CivArchive
    Preview 112379617
    Preview 112379800
    Preview 112379823
    Preview 112379845
    Preview 112379897
    Preview 112379959
    Preview 112380059
    Preview 112380086
    Preview 112380124
    Preview 112380134
    Preview 112380234
    Preview 112380248
    Preview 112380332
    Preview 112380333
    Preview 112380477
    Preview 112380315
    Preview 112380314
    Preview 112380095
    Preview 112379678
    Preview 115096128

    做好工具人 服务艺术家 Forever in memory of METAFILM Studio founder Mr. Yuan Bo 

    Metafilm`s CATEGORICALINDEX 》

    Grouped by base model, bases ordered by first production use (newest first); brands inside each base ordered by heat (total downloads); works inside each brand newest first. 282 works across 21 base models.

    Type tags: fine-tune / LoRA / workflow / ControlNet. Data collected 2026-10-23.

    ▍RedCraft 红潮 ⌝ ▍Dark Beast 黑兽 ⌝ ▍FineView 欢视 ⌝ 
    ▍红潮|RED 室内设计 Interior Artistry ⌝   ▍AiARTiST 工业设计 Industrial ⌝  

    ▍CADS Workflows AIGC工业设计流水线
    ▍AiARTiST Architecture ▍AiARTiST Chinese Art
    ▍Character LoRAs  ▍UNIT Base Reworks 会员模型
    ▍MIRACLE Imaging ▍Tooling · Acceleration · Workflows


    KREA 2 / KREA2 Turbo Fine-tune / LoRas

    First used 2026-06-24 · 4 brands · 15 works · 187,368 downloads

    ▍FineView 欢视

    1 works · 5 downloads

    ▍ 欢视 · KREA 2 / KREA2 Turbo · KREA2 IC-Turbo 1.0 ⌝

    Open → Civitai · KREA2 IC-Turbo 1.0

    Dark Beast 5C K2T × Krea 2 Identity Edit 融合版;EULER / 8 步 / CFG 1 / 尺寸 ≤2MP,带 In-Context 身份锁定。
    Fusion of Dark Beast 5C K2T with Krea 2 Identity Edit; Euler, 8 steps, CFG 1, ≤2MP, with in-context identity locking.

    2026-10-23 · fine-tune


    ▍RedCraft 红潮

    9 works · 130,143 downloads

    ▍ 红潮 · Krea 2 Turbo · KREA2RED 剧创 Story 2 ⌝

    Open → Civitai · KREA2RED 剧创Story2

    红潮剧创Story2美颜版,Krea 2 Turbo 上做剧集人物。
    BeautyLock Drama Story2 cut of REDGPT on Krea 2 Turbo, aimed at consistent dramatic characters and scenes.

    2026-10-09 · fine-tune · 882 downloads · 110 likes · 178 images · 1,451 image likes


    ▍ 红潮 · Krea 2 Turbo · 赤佬 5.0 BUZZ DEAL (K2T)商务版 ⌝

    Open → Civitai · 赤佬5 BUZZ DEAL (K2T)商务版

    Krea 2 Turbo 单模型Versions,面向商业文生图素材生产。
    Krea 2 Turbo single-model build for commercial text-to-image asset production.

    2026-09-16 · fine-tune · 347 images · 3,501 image likes


    ▍ 红潮 · Krea 2 Turbo · 赤佬 4.0 MASO (K2T) DUAL ⌝

    Open → Civitai · 赤佬 4.0 受虐版 K2T DUAL

    Krea 2 Turbo 双模型降噪版,用随机性换听话,主攻漫剧素材。
    Krea 2 Turbo DUAL hybrid trading prompt obedience for high-noise randomness, aimed at manga-style drama assets.

    2026-09-09 · fine-tune · 12,345 downloads · 376 likes · 50 images · 1,386 image likes


    ▍ 红潮 · Krea 2 Turbo · 赤佬 3.0 (Krea2) ⌝

    Open → Civitai · 赤佬 3.0 (Krea2)

    赤佬无码版 Krea 2 混合模型,CFG 1、8-12 步,画面效果调优。
    Uncensored Krea 2 mix from the 赤佬 line, CFG 1 in 8-12 steps with a visual-effects pass.

    2026-07-17 · fine-tune · 53,368 downloads · 1,598 likes · 104 images · 2,457 image likes


    ▍ 红潮 · Krea 2 Turbo · KREA2RED AE剧创 Story 1

    Open → Civitai · KREA2RED AE剧创

    Krea 2 Turbo 双权重版,高低噪交叉采样,多样性留给剧创画面。
    Two-file high/low-noise Krea 2 Turbo fine-tune that cross-samples to keep raw diversity in drama visuals.

    2026-09-02 · fine-tune · 440 images · 3,229 image likes


    ▍ 红潮 · Krea 2 Turbo · GPT逼真版 INT4/INT8

    Open → Civitai · GPT逼真版 INT4/INT8 Convrot

    Krea 2 量化写实Base model,用高噪/低噪双权重交替采样出影视剧感画面。
    Quantised Krea 2 checkpoint from REDGPT for realistic drama stills, shipped as paired high-noise and low-noise weights.

    2026-07-18 · fine-tune · 9,197 downloads · 428 likes · 18 images · 1,379 image likes

    ▍ 红潮 · Krea 2 Turbo · KREA2 GPT逼真版 FP8

    Open → Civitai · KREA2 GPT逼真版

    红潮 Krea 2 逼真版影视向Base model,无码出图,供角色图与分镜素材。
    Photoreal, uncensored Krea 2 checkpoint from RedCraft made for film and drama stills, character art and storyboards.

    2026-07-13 · fine-tune · 12,175 downloads · 676 likes · 102 images · 3,320 image likes


    ▍ 红潮 · Krea 2 Turbo · 赤佬 2.0

    Open → Civitai · 赤佬2 (Krea2)Edition

    Krea 2 上的无码混合版,CFG 1、八步出图。
    No-mosaic Krea 2 mix tuned for fast eight-step output at CFG 1.

    2026-07-01 · fine-tune · 19,436 downloads · 832 likes · 187 images · 3,059 image likes

    ▍ 红潮 · Krea 2 Turbo · 赤佬 1.1

    Open → Civitai · 赤佬1.1 INT8 & INT4

    Raw+Turbo 合并的无审查 Krea 2 模型,CFG 1、十步出图。
    Raw plus Turbo merged Krea 2 checkpoint with zero censorship, running CFG 1 in ten steps.

    2026-06-24 · fine-tune · 22,676 downloads · 992 likes · 278 images · 2,929 image likes


    ▍Dark Beast 黑兽

    4 works · 52,063 downloads

    ▍ 黑兽 · Krea 2 Turbo · Beast 5C 黑兽5.0 ⌝

    Open → Civitai · Beast 5C (OCT→SALE ) K2T

    无审查 Krea 2 Turbo 大模型,八步出彩色年轻角色。(4.0 因审查取消)
    Uncensored Krea 2 Turbo checkpoint for colorful youth characters, single 8-step pass. (4.0 Canceled)

    2026-09-28 · fine-tune · 3,939 downloads · 446 likes · 279 images · 4,244 image likes


    ▍ 黑兽 · Krea 2 Turbo · Dark Beast 3 黑兽3.0 ⌝ 

    Open → Civitai · Dark Beast 3 黑兽3.0

    Krea 2 上的黑兽 3.0 硬核版,无审查、逐像素质感、无马赛克。
    Krea 2 hardcore Dark Beast 3 edition: uncensored, per-pixel texture, zero mosaics.

    2026-07-28 · fine-tune · 12,995 downloads · 519 likes · 163 images · 3,202 image likes


    ▍ 黑兽 · Krea 2 Turbo · Dark Beast 2 黑兽K2T FP8

    Open → Civitai · Dark Beast 2 黑兽K2T FP8

    Krea 2 底的黑兽硬核版,FP8 权重,无审查、无马赛克。
    Krea 2 hardcore Dark Beast edition in FP8: uncensored, detail-obsessed, zero mosaics.

    2026-06-28 · fine-tune · 27,696 downloads · 1,181 likes · 205 images · 3,189 image likes

    ▍ 黑兽 · Krea 2 Turbo · KREA2 黑兽1.1 INT8 Convrot

    Open → Civitai · KREA2 黑兽1.1 INT8 Convrot

    Krea 2 黑兽无审查版,INT8 Convrot 量化,零马赛克。
    Uncensored Krea 2 Dark Beast edition, INT8 Convrot for ComfyUI, aiming at explicit detail with zero mosaics.

    2026-07-02 · fine-tune · 7,433 downloads · 289 likes · 81 images · 1,311 image likes


    ▍Tooling · Acceleration · Workflows

    1 works · 5,157 downloads

    ▍ 工具 · Krea 2 Turbo · Workflow (赤佬版) ⌝

    Open → Civitai · Boogu & Krea2(赤佬版)

    多种模型混合生成 的 ComfyUI Workflows:密集文字+实时增强。
    ComfyUI workflow pairing any fine-tunes with Krea2 for dense text plus real-time polish.

    2026-06-24 · workflow · 5,157 downloads · 104 likes · 35 images · 261 image likes


    Qwen image 2.1 Fine-tune / LoRas

    First used 2026-09-20 · 3 brands · 8 works · 23,287 downloads

    ▍FineView

    1 works · 778 downloads

    ▍ 欢视 · Qwen 2.1 · QW21 Turbo Fixed ⌝

    Open → Civitai · QW21 Turbo Fixed

    Qwen-Image-2.1 的 RED 纹理修复微调版,去棋盘格与 NaN。
    REDQW21 Turbo texture-fixed fine-tune of Qwen-Image-2.1, targeting clean textures without VAE artifacts.

    2026-10-15 · fine-tune · 47 images · 569 image likes


    ▍RedCraft

    3 works · 16,736 downloads

    ▍ 红潮 · Qwen 2.1 · RedCraft 3.0 (FixQW21T)

    Open → Civitai · RedCraft 3.0 (FixQW21T)

    Qwen 2.1 大改微调,修复 Turbo 质感,支持多图参考改图。
    Heavily modified Qwen-Image 2.1 fine-tune fixing Turbo texture, with text-to-image and multi-reference editing.

