢@wikeeyang @亮亮rayne @Gluttony100 @Colour 及所有人）etc实测样张对比（nano2/Zi-Turbo/REDZ1.5）Comparison samples实测样张对比 2（Dreamina/Zi-Turbo/REDZ1.5）Comparison samples 2例图参数及对比工作流稍后上传到返图区 The example parameters and comparison workflow will be uploaded to the return to image area later

Files with SHA256: f0a36d3c82aacd3b5116ace3f01ca7adbe26232a912ffc7af5c4af40dac5a3d9