    2026-10-15 · fine-tune · 3,964 downloads · 137 likes · 60 images · 719 image likes

    ▍ 红潮 · Qwen 2.1 · REDQW21(UNLOCKED)v2 TI2I

    Open → Civitai · REDQW21(UNLOCKED)v2 TI2I

    Qwen 2.1 解锁版 v2,在文生图之外补上多参图生图改图。
    v2 of the unlocked Qwen 2.1 fine-tune, adding multi-reference text-image-to-image editing on top of text-to-image.

    2026-09-30 · fine-tune · 5,118 downloads · 196 likes · 139 images · 1,445 image likes

    ▍ 红潮 · Qwen 2.1 · REDQW21(UNLOCKED)v1 T2I

    Open → Civitai · REDQW21(UNLOCKED)v1 T2I

    Qwen 2.1 解锁版 v1:文生图加多图参考改图,配Workflows。
    Unlocked Qwen 2.1 fine-tune for text-to-image and multi-reference editing, shipped with a matching ComfyUI workflow.

    2026-09-24 · fine-tune · 7,654 downloads · 245 likes · 84 images · 931 image likes


    ▍Tooling · Acceleration · Workflows

    4 works · 5,773 downloads

    ▍ 工具 · Qwen 2.1 · REDQW21(UNLOCKED) 💴

    Open → Civitai · REDQW21(UNLOCKED) 💴

    REDQW21 UNLOCKED 挑战赛的付费对齐Workflows,仅供研究评估。
    Paid aligned Qwen-Image 2.1 workflow entered in the REDQW21 UNLOCKED challenge, research and evaluation only.

    2026-09-24 · workflow · 45 images · 549 image likes

    ▍ 工具 · Qwen 2.1 · TURBO(Viggle)QW21 v0.2.1

    Open → Civitai · TURBO(Viggle)QW21 v0.2.1

    Viggle 的 Qwen-Image 2.1 四步蒸馏版,文生图加参考改图。
    Viggle's 4-step Qwen-Image 2.1 turbo repost: text-to-image and 1-3 reference edits, no CFG.

    2026-09-23 · LoRA · 4,653 downloads · 173 likes · 28 images · 733 image likes

    ▍ 工具 · Qwen 2.1 · QW21(DLSS5)LUT NV20-50

    Open → Civitai · QW21(DLSS5)LUT NV20-50

    Qwen Image 2.1 Workflows,十图参考改图,incl. DLSS5、LUT。
    Qwen Image 2.1 workflow with EasyCache, ten-image reference edits, DLSS5 enhancement and a LUT loader.

    2026-09-20 · workflow · 316 downloads · 29 likes · 36 images · 439 image likes

    ▍ 工具 · Qwen 2.1 · QW21加速器(Native acceler)

    Open → Civitai · QW21加速器(Native acceler)

    Qwen Image 2.1 素版Workflows,原生 EasyCache 压步数。
    Lean Qwen Image 2.1 workflow on ComfyUI's native EasyCache, plus multi-reference, VSR upscale and batch LoRA nodes.

    2026-09-20 · workflow · 748 downloads · 37 likes · 82 images · 679 image likes


    MiniMax H3 Fine-tune / LoRas

    First used 2026-08-09 · 4 brands · 7 works · 31,409 downloads

    ▍FineView

    1 works · 2 downloads

    ▍ 欢视 · MiniMax H3 · ASOP H3 Director💰

    Open → Civitai · ASOP H3 Director💰

    MiniMax H3 的付费导演版服务包,自研推理框架,30 秒以上单次成片。
    Paid MiniMax H3 director package with a proprietary framework for 30s+ single-pass short-drama work.

    2026-10-02 · fine-tune · 2 downloads · 79 likes · 36 images · 3,711 image likes


    ▍RedCraft

    2 works · 11,363 downloads

    ▍ 红潮 · MiniMax H3 · 💰H3 RED-MAAS(ToB)

    Open → Civitai · 💰H3 RED-MAAS(ToB)

    H3 出草稿、LTX-2.5 放大的双模型方案,16GB 显存出 2K。
    Two-model H3 plus LTX-2.5 pipeline for 2K video: fast semantic draft, then IC-guided upscale.

    2026-08-24 · fine-tune · 10 downloads · 340 likes


    ▍Tooling · Acceleration · Workflows

    3 works · 19,064 downloads

    ▍ 工具 · MiniMax H3 · H3-LTX25-神速重绘⚡

    Open → Civitai · H3-LTX25-神速重绘⚡

    H3 先出带声草稿,LTX-2.5 重绘,RIFE 插帧到 48fps。
    H3 drafts a clip with audio, LTX-2.5 IC-LoRA re-renders it as V2V, RIFE lifts it to 48 fps.

    2026-08-24 · workflow · 74 downloads · 104 likes

    ▍ 工具 · MiniMax H3 · MiniMax H3 Lightx2v 8step

    Open → Civitai · MiniMax H3 Lightx2v 8step

    LightX2V 蒸馏的 H3 8 步加速Workflows,配剪枝 20B FL2VA。
    MiniMax H3 eight-step turbo workflow built on LightX2V distillation, for pruned 20B FL2VA and NFE comparisons.

    2026-08-09 · workflow · 5,837 downloads 860 image likes

    ▍ 工具 · MiniMax H3 · lightx2v / Minimax-h3 1.0

    Open → Civitai · lightx2v / Minimax-h3 1.0

    LightX2V 的 H3 少步蒸馏,配剪枝 20B FL2VA。
    LightX2V's MiniMax-H3-Turbo distillation for pruned 20B FL2VA video, with NFE and LoRA comparisons.

    2026-08-09 · LoRA · 13,153 downloads · 516 likes


    ▍RedCraft / Dark Beast Collection

    1 works · 980 downloads

    ▍ LoRAs合集 · MiniMax H3 · H3 Exploded view

    Open → Civitai · H3 Exploded view

    MiniMax H3 拆解动画 LoRA,主体炸散成零件再原路装回。
    MiniMax H3 LoRA that makes a subject explode into parts, scatter wide, then reassemble.

    2026-10-02 · LoRA · 980 downloads · 148 likes


    MOSS Image Fine-tune / LoRas

    First used 2026-10-18 · 2 brands · 3 works · 62 downloads

    ▍FineView

    2 works · 47 downloads

    ▍ 欢视 · MOSS Image · MOSS 2 Asian 6NFE

    Open → Civitai · MOSS 2 Asian 6NFE

    MOSS 2 东亚专属版,6 步出图,带合并加速与 LCS。
    East Asian tuned MOSS 2 checkpoint at 6 NFE, with merged acceleration and LCS color control.

    2026-10-19 · fine-tune · 25 downloads · 25 likes · 21 images · 248 image likes

    ▍ 欢视 · MOSS Image · MOSS-Image v1.0 6步NSFW

    Open → Civitai · MOSS-Image v1.0 6步NSFW

    MOSS-Image 6 步 NSFW 基版,配 LCS 控色控锐。
    6-step NSFW MOSS-Image build on Euler/Simple and CFG=1, with LCS nodes for color and sharpness.

    2026-10-18 · fine-tune · 22 downloads · 30 likes · 33 images · 411 image likes


    ▍Tooling · Acceleration · Workflows

    1 works · 15 downloads

    ▍ 工具 · MOSS Image · MOSS 6-Steps Sharpen(LCS)

    Open → Civitai · MOSS 6-Steps Sharpen(LCS)

    MOSS Image 的 6 步Workflows,用 LCS 在采样期调色加锐。
    MOSS Image six-step workflow with LCS nodes adding training-free colour and sharpness control during sampling.

    2026-10-18 · workflow · 15 downloads · 17 likes · 89 images · 474 image likes


    Illustrious Fine-tune / LoRas

    First used 2024-11-28 · 2 brands · 9 works · 56,874 downloads

    ▍RedCraft

    4 works · 43,463 downloads

    ▍ 红潮 · Illustrious · Illust3Relustion🔥

    Open → Civitai · Illust3Relustion🔥

    Illustrious 3 Base model,按 2M 像素出动漫人物图,CFG 5.5。
    Illustrious 3 Relustion checkpoint designed for Hi-RES 2M output at CFG 5.5 and around 30 steps.

    2025-03-13 · fine-tune · 28,552 downloads · 1,845 likes · 102 images · 4,148 image likes

    ▍ 红潮 · Illustrious · 光輝2Relustion[IL]

    Open → Civitai · 光輝2Relustion[IL]

    复古真实系 Illustrious 微调,CFG 5、25~30 步出图。
    Retro East-Asian photoreal Illustrious fine-tune with SD-era texture, best at CFG 5 and 25-30 steps.

    2025-01-19 · fine-tune · 6,119 downloads · 396 likes · 147 images · 592 image likes

    ▍ 红潮 · Illustrious · Relustion ULTRA🔥

    Open → Civitai · Relustion ULTRA🔥

    整合 CADS3 NSFW 权重的 Illustrious 极限写实Base model。
    Extreme-realism Illustrious checkpoint merging CADS3 NSFW weights for harder detail and high-resolution output.

    2024-12-11 · fine-tune · 5,052 downloads · 233 likes · 223 images · 634 image likes

    ▍ 红潮 · Illustrious · Relustion IL NSFW

    Open → Civitai · Relustion IL NSFW

    Illustrious 上的亚洲写实 NSFW Base model,比 V1 更厚实。
    Illustrious-based Asian realistic NSFW checkpoint with a bolder, less washed-out look than v1.

    2024-11-28 · fine-tune · 3,740 downloads · 192 likes · 101 images · 364 image likes


    ▍RedCraft LoRA

    5 works · 13,411 downloads

    ▍ 红潮 LoRA · Illustrious · vol.6 Ripping-off (Re3IL)

    Open → Civitai · vol.6 Ripping-off (Re3IL)

    LSP 的脱衣版 LoRA,基于 Relustion3IL,做撕衣镜头。
    Clothing-removal LSP LoRA for Illustrious/Relustion3IL, aimed at undressing shots and video generation.

    2025-04-29 · LoRA · 1,417 downloads · 103 likes · 14 images · 159 image likes

    ▍ 红潮 LoRA · Illustrious · vol.2 barefoot(IL)

    Open → Civitai · vol.2 barefoot(IL)

    LSP 的 Illustrious 版,配红潮 CADS IL,裸足触发词照旧。
    Illustrious cut of the LSP LoRA for CADS Relustion IL, keeping the barefoot trigger on an NSFW Asian base.

    2024-12-31 · LoRA · 1,589 downloads · 126 likes · 44 images · 230 image likes

    ▍ 红潮 LoRA · Illustrious · vol.6 CIVIL-HOT(IL)🔥

    Open → Civitai · vol.6 CIVIL-HOT(IL)🔥

    Illustrious 版红潮 LoRA,做 CIVIL-HOT 片商封面。
    Illustrious LoRA for CIVIL-HOT JAV covers, matched to the RedCraft ULTRA realistic base.

    2024-12-25 · LoRA · 6,813 downloads · 606 likes · 33 images · 293 image likes

    ▍ 红潮 LoRA · Illustrious · vol.1 AiPONDO (IL)

    Open → Civitai · vol.1 AiPONDO (IL)

    Illustrious Base model的 AiPONDO 厂牌无码成人向 LoRA。
    Illustrious LoRA for uncensored adult scenes in the AiPONDO label style.

    2024-12-18 · LoRA · 1,087 downloads · 83 likes · 71 images · 131 image likes

    ▍ 红潮 LoRA · Illustrious · vol.4 AiONE (IL)🔥

    Open → Civitai · vol.4 AiONE (IL)🔥

    Illustrious 版红潮 LoRA,做 AI ONE 片商的图。
    Illustrious LoRA for the AI ONE label, tuned to the RedCraft Relustion IL NSFW base.

    2024-12-13 · LoRA · 2,505 downloads · 227 likes · 12 images · 139 image likes


    LTXV 2.5 / 2.3 Fine-tune / LoRas

    First used 2026-07-17 · 2 brands · 4 works · 27,890 downloads

    ▍FineView

    2 works · 19,895 downloads

    ▍ 欢视 · LTXV 2.5 · LTX 2.5 REDGraft(NSFW)

    Open → Civitai · LTX 2.5 REDGraft(NSFW)

    LTX-2.5 权重移植版 REDGraft,Fast 2K,标 NSFW。
    REDGraft, a sulphur2 same-architecture port of LTX-2.5 for fast 2K video; tagged NSFW.

    2026-08-21 · fine-tune · 10,912 downloads · 313 likes

    ▍ 欢视 · LTXV 2.3 · LTXV 2.3高速版 INT4 ConvRot

    Open → Civitai · LTXV 2.3高速版 INT4 ConvRot

    LTXV 2.3 的 INT4 ConvRot 量化高速版,原生节点直出短视频。
    INT4 ConvRot quantized LTXV 2.3 checkpoint for ComfyUI 0.27, made for fast 25-frame clips.

    2026-07-19 · fine-tune · 8,983 downloads · 402 likes


    ▍Tooling · Acceleration · Workflows

    2 works · 7,995 downloads

    ▍ 工具 · LTXV 2.5 · LTX25 Pruned 8+3 Fast

    Open → Civitai · LTX25 Pruned 8+3 Fast

    LTXV 2.5 剪枝版少步Workflows,8 步加 3 步两段采样快出片。
    Pruned LTXV 2.5 workflow running a fast two-stage 8-plus-3 step pass for quick video drafts.

    2026-08-18 · workflow · 3,049 downloads · 88 likes

    ▍ 工具 · LTXV 2.3 · LTX2.3REDMixKrea2

    Open → Civitai · LTX2.3REDMixKrea2

    LTXV 2.3 的 REDMix + Krea2 视频Workflows,属工具类合集。
    LTXV 2.3 video workflow pairing a REDMix video model with Krea2, part of the author's ComfyUI workflow shelf.

    2026-07-17 · workflow · 4,946 downloads · 85 likes


    Wan Video Fine-tune / LoRas

    First used 2025-02-26 · 2 brands · 13 works · 143,535 downloads

    ▍FineView

    12 works · 141,211 downloads

    ▍ 欢视 · Wan Video · CausVid14B(适配VACE)🔥

    Open → Civitai · CausVid14B(适配VACE)🔥

    WAN 14B 的 CausVid DMD2 加速版,适配 VACE 参考帧。
    CausVid DMD2 accelerator for WAN 14B: 6-16 step video, best adapted to WAN-VACE reference frames.

    2025-05-12 · fine-tune · 15,618 downloads · 896 likes · 3 images · 1,276 image likes

    ▍ 欢视 · Wan Video · 首尾帧-FLF2V-14B-720P

    Open → Civitai · 首尾帧-FLF2V-14B-720P

    Wan 2.1 官方首尾帧 14B 720P 视频模型,两张定帧补中间动态。
    Wan 2.1 official first-and-last-frame 14B video checkpoint at 720P for ComfyUI transitions.

    2025-04-20 · fine-tune · 3,616 downloads · 359 likes · 2 images · 292 image likes

    ▍ 欢视 · Wan Video · ComfyORG i2v-720p GGUF

    Open → Civitai · ComfyORG i2v-720p GGUF

    Wan 2.1 图生视频 720p 权重的 ComfyUI 重打包版。
    Comfy-Org repack of Wan 2.1 image-to-video 720p for ComfyUI, with GGUF and fp8 files for smaller cards.

    2025-03-18 · fine-tune · 3,741 downloads · 215 likes

    ▍ 欢视 · Wan Video · ComfyORG i2v-480p GGUF

    Open → Civitai · ComfyORG i2v-480p GGUF

    Comfy-Org 重打包的 Wan 2.1 图生视频 480p GGUF 版。
    Comfy-Org's repackaged Wan 2.1 image-to-video 480p model in GGUF form, with file placement notes.

    2025-03-05 · fine-tune · 13,614 downloads · 843 likes · 5 images · 610 image likes

    ▍ 欢视 · Wan Video · ComfyORG wan_2.1_vae

    Open → Civitai · ComfyORG wan_2.1_vae

    Wan 2.1 视频 VAE 单文件版,Comfy-Org 重打包。
    The Wan 2.1 video VAE on its own, repacked by Comfy-Org for ComfyUI's vae folder.

    2025-03-04 · fine-tune · 5,849 downloads · 95 likes · 12 images · 223 image likes

    ▍ 欢视 · Wan Video · Kijai-Wan2.1-I2V-14B-720P

    Open → Civitai · Kijai-Wan2.1-I2V-14B-720P

    Kijai 打包的 Wan 2.1 14B 图生视频 720P 权重。
    Kijai's ComfyUI packaging of Wan 2.1 14B image-to-video at 720p, the higher-resolution half of the pair.

    2025-02-26 · fine-tune · 25,528 downloads · 1,944 likes · 7 images · 892 image likes

    ▍ 欢视 · Wan Video · Kijai-Wan2.1-I2V-14B-480P

    Open → Civitai · Kijai-Wan2.1-I2V-14B-480P

    Kijai 打包的 Wan 2.1 14B 图生视频 480P 权重。
    Kijai's ComfyUI packaging of Wan 2.1 14B image-to-video at 480p, tested at 512×512 with 81 frames.

    2025-02-26 · fine-tune · 43,002 downloads · 3,726 likes

    ▍ 欢视 · Wan Video · Kijai-Wan2.1-T2V-14B

    Open → Civitai · Kijai-Wan2.1-T2V-14B

    Kijai 打包的 Wan 2.1 文本生视频 14B 主干。
    Kijai's ComfyUI packaging of Wan 2.1 text-to-video 14B, the full-size model for prompt-driven shots.

    2025-02-26 · fine-tune · 8,364 downloads · 410 likes

    ▍ 欢视 · Wan Video · Kijai-Wan2.1-T2V-1.3B

    Open → Civitai · Kijai-Wan2.1-T2V-1.3B

    Kijai 打包的 Wan 2.1 文本生视频 1.3B,轻量可跑。
    Kijai's ComfyUI packaging of Wan 2.1 text-to-video 1.3B, the light model that runs on consumer GPUs.

    2025-02-26 · fine-tune · 5,793 downloads · 132 likes

    ▍ 欢视 · Wan Video · Kijai_Wan2_1_VAE

    Open → Civitai · Kijai_Wan2_1_VAE

    Kijai 打包的 Wan 2.1 视频 VAE,bf16 单文件。
    Wan 2.1 video VAE in bf16, Kijai's standalone repack for ComfyUI's vae folder.

    2025-02-26 · fine-tune · 7,930 downloads · 87 likes

    ▍ 欢视 · Wan Video · Kijai-umt5-xxl-enc-Wan

    Open → Civitai · Kijai-umt5-xxl-enc-Wan

    Wan 2.1 用的 umt5-xxl 文本编码器 bf16 版。
    Wan 2.1's umt5-xxl text encoder in bf16, a standalone safetensors for ComfyUI's text_encoders folder.

    2025-02-26 · fine-tune · 4,321 downloads · 41 likes

    ▍ 欢视 · Wan Video · Kijai-open-clipl-vit-Wan

    Open → Civitai · Kijai-open-clipl-vit-Wan

    Kijai 打包的 CLIP 视觉编码器 fp16,WanVideo 专用。
    CLIP vision encoder (open-clip xlm-roberta-large ViT-H/14) in fp16, packaged by Kijai for the WanVideo wrapper.

    2025-02-26 · fine-tune · 3,835 downloads · 35 likes


    ▍WAN LoRA

    1 works · 2,324 downloads

    ▍ 红潮 LoRA · Wan Video · vol.7 2Girls1Dick(WAN)

    Open → Civitai · vol.7 2Girls1Dick(WAN)

    WAN 2.1 图生视频 LoRA,做 JAV 风格的双女 POV 口交镜头。
    WAN 2.1 image-to-video LoRA for a two-girl POV oral scene in JAV style.

    2025-04-22 · LoRA · 2,324 downloads · 226 likes · 11 images · 160 image likes


    ZImage Base / Turbo Fine-tune / LoRas

    First used 2025-12-02 · 6 brands · 27 works · 229,998 downloads

    ▍RedCraft

    7 works · 89,362 downloads

    ▍ 红潮 · ZImage · RedZiT2⚡️红潮2倍速 2026HD

    Open → Civitai · RedZiT2⚡️红潮2倍速 2026HD

    ZImage Turbo 的 2026 高清重制,INT8 规格,面向商用。
    2026 HD rebuild of the ZImage Turbo line on an INT8 Convrot spec, aimed at commercial HD output.

    2026-07-05 · fine-tune · 6,813 downloads · 218 likes · 83 images · 856 image likes

    ▍ 红潮 · ZImage · ZIB Distilled⚡️FUNxAGILE

    Open → Civitai · ZIB Distilled⚡️FUNxAGILE

    减掉 DPO 与蒸馏权重的 Z-Image Versions,随机性更强。
    Z-Image AGILE build with reduced DPO and distilled weight, tuned for variety and richer compositions.

    2026-03-08 · fine-tune · 9,704 downloads · 378 likes · 143 images · 1,660 image likes

    ▍ 红潮 · ZImage · ZIB Distilled⚡️DPO Veris

    Open → Civitai · ZIB Distilled⚡️DPO Veris

    ZIB 引擎加 DPO 调校的 Z-Image 蒸馏Base model,光影更干净。
    Z-Image distilled checkpoint on the ZIB engine with DPO tuning for cleaner light and color.

    2026-02-26 · fine-tune · 5,593 downloads · 315 likes · 119 images · 1,181 image likes

    ▍ 红潮 · ZImage · ZIB Distilled⚡️DX3 Lucis

    Open → Civitai · ZIB Distilled⚡️DX3 Lucis

    5 步出图的 Z-Image 蒸馏Base model,用于快速验证 LoRA。
    Five-step Z-Image distilled checkpoint for instant LoRA prototyping and high-entropy pre-sampling.

    2026-02-11 · fine-tune · 16,988 downloads · 428 likes · 194 images · 1,328 image likes

    ▍ 红潮 · ZImage · ZIB Distilled⚡️RedDX2

    Open → Civitai · ZIB Distilled⚡️RedDX2

    红潮 RedDX2 蒸馏版 Z-Image 检查点,CFG 1.0 十步出图。
    Distilled Z-Image checkpoint of the RedCraft RedDX2 line, running at CFG 1.0 in about ten steps.

    2026-02-05 · fine-tune · 5,758 downloads · 258 likes · 167 images · 1,128 image likes

    ▍ 红潮 · ZImage · REDZIB⚡️DX1加速版

    Open → Civitai · REDZIB⚡️DX1加速版

    Z-Image 原版的蒸馏加速版,保留原味风格与 LoRA 兼容性。
    Distilled speed-up of the original Z-Image (non-Turbo) base, keeping its look and LoRA compatibility.

    2026-01-30 · fine-tune · 5,514 downloads · 250 likes · 67 images · 1,193 image likes

    ▍ 红潮 · ZImage · RedZiT⚡️1.5AIO

    Open → Civitai · RedZiT⚡️1.5AIO

    蒸馏自真实照片采样的 ZImage Turbo 模型,有多档权重形式。
    ZImage Turbo checkpoint distilled from Nano Banana Pro photo sampling, shipping as AIO, BF16/FP8 and LoRA.

    2025-12-02 · fine-tune · 38,992 downloads · 1,738 likes · 267 images · 2,076 image likes


    ▍Dark Beast

    8 works · 86,346 downloads

    ▍ 黑兽 · ZImage · DBZiT9🟥DIM RClaw

    Open → Civitai · DBZiT9🟥DIM RClaw

    黑兽 9 融进 Moody Porn Mix 8,为求又快又少出错而调。
    Dark Beast 9 fused onto a Moody Porn Mix 8 backbone, tuned for fast, low-error output.

    2026-03-20 · fine-tune · 17,985 downloads · 657 likes · 56 images · 843 image likes

    ▍ 黑兽 · ZImage · DBZiT8🟥SDA@FOK

    Open → Civitai · DBZiT8🟥SDA@FOK

    ZiT 黑兽版,合入 SDA 训练的 LoKr,专治 8 步出图同质化。
    ZiT black-beast checkpoint merging an SDA-trained LoKr to undo diversity collapse at 8 steps.

    2026-03-15 · fine-tune · 6,119 downloads · 280 likes · 84 images · 679 image likes

    ▍ 黑兽 · ZImage · DarkBlitZ6🟥BeyondNSFW

    Open → Civitai · DarkBlitZ6🟥BeyondNSFW

    ZImage Turbo 的 NSFW 解锁版,色彩更艳、纹理更锐。
    NSFW-unlocked ZImage Turbo checkpoint on Beyond Reality 3, with richer color and no trigger words.

    2026-02-18 · fine-tune · 32,420 downloads · 1,152 likes · 173 images · 1,596 image likes

    ▍ 黑兽 · ZImage · DarkBlitZ🟥ZIT 5steps

    Open → Civitai · DarkBlitZ🟥ZIT 5steps

    黑兽 ZImage Turbo 的 5 步蒸馏版,CFG=1 直出。
    ZImage Turbo dark Beast distillation tuned for 5 steps at fixed CFG=1, built purely for speed.

    2026-02-12 · fine-tune · 7,334 downloads · 274 likes · 261 images · 1,849 image likes

    ▍ 黑兽 · ZImage · DBZ IV⚡️Allstar

    Open → Civitai · DBZ IV⚡️Allstar

    ZiT 黑兽最终版,基于 RedZ1.5 重修畸变,稳定好挂 LoRA。
    Closing Dark Beast build on ZiT, rebuilt from RedZ1.5 to correct distortion — stable and LoRA-friendly.

    2026-01-27 · fine-tune · 7,338 downloads · 532 likes · 47 images · 920 image likes

    ▍ 黑兽 · ZImage · DBZ-4K⚡️v3.1

    Open → Civitai · DBZ-4K⚡️v3.1

    ZiT 融合模型,主攻超写实情色影像,可出 16M 像素大图。
    Merged ZiT model for hyper-real erotic imagery, using Dype scale and SeedVR2 toward 16M-pixel output.

    2026-01-16 · fine-tune · 4,751 downloads · 270 likes · 92 images · 1,414 image likes

    ▍ 黑兽 · ZImage · DBZ-LSP⚡️v2.5

    Open → Civitai · DBZ-LSP⚡️v2.5

    把 LSP 小姿势题材做成 ZiT Turbo Base model版,主打标签抽卡。
    The LSP position concept as a ZiT Turbo-base checkpoint, built for relentless tag gacha.

    2026-01-08 · fine-tune · 3,881 downloads · 227 likes · 149 images · 1,070 image likes

    ▍ 黑兽 · ZImage · DBZ-AIO⚡v1.0

    Open → Civitai · DBZ-AIO⚡v1.0

    ZImageTurbo 底的黑兽 AIO 整合版,作者未留本版Notes。
    All-in-one Dark Beast checkpoint on ZImageTurbo; the author left no version notes for this build.

    2025-12-21 · fine-tune · 6,518 downloads · 451 likes · 361 images · 1,769 image likes


    ▍Tooling · Acceleration · Workflows

    5 works · 27,391 downloads

    ▍ 工具 · ZImage · 🟥Z&Klein9b🟦Hybrid BFS

    Open → Civitai · 🟥Z&Klein9b🟦Hybrid BFS

    Klein 9B 加速版上的 BFS 换脸Workflows,修掉肖像参考不跟图。
    Klein 9B accelerated workflow built for BFS face swaps, with a fix for portrait reference drift.

    2026-03-19 · workflow · 3,661 downloads · 118 likes · 65 images · 647 image likes

    ▍ 工具 · ZImage · DarkBeastZit8🟥SDA WF

    Open → Civitai · DarkBeastZit8🟥SDA WF

    ZImage Turbo 上的暗兽系Workflows,用 LCS 节点在采样期调色加锐。
    ZImage Turbo workflow from the Dark Beast line, using LCS nodes to steer colour and sharpness mid-sampling.

    2026-03-15 · workflow · 4,866 downloads · 132 likes · 29 images · 331 image likes

    ▍ 工具 · ZImage · ZIB Distilled

    Open → Civitai · ZIB Distilled

    细节滑杆 LoRA 的 ZIB / Distilled 版:只调权重控细节。
    The ZIB and ZImage Distilled build of the DetailDeamonZ detail-slider LoRA.

    2026-02-09 · LoRA · 6,125 downloads · 343 likes · 83 images · 575 image likes

    ▍ 工具 · ZImage · DetailDeamonZ Slider For Z

    Open → Civitai · DetailDeamonZ-SliderLoRA

    给 ZImageTurbo 的细节滑杆 LoRA,干净与细节爆炸随权重切换。
    A detail-slider LoRA for ZImageTurbo that trades clean minimalism for extreme detail.

    2025-12-09 · LoRA · 11,979 downloads · 812 likes · 235 images · 974 image likes

    ▍ DetailDeamonZ-Slider 细节魔控LoRA for ZImageTurbo & Re

    Open → Liblib · DetailDeamonZ.1

    ZImageTurbo 的细节滑杆 LoRA,靠权重正负调细节多少。
    Detail slider LoRA for ZImageTurbo and RedZ1.5: negative weight adds detail, positive weight cleans it up.

    2025-12-09 · LoRA · 760 downloads · 22 likes


    ▍Character LoRA

    4 works · 15,195 downloads

    ▍ 角色 · ZImage · v1.0 Rank128

    Open → Civitai · v1.0 Rank128

    Kawaii Kinky 的 rank 128 版,给成年写真加一层撩人情绪。
    Kawaii Kinky rank-128 LoRA layering a teasing, seductive attitude onto adult ZImage Turbo portraits.

    2025-12-20 · LoRA · 5,429 downloads · 478 likes · 399 images · 759 image likes

    ▍ 角色 · ZImage · v1.0 Lossless

    Open → Civitai · v1.0 Lossless

    Kawaii Kinky v1 无损权重,11.46 GB 成年向人像。
    Kawaii Kinky v1 as an 11.46 GB FP16 full-difference LoRA for seductive adult portraits on ZImage Turbo.

    2025-12-17 · LoRA · 772 downloads · 59 likes · 39 images · 91 image likes

    ▍ 角色 · ZImage · Teeny Moody 小情绪 For ZImage

    Open → Civitai · v1.0

    ZImage Turbo 的情绪滤镜,让日常人像多一层「小情绪」。
    Teeny Moody LoRA giving everyday ZImage Turbo portraits a soft, melancholy expression filter.

    2025-12-11 · LoRA · 8,961 downloads · 620 likes · 109 images · 832 image likes

    ▍ Teeny Moody 小情绪 LoRA For ZImageTurbo & RedZ1.5

    Open → Liblib · Teeny Moody v1

    Z-Image 的表情 LoRA,给 20 岁上下的人像加一层小情绪微表情。
    Z-Image expression LoRA for tiny moody micro-expressions on young-adult portraits.

    2025-12-09 · LoRA · 33 downloads · 1 likes


    ▍Zimage LoRA

    2 works · 8,557 downloads

    ▍ 红潮 LoRA · ZImage · vol.7 ZIB⚡️JAV Girls

    Open → Civitai · vol.7 ZIB⚡️JAV Girls

    ZImage 版 JAV 女优风格 LoRA,用触发词还原日系成人片气质。
    JAV-style character LoRA for RedCraft REDZimage DX1 on ZImageTurbo, adding Japanese adult-video performer looks.

    2026-02-02 · LoRA · 2,322 downloads · 218 likes · 28 images · 415 image likes

    ▍ 红潮 LoRA · ZImage · LSP⚡️Z Full-Rank 1.0

    Open → Civitai · LSP⚡️Z Full-Rank 1.0

    LSP 姿势 LoRA 的无损全秩版,适配 Nunchaku SVDQ。
    Lossless full-rank FP16 build of the LSP sex-position LoRA, tuned for Nunchaku SVDQ Z-Image.

    2026-01-04 · LoRA · 6,235 downloads · 426 likes · 76 images · 617 image likes


    ▍RedCraft / Dark Beast Collection

    1 works · 3,147 downloads

    ▍ 红潮合集 · ZImage · DBZiT8🟥SDA LoRA

    Open → Civitai · DBZiT8🟥SDA LoRA

    Dark Beast Character LoRA,跑在 ZImageTurbo Base model上。
    Dark Beast character LoRA for ZImageTurbo, the ZIT counterpart to the Klein 9B release.

    2026-03-20 · LoRA · 3,147 downloads · 169 likes · 6 images · 110 image likes


    Flux.2 Klein 9B Fine-tune / LoRas

    First used 2026-02-03 · 3 brands · 6 works · 48,518 downloads

    ▍Dark Beast

    4 works · 36,705 downloads

    ▍ 黑兽 · Flux.2 Klein 9B · DBKleinV2🟦BFS

    Open → Civitai · DBKleinV2🟦BFS

    主打换脸的 Flux.2 Klein 9B 大模型,内置 BFS 保身份。
    Face-swap focused Flux.2 Klein 9B checkpoint with BFS for stable identity transfer.

    2026-03-03 · fine-tune · 21,451 downloads · 658 likes · 14 images · 352 image likes

    ▍ 黑兽 · Flux.2 Klein 9B · DBK 9b V1 BASE

    Open → Civitai · DBK 9b V1 BASE

    黑兽线的未加速 Klein 9B 原始权重,供继续训练使用。
    Unaccelerated Flux.2 Klein 9B base weights of the Dark Beast line, released for training.

    2026-02-19 · fine-tune · 906 downloads · 41 likes · 6 images · 19 image likes

    ▍ 黑兽 · Flux.2 Klein 9B · DBK🟦BlitZ V1.5

    Open → Civitai · DBK🟦BlitZ V1.5

    黑兽风格 Klein 9B 极速版,5 步、CFG=1 直出。
    Dark Beast style on Flux.2 Klein 9B, distilled to run in 5 steps at fixed CFG=1.

    2026-02-08 · fine-tune · 11,171 downloads · 272 likes · 34 images · 261 image likes

    ▍ 黑兽 · Flux.2 Klein 9B · DBK🟦Mixed V1

    Open → Civitai · DBK🟦Mixed V1

    黑兽风格 Klein 9B 折中版,10–20 步、CFG=1。
    Dark Beast style on Flux.2 Klein 9B, tuned for 10–20 steps at CFG=1.

    2026-02-03 · fine-tune · 3,177 downloads · 147 likes · 64 images · 388 image likes


    ▍Tooling · Acceleration · Workflows

    1 works · 6,271 downloads

    ▍ 工具 · Flux.2 Klein 9B · DBKlein9b🟦 V2.0 BFS

    Open → Civitai · DBKlein9b🟦 V2.0 BFS

    Klein 9B 上的面交换Workflows:Dark Beast V2.0 BFS。
    A face-swap workflow built on Dark Beast KLEIN 9b V2.0 BFS and the fast FLUX.2 Klein 9B.

    2026-03-03 · workflow · 6,271 downloads · 100 likes · 2 images · 28 image likes


    ▍RedCraft / Dark Beast Collection

    1 works · 5,542 downloads

    ▍ 红潮合集 · Flux.2 Klein 9B · DBKlein9b🟦 V2.0 LoRA

    Open → Civitai · DBKlein9b🟦 V2.0 LoRA

    FLUX.2 Klein 9B 换脸 LoRA,替换后仍保住身份、表情与光线。
    Face-swap LoRA for FLUX.2 Klein 9B that keeps identity, expression and lighting intact.

    2026-03-05 · LoRA · 5,542 downloads · 233 likes · 2 images · 61 image likes


    Flux.1 Dev Fine-tune / LoRas

    First used 2024-08-15 · 9 brands · 47 works · 112,421 downloads

    ▍RedCraft

    24 works · 74,063 downloads

    ▍ 红潮 · Flux.1 D · RED-OMNI [Flux.1&DreamO]

    Open → Civitai · RED-OMNI [Flux.1&DreamO]

    合入 DreamO 的 Flux.1 Base model,偏 2.5D 成人向。
    Flux.1 checkpoint merging DreamO's in-Kontext LoRA, aimed at 2.5D NSFW work with ComfyUI nodes bundled.

    2025-05-30 · fine-tune · 6,806 downloads · 320 likes · 16 images · 230 image likes

    ▍ 红潮 · Flux.1 D · RED-Edit (ICEdit)1.1🔥多任务

    Open → Civitai · RED-Edit (ICEdit)1.1🔥多任务

    红潮 IC 编辑模型 1.1 多任务版,基于 Flux.1 D 改图。
    Flux.1 D in-context image editing checkpoint from the RedCraft line, version 1.1, tagged multi-task.

    2025-05-04 · fine-tune · 5,149 downloads · 317 likes · 75 images · 909 image likes

    ▍ 红潮 · Flux.1 D · RED-UNO(In-Context)

    Open → Civitai · RED-UNO(In-Context)

    配 UNO in-context 流程的 Flux.1 微调,主打多主体一致。
    Flux.1 finetune made to run UNO's in-context pipeline for consistent multi-subject generation.

    2025-04-28 · fine-tune · 1,456 downloads · 235 likes · 26 images · 436 image likes

    ▍ 红潮 · Flux.1 D · Reveal5[SFW]ULTRA

    Open → Civitai · Reveal5[SFW]ULTRA

    锁定 NSFW 的 Flux ULTRA Base model,15 步,带高对比摄影触发词。
    SFW-locked Flux ULTRA checkpoint with de-distilled trainable weights and a high-contrast trigger.

    2025-03-31 · fine-tune · 7,534 downloads · 453 likes · 167 images · 1,511 image likes

    ▍ 红潮 · Flux.1 D · RealReveal5

    Open → Civitai · RealReveal5

    Flux.1 写实Base model,做超高清材质,兼容并可训 LoRA,NSFW 解锁。
    De-distilled Flux.1 D realism checkpoint for ultra-HD material, F.1 LoRA compatible, trainable, NSFW unlocked.

    2025-03-21 · fine-tune · 8,194 downloads · 506 likes · 305 images · 2,182 image likes

    ▍ 红潮 · Flux.1 D · RED[创意] Epicus BIG Movie

    Open → Civitai · RED[创意] Epicus BIG Movie

    Flux.1 反蒸馏创意Base model,NSFW 解锁、LoRA 通用,主打随机性。
    De-distilled Flux.1 D creative fine-tune made for variety over polish, NSFW unlocked, any F.1 LoRA welcome.

    2025-03-02 · fine-tune · 453 downloads · 64 likes · 221 images · 874 image likes

    ▍ 红潮 · Flux.1 D · F.1[修图] Fill.NSFW

    Open → Civitai · F.1[修图] Fill.NSFW

    FLUX.1 内补 Fill 模型,解锁 NSFW,主要修女性肢体。
    FLUX.1 Fill inpainting model that unlocks NSFW concepts, mainly for repairing female anatomy and organs.

    2025-02-22 · fine-tune · 5,990 downloads · 304 likes · 20 images · 124 image likes

    ▍ 红潮 · Flux.1 D · 🌹NewReveal[F.1]ULTRA🌹

    Open → Civitai · 🌹NewReveal[F.1]ULTRA🌹

    去蒸馏 Flux.1 D 检查点,NSFW 解锁,二十步出图。
    De-distilled Flux.1 D checkpoint with FLUX Aesthetics enhancement, NSFW unlocked and tuned for 20 steps.

    2025-02-09 · fine-tune · 5,556 downloads · 435 likes · 837 images · 3,237 image likes

    ▍ 红潮 · Flux.1 D · New REVE[A]L 🏛🔞🔥

    Open → Civitai · New REVE[A]L 🏛🔞🔥

    反蒸馏 F.1 Dev Base model,解锁 NSFW,全系 LoRA 兼容。
    De-distilled F.1 Dev checkpoint with NSFW unlocked, full LoRA compatibility and 4K-grade material weights.

    2025-01-23 · fine-tune · 10,112 downloads · 623 likes · 482 images · 3,162 image likes

    ▍ 红潮 · Flux.1 D · RED.2 BF16 (ArtAug)

    Open → Civitai · RED.2 BF16 (ArtAug)

    红潮的 BF16 美学Base model,用 ArtAug 循环调过画质。
    BF16 aesthetic Flux.1 dev checkpoint in the RedCraft line, its quality tuned by the ArtAug loop.

    2025-01-08 · fine-tune · 2,107 downloads · 138 likes · 91 images · 821 image likes

    ▍ 红潮 · Flux.1 D · TURBO Reveal(HOTFix2.1)🔥

    Open → Civitai · TURBO Reveal(HOTFix2.1)🔥

    Reveal2 Turbo 的 Flux Versions,8-10 步出图,人像更广。
    Fast Reveal2 Turbo Flux checkpoint for 8-10 steps, portrait-expanded and distilled with Turbo-Alpha.

    2024-12-25 · fine-tune · 2,132 downloads · 153 likes · 254 images · 940 image likes

    ▍ 红潮 · Flux.1 D · Reveal ULTRA [v3.5]🔥

    Open → Civitai · Reveal ULTRA [v3.5]🔥

    Flux 上的 NSFW ULTRA Base model,动作更复杂、LoRA 更适配。
    Flux NSFW ULTRA checkpoint with richer motion and stronger LoRA support, updated live through v3.5.

    2024-12-17 · fine-tune · 6,163 downloads · 430 likes · 356 images · 1,501 image likes

    ▍ 红潮 · Flux.1 D · Reveal NSFW

    Open → Civitai · Reveal NSFW

    Flux 上的 NSFW Base model,10-20 步蒸馏,成人场景适配强。
    Flux-dev-based NSFW checkpoint distilled for 10-20 steps, tuned for adaptive adult scenes and portraits.

    2024-11-22 · fine-tune · 8,995 downloads · 794 likes · 535 images · 2,213 image likes

    ▍ 红潮 · Flux.1 D · Red.1 Q4 GGUF

    Open → Civitai · Red.1 Q4 GGUF

    Red.1 的 Q4 GGUF 量化版,按 1024 分辨率推理,省显存。
    Q4 GGUF quantised build of the Red.1 style model, tuned for 1024 x 1024 inference and low VRAM.

    2024-11-19 · fine-tune · 944 downloads · 25 likes · 31 images · 113 image likes

    ▍ 红潮 · Flux.1 D · Red.1 BF16

    Open → Civitai · Red.1 BF16

    Red.1 的全精度 BF16 扩散权重,供 ComfyUI 单独加载。
    Full BF16 diffusion weights of the Red.1 style model, merged from two Schnell de-distilled fine-tunes for ComfyUI.

    2024-11-18 · fine-tune · 599 downloads · 38 likes · 132 images · 502 image likes


    ▍ 红潮|RED 室内艺术 Interior Artistry

    Open → Liblib · 红InteriorArtist

    红潮室内艺术检查点,东方灵感的空间效果图,CFG 1、4-8 步出图。
    A Flux.1 interior-design checkpoint from the RedCraft line: CFG 1, 4-8 steps, clean prompt following.

    2025-03-13 · fine-tune · 1,400 downloads · 1,881 likes

    ▍ 红潮|RED_气韵权局_Aura of Authority

    Open → Liblib · Authority Aura

    Flux 室内 LoRA,做有掌控感的场域:墨黑配铜色、方正布局、混凝土加深木。
    Flux.1 D interior LoRA for rooms that read as power: ink black and copper, square layouts, concrete and dark wood.

    2025-03-24 · LoRA

    ▍ 红潮|RED_源木寻味_Original Aroma

    Open → Liblib · Original Aroma

    Flux 室内 LoRA,做温暖的极简:木质柜体配木桌,靠木纹压住单色墙。
    Flux.1 D interior LoRA for warm minimalism: timber cabinets and tables against smooth monochrome walls, grain doing the work.

    2025-03-21 · LoRA

    ▍ 红潮|RED_奢秘永恒_Opus Arcanum

    Open → Liblib · Opus Arcanum

    红潮「奢秘·永恒」:暗调奢华、隐秘氛围的 FLUX.1 室内 LoRA。
    FLUX.1 LoRA for dark, secretive luxury interiors, the Opus Arcanum entry in the 红潮 series.

    2025-03-20 · LoRA

    ▍ 红潮|RED_任性简活_Arbitrary Living

    Open → Liblib · Arbitraryliving

    红潮「任性·简活」:极简北欧风居住空间的 FLUX.1 室内 LoRA。
    FLUX.1 interior LoRA for minimalist Nordic living spaces, from the RedCraft line.

    2025-03-20 · LoRA

    ▍ 红潮|RED_朴墅合院_Oriental Villa

    Open → Liblib · Oriental Villa

    F.1 上的东方合院 LoRA,大地色系、朴素乡村生活。
    A Flux LoRA for oriental rustic courtyard villas: earthen tones and plain rural living.

    2025-03-18 · LoRA

    ▍ 红潮|RED_现代隐居_Modern Hermitage

    Open → Liblib · ModernHermitage

    Flux 室内 LoRA,做现代隐居感:低矮大地色坐具、自然光与手织质感。
    Flux.1 D interior LoRA for a modern hermitage: low earth-toned seating, flooded daylight, hand-woven texture.

    2025-03-17 · LoRA

    ▍ 红潮|RED_源本生活_Natural Source

    Open → Liblib · Natural Source

    Flux 室内 LoRA,做禅意的东方空间:单色底子、一丛绿叶、藤编木椅。
    Flux.1 D interior LoRA for zen-leaning eastern rooms: monochrome shell, green leaves, rattan-backed timber chairs.

    2025-03-16 · LoRA

    ▍ 红潮|RED_杼影玄光_Dynamic Reflection

    Open → Liblib · Dyna-reflection

    Flux 室内 LoRA,做东方高端别墅,丝绢层叠与琉璃金属光影最吃重。
    Flux.1 D interior LoRA for high-end oriental villas: layered silk, glazed inlay and patinated metal in a parametric frame.

    2025-03-15 · LoRA


    ▍Flux.1 LoRA

    8 works · 27,368 downloads

    ▍ 红潮 LoRA · Flux.1 D · vol.5 submission(麦橘超然)🔥

    Open → Civitai · vol.5 submission(麦橘超然)🔥

    LSP 的 Flux.1 Dev 顺从向Versions,配麦橘超然Base model与主观视角。
    Submission-themed Flux.1 Dev cut of LSP, written for majicFlus and POV framing with a bound subject.

    2025-01-06 · LoRA · 7,174 downloads · 592 likes · 73 images · 551 image likes

    ▍ 红潮 LoRA · Flux.1 D · vol.1 barefoot(F1)🔥

    Open → Civitai · vol.1 barefoot(F1)🔥

    NSFW Flux.1 Dev 赤足版 LSP LoRA,姿势重点放在裸足上。
    Barefoot LSP LoRA cut for NSFW Flux.1 Dev checkpoints, biasing explicit poses toward visible bare feet.

    2024-12-31 · LoRA · 5,023 downloads · 425 likes · 62 images · 271 image likes

    ▍ 红潮 LoRA · Flux.1 D · vol.6 SOFT(ON)DIFF(F.1)🔥

    Open → Civitai · vol.6 SOFT(ON)DIFF(F.1)🔥

    Flux 版红潮 LoRA,把 SOFT DIFFUSION 排成时尚杂志封面。
    Flux LoRA rendering SOFT DIFFUSION adult video covers as fashion magazine spreads.

    2024-12-22 · LoRA · 1,869 downloads · 159 likes · 37 images · 245 image likes

    ▍ 红潮 LoRA · Flux.1 D · vol.2 CIVIL-HOT(F.1)🔥

    Open → Civitai · vol.2 CIVIL-HOT(F.1)🔥

    FLUX.1 Dev 的 CIVIL-HOT 无码成人向 LoRA。
    FLUX.1 Dev LoRA for uncensored romantic-action and body-art scenes, CIVIL-HOT theme.

    2024-12-19 · LoRA · 1,629 downloads · 103 likes · 43 images · 447 image likes

    ▍ 红潮 LoRA · Flux.1 D · vol.3 AiPONDO(F.1)

    Open → Civitai · vol.3 AiPONDO(F.1)

    AiPONDO 主题的 FLUX.1 Dev 无码成人向 LoRA。
    AiPONDO-themed FLUX.1 Dev LoRA for uncensored adult scenes, same settings as vol.2.

    2024-12-19 · LoRA · 1,773 downloads · 105 likes · 67 images · 135 image likes

    ▍ 红潮 LoRA · Flux.1 D · vol.2 PRESTAGE

    Open → Civitai · vol.2 PRESTAGE

    Flux 版红潮 LoRA,排标题为 PRESTAGE 的时尚杂志风 AV 封面。
    Flux LoRA building fashion-magazine style JAV covers titled PRESTAGE.

    2024-12-12 · LoRA · 1,480 downloads · 102 likes · 12 images · 65 image likes

    ▍ 红潮 LoRA · Flux.1 D · vol.3 TAPPAN

    Open → Civitai · vol.3 TAPPAN

    Flux 版红潮 LoRA,出标题为 TAPPAN 的 AV 杂志封面。
    Flux LoRA laying out JAV magazine covers titled TAPPAN on the Reveal NSFW base.

    2024-12-12 · LoRA · 1,239 downloads · 72 likes · 54 images · 147 image likes

    ▍ 红潮 LoRA · Flux.1 D · vol.1 AiDEAPOCKET🔥

    Open → Civitai · vol.1 AiDEAPOCKET🔥

    Flux 版红潮 LoRA,排标题为 AIDEA POCKET 的 AV 杂志封面。
    Flux LoRA rendering JAV magazine covers titled AIDEA POCKET in a fashion-magazine layout.

    2024-12-11 · LoRA · 7,181 downloads · 1,182 likes · 94 images · 1,219 image likes


    ▍AiARTiST Chinese Art

    1 works · 3,200 downloads

    ▍ 新世界电商超模 F.1-SuperModel 东方审美

    Open → Liblib

    2024-08-21 · fine-tune · 3,200 downloads · 225 likes


    ▍Others (Liblib)

    5 works · 2,800 downloads

    ▍ REVEAL ULTRA

    Open → Liblib

    2025-02-11 · fine-tune · 250 downloads · 43 likes

    ▍ 探店打卡 F.1-VKOC Checkin_元影虚拟网红

    Open → Liblib · 角色IPv3

    角色 IP 形象的第三版,比 v2 晚一天Published的 FLUX.1 LoRA。
    Third character-IP build of the XIAOYI persona on FLUX.1, dated one day after v2.

    2024-09-27 · LoRA · 0 downloads · 0 likes

    ▍ 数据可视化 F.1-Visualdata 数据报表设计

    Open → Liblib · DataVIS

    把提示词变成数据报表版式的 FLUX.1 LoRA,英文信息密度高。
    FLUX.1 LoRA for data-visualisation report layouts, generating dense dashboards with English labels.

    2024-09-02 · LoRA · 309 downloads · 115 likes

    ▍ 界面设计 F.1-UxDesigner 应用交互设计

    Open → Liblib · UxD

    FLUX.1 上的 App 界面与交互设计 LoRA,出 UI 稿用。
    FLUX.1 LoRA for app UI and interaction design mockups, driven by the APP UI design trigger.

    2024-08-24 · LoRA · 241 downloads · 68 likes

    ▍ 灵感源流加速 F.1-METAF.1ES Hyper

    Open → Liblib

    2024-08-18 · LoRA · 2,000 downloads · 584 likes


    ▍UNIT Base Reworks 会员模型

    4 works · 2,342 downloads

    ▍ 元影内测FLUX(非公开)_极致逼真

    Open → Liblib · 茅台v3.0

    茅台虚拟网红第三版,Flux.1 D Base model,触发词 Moutai。
    Third Moutai virtual-influencer LoRA on Flux.1 D, triggered by the single word Moutai.

    2024-09-07 · LoRA · 0 downloads · 0 likes

    ▍ UNIT-F.1-ArchTechnic-DiT

    Open → Liblib · 首席架构师Hyper8

    Flux.1 D 上的建筑结构 LoRA,走 8 步 Hyper 蒸馏。
    Architecture and structure LoRA on Flux.1 D, running the 8-step Hyper distillation line.

    2024-08-17 · LoRA · 0 downloads · 0 likes

    ▍ UNIT-F.1-MandelaEffect-LoRA

    Open → Liblib · 曼德拉效应

    Flux.1 D 上的「曼德拉效应」LoRA,做错位诡异感。
    Flux.1 D LoRA for Mandela-Effect uncanny imagery, tuned for Liblib's online WebUI.

    2024-08-15 · LoRA · 142 downloads · 68 likes

    ▍ F.1-Dev&Schnell-8步高品质加速版FP8

    Open → Liblib

    2024-08-15 · fine-tune · 2,200 downloads · 121 likes


    ▍RedCraft / Dark Beast LoRa Collection

    1 works · 1,479 downloads

    ▍ 红潮合集 · Flux.1 D · RED1 F1 2024

    Open → Civitai · RED1 F1 2024

    Flux.1 D 补画质 LoRA,找回 GGUF 量化丢掉的细节材质。
    Light Flux.1 D LoRA that restores detail and material texture lost after GGUF quantization.

    2024-11-19 · LoRA · 1,479 downloads · 61 likes · 84 images · 291 image likes


    ▍Tooling · Acceleration · Workflows

    1 works · 628 downloads

    ▍ 工具 · Flux.1 D · STEP1xEDIT-nodes

    Open → Civitai · STEP1xEDIT-nodes

    在 ComfyUI 里跑 Step1X-Edit 图像编辑的节点包与示例Workflows。
    ComfyUI custom nodes and example workflow that run the Step1X-Edit image editor on Flux.1 D.

    2025-05-05 · workflow · 628 downloads · 38 likes · 16 images · 142 image likes


    ▍AiARTiST Concept / Festival

    1 works · 278 downloads

    ▍ 概念 · Flux.1 D · The Year of the Snake

    Open → Civitai · The Year of the Snake

    蛇年新春节庆 LoRA,给 FLUX.1 Dev 画面加中式年味装饰。
    Lunar snake-year festival LoRA for FLUX.1 Dev: red and green ornament, festive motifs.

    2025-01-14 · LoRA · 278 downloads · 25 likes · 43 images · 301 image likes


    ▍AiARTiST Architecture

    2 works · 263 downloads

    ▍ 万物花开 F.1-CITYFlowers_鲜花地标建筑

    Open → Liblib · v1.0

    Flux.1 D 地标花艺 LoRA,把城市建筑做成开花雕塑。
    Flux.1 D LoRA for flower-covered landmarks: city buildings and towers wrapped in blooms.

    2024-09-25 · LoRA · 0 downloads · 0 likes

    ▍ 建筑科技 F.1-ArchTechnic 建筑室内设计

    Open → Liblib · DEVnoT5lib在线

    建筑与室内设计的 FLUX.1 Base model,本版为 DEV 去 T5 在线版。
    FLUX.1 architectural and interior design checkpoint, shipped as a DEV build without T5 for Liblib online use.

    2024-08-18 · fine-tune · 263 downloads · 101 likes


    Flux.1 Schnell Fine-tune / LoRas

    First used 2024-11-24 · 1 brands · 4 works · 7,669 downloads

    ▍RedCraft

    4 works · 7,669 downloads

    ▍ 红潮 · Flux.1 S · RASCH.3 (RUSH·Reveal)🔥

    Open → Civitai · RASCH.3 (RUSH·Reveal)🔥

    Schnell 系精修、放大Base model,给各家出图补细节。
    Schnell-based refiner and detail upscaler for IL, Pony, XL, MJ and SD15 images at 10 steps.

    2025-01-29 · fine-tune · 2,597 downloads · 101 likes · 140 images · 652 image likes

    ▍ 红潮 · Flux.1 S · REFLEX NSFW

    Open → Civitai · REFLEX NSFW

    红潮 REFLEX 成人向版,基于快速 Schnell Base model出图。
    RedCraft REFLEX adult checkpoint on the fast Flux.1 Schnell base, NSFW unlocked for high-volume work.

    2024-12-04 · fine-tune · 3,568 downloads · 146 likes · 142 images · 687 image likes

    ▍ 红潮 · Flux.1 S · RASCH.2

    Open → Civitai · RASCH.2

    Schnell 微调版,4–8 步出图,颗粒更少。
    Flux Schnell finetune re-distilled over De-distilledMixTuned-V1, running at 4–8 steps with less grain.

    2024-11-30 · fine-tune · 902 downloads · 65 likes · 202 images · 1,324 image likes

    ▍ 红潮 · Flux.1 S · RASCH.1 Forge

    Open → Civitai · RASCH.1 Forge

    Forge 向的 Schnell 微调版,锁定 10 步推理换细节。
    Forge-oriented Schnell finetune, pruned without T5, locked to a 10-step inference process.

    2024-11-24 · fine-tune · 602 downloads · 35 likes · 145 images · 2,250 image likes


    Pony Fine-tune / LoRas

    First used 2024-05-13 · 3 brands · 10 works · 41,682 downloads

    ▍RedCraft

    4 works · 22,964 downloads

    ▍ 红潮 · Pony · PONY3Relustion PRO🔥

    Open → Civitai · PONY3Relustion PRO🔥

    红潮 Pony 线 PRO 版,走东亚真实系。
    RedCraft's PRO Pony checkpoint for East Asian realistic work, built on the author's stated 3300W leak dataset.

    2025-03-03 · fine-tune · 3,135 downloads · 465 likes · 186 images · 1,171 image likes

    ▍ 红潮 · Pony · PONY2Relustion(FP32)🔥

    Open → Civitai · PONY2Relustion(FP32)🔥

    红潮 Pony 线的 FP32 版,主打东亚真实系画质。
    FP32 build of the RedCraft Pony line, pitched as a 32-bit East Asian realistic Pony checkpoint.

    2025-01-11 · fine-tune · 7,583 downloads · 528 likes · 541 images · 1,300 image likes

    ▍ 红潮 · Pony · PONY Relustion🔥

    Open → Civitai · PONY Relustion🔥

    PONY Relustion 首发 FP16 版,高画质 Pony 创意Base model。
    FP16 first release of the PONY Relustion series, a Pony-based creative checkpoint with adaptive aspect ratios.

    2024-12-20 · fine-tune · 5,708 downloads · 377 likes · 313 images · 1,820 image likes

    ▍ 红潮 · Pony · Hyper PONY 2024

    Open → Civitai · Hyper PONY 2024

    Hyper-SDXL 加速的 Pony Base model,10 步、CFG 2 即可出稿。
    Hyper-SDXL accelerated Pony checkpoint for fast design renders at roughly 10 steps and CFG 2.

    2024-05-13 · fine-tune · 6,538 downloads · 610 likes · 74 images · 318 image likes


    ▍PONY LoRAs

    5 works · 8,218 downloads

    ▍ 红潮 LoRA · Pony · vol.4 submission(PONY)

    Open → Civitai · vol.4 submission(PONY)

    LSP 的 Pony 顺从向Versions,出被支配一方的角色扮演场景。
    Submission-themed Pony cut of LSP, aimed at role-play scenes with a submissive partner on PONY Relustion.

    2025-01-06 · LoRA · 1,928 downloads · 131 likes · 98 images · 307 image likes

    ▍ 红潮 LoRA · Pony · vol.3 barefoot(PONY)🔥

    Open → Civitai · vol.3 barefoot(PONY)🔥

    Pony 赤足版 LSP LoRA,配红潮 Relustion,动作稳、器官准。
    Pony barefoot cut of the LSP LoRA for PONY Relustion, tuned for stable motion and accurate anatomy.

    2025-01-02 · LoRA · 1,726 downloads · 124 likes · 55 images · 181 image likes

    ▍ 红潮 LoRA · Pony · vol.5 CIVIL-HOT(PONY)

    Open → Civitai · vol.5 CIVIL-HOT(PONY)

    Pony 版红潮 LoRA,生成 CIVIL-HOT 片商的封面与盘套版式。
    Pony LoRA producing CIVIL-HOT style JAV cover art and sleeve layouts.

    2024-12-23 · LoRA · 2,281 downloads · 175 likes · 46 images · 156 image likes

    ▍ 红潮 LoRA · Pony · vol.4 AiPONDO(PONY)🔥

    Open → Civitai · vol.4 AiPONDO(PONY)🔥

    Pony 版红潮 LoRA,用来出 AIPONDO 片商的封面与包装视觉。
    Pony LoRA for JAV-style cover and packaging art built around the AIPONDO label.

    2024-12-20 · LoRA · 1,271 downloads · 76 likes · 20 images · 75 image likes

    ▍ 红潮 LoRA · Pony · vol.5 SILD(PONY)

    Open → Civitai · vol.5 SILD(PONY)

    Pony 版红潮 LoRA,做 SILD 刊名,配 iNiverse Mix 写实Base model。
    Pony LoRA for the SILD label, built on the iNiverse Mix Pony realistic base.

    2024-12-14 · LoRA · 1,012 downloads · 64 likes · 36 images · 106 image likes


    SD 3 Fine-tune / LoRas

    First used 2024-06-16 · 3 brands · 4 works · 1,674 downloads

    ▍AiARTiST Industrial

    1 works · 701 downloads

    ▍ SD3-1GIRL-人像LoRA-MIST-AiARTiST

    Open → Liblib · 1GIRL

    SD3 测试版人像 LoRA,1girl 触发,给 sd3_medium 用。
    Early SD3 portrait LoRA triggered by 1girl, built for sd3_medium tests; still a limited beta.

    2024-06-16 · LoRA · 701 downloads · 34 likes


    ▍Tooling · Acceleration · Workflows

    2 works · 683 downloads

    ▍ SD3-ControlNet整理-InstantX-CADS

    Open → Liblib

    2024-06-20 · ControlNet · 176 downloads · 15 likes

    ▍ SD3-元影制片-METAFILMS3-AiARTiST

    Open → Liblib

    2024-06-17 · workflow · 507 downloads · 54 likes


    ▍AiARTiST Architecture

    1 works · 290 downloads

    ▍ SD3-艺术海报构图设计LoRA-CADS-AiARTiST

    Open → Liblib

    2024-06-16 · LoRA · 290 downloads · 23 likes


    SDXL Fine-tune / LoRas

    First used 2023-12-31 · 9 brands · 77 works · 59,036 downloads

    ▍AiARTiST Industrial

    34 works · 21,791 downloads

    ▍ AiARTiST-沉浸XL-ImmersionXL-LoRA

    Open → Liblib · immersion

    SDXL 上的氛围型 LoRA,给 AIGC 体验类画面加沉浸代入感。
    SDXL LoRA that pushes immersive, you-are-there atmosphere into AIGC experience scenes.

    2024-05-12 · LoRA · 73 downloads · 18 likes

    ▍ AiARTiST-我也学一招-3DCartoon IP XL

    Open → Liblib · martialarts

    SDXL 的 3D 卡通 IP LoRA,为武侠短视频赛把人物做成立体手办感。
    SDXL 3D-cartoon IP LoRA for a wuxia short-video contest, turning characters into toy-like 3D figures.

    2024-04-25 · LoRA · 405 downloads · 49 likes

    ▍ AiARTiST-CybercarXL-未来载具XL

    Open → Liblib · UNIT1

    SDXL 上的未来载具 LoRA,出霓虹纳米朋克风的赛博车。
    SDXL LoRA for futuristic vehicles and cyber cars, tuned for neon nanopunk car renders.

    2024-04-17 · LoRA · 2,400 downloads · 362 likes

    ▍ AiARTiST 中老年女装肖像摄影XL LoRA

    Open → Liblib · NvZhuang

    拍中老年女装的 SDXL LoRA,出奢华服装与影棚光的商业肖像。
    An SDXL LoRA for mature women's fashion portraits — luxury outfits, studio light, editorial framing.

    2024-04-15 · LoRA · 214 downloads · 46 likes

    ▍ AiARTiST-AI科学家XL-A.I.Scientist

    Open → Liblib · UNIT1

    SDXL 上的Character LoRA,把人物做成体内透出电路与机械的 AI 科学家。
    SDXL LoRA that dresses a portrait subject as an AI scientist, circuitry and mechanics showing through a translucent body.

    2024-04-15 · LoRA · 1,400 downloads · 189 likes

    ▍ AiARTiST 餐刀设计及场景KV XL LoRA

    Open → Liblib · Diningknife

    餐刀设计与场景 KV 的 SDXL LoRA,配 CADS2.0 商用Base model。
    Dining-knife LoRA for cutlery design and its scene key visual, paired with the CADS2.0 commercial base.

    2024-04-01 · LoRA · 104 downloads · 42 likes

    ▍ AiARTiST 简约实景自定义风格展台 XL LoRA

    Open → Liblib · Productstage

    SDXL 产品展台 LoRA,不锁色调与道具,配 CADS2.0 商业Base model出图。
    Loose SDXL LoRA for product exhibition stages: no fixed palette or props, made for the CADS2.0 commercial base.

    2024-04-01 · LoRA · 533 downloads · 226 likes

    ▍ AiARTiST 香水化妆品场景KV XL LoRA

    Open → Liblib · BottleScene

    香水化妆品瓶身 KV 的 SDXL LoRA,配 CADS2.0 商用Base model。
    Perfume and cosmetics bottle LoRA for scene key visuals on the CADS2.0 base, framed for 1920x1080 and up.

    2024-04-01 · LoRA · 593 downloads · 243 likes

    ▍ AiARTiST 国学大师肖像摄影XL LoRA

    Open → Liblib · ChineseMaster

    拍国学与传统文化大师肖像的 SDXL LoRA,出长者学者的端正面孔。
    An SDXL LoRA for portraits of Chinese traditional-culture masters — older scholars in dignified settings.

    2024-04-01 · LoRA · 321 downloads · 71 likes

    ▍ AiARTiST 分体式水龙头设计效果图 LoRA XLVersions

    Open → Liblib · Faucet

    出分体式水龙头设计效果图的 SDXL LoRA,常配合涂鸦线稿控制造型。
    An SDXL LoRA rendering split-type faucet designs, usually driven by a scribble ControlNet sketch.

    2024-03-23 · LoRA · 71 downloads · 16 likes

    ▍ AiARTiST 木质猫塔参考图设计器 LoRA XLVersions

    Open → Liblib · CatTower

    AiARTiST 的 SDXL LoRA,把木质猫爬架出成可参考的产品图。
    AiARTiST SDXL LoRA that renders wooden cat towers as pet-furniture product reference.

    2024-03-23 · LoRA · 52 downloads · 12 likes

    ▍ AiARTiST 母婴电商场景素材 LoRA XLVersions

    Open → Liblib · Mom&baby

    AiARTiST 的 SDXL LoRA,做母婴电商场景画面,情绪偏温暖。
    AiARTiST SDXL LoRA for mother-and-baby e-commerce scenes in a warm, tender mood.

    2024-03-23 · LoRA · 123 downloads · 46 likes

    ▍ AiARTiST XL 3D黏土世界场景 电商展台 LoRA

    Open → Liblib · 3DCartoon

    SDXL 的 3D 黏土电商展台 LoRA,产品场景配微距光。
    SDXL LoRA for clay-style e-commerce stages: product scenes, plinths and macro light, no people.

    2024-03-22 · LoRA · 261 downloads · 60 likes

    ▍ AiARTiST 饰品耳钉模特穿戴效果图 LoRA XLVersions

    Open → Liblib · Earstud

    耳钉佩戴效果 LoRA,把花朵耳钉戴到真人身上,出街拍风效果图。
    Ear-stud LoRA that puts flower-shaped jewellery on a live model in a street scene, as a wearing mock-up.

    2024-03-22 · LoRA · 137 downloads · 37 likes

    ▍ AiARTiST 绝尘 Ancient general XL

    Open → Liblib · 绝尘

    古代将军Character LoRA,画披甲骑马夜战,可挂任意 SDXL Base model。
    Character LoRA for armoured ancient Chinese generals on horseback, built to sit on any SDXL checkpoint.

    2024-03-21 · LoRA · 970 downloads · 119 likes

    ▍ AiARTiST 养生壶外观辅助设计 LoRA XLVersions

    Open → Liblib · Health kettle

    做养生壶外观方案的 SDXL LoRA,输出可直接提案的产品效果图。
    An SDXL LoRA sketching electric health-kettle exteriors for design boards and kitchen renders.

    2024-03-21 · LoRA · 55 downloads · 10 likes

    ▍ AiARTiST 电动按摩椅辅助设计 LoRA XLVersions

    Open → Liblib · Massage

    AiARTiST 的按摩椅外观设计 LoRA,默认把椅子放进室内外场景出概念图。
    AiARTiST SDXL LoRA for massaging-chair industrial design, placing chair concepts in real room scenes.

    2024-03-21 · LoRA · 61 downloads · 14 likes

    ▍ AiARTiST 新鲜食材电商摆台素材 LoRA XLVersions

    Open → Liblib · Freshfood

    AiARTiST 的 SDXL LoRA,专做生鲜食材的电商摆台画面。
    AiARTiST SDXL LoRA that builds fresh-food table setups for e-commerce listings.

    2024-03-20 · LoRA · 727 downloads · 206 likes

    ▍ AiARTiST 中医药材摆台生成器 LoRA XLVersions

    Open → Liblib · Herbs

    把中药材摆成桌面陈列的 SDXL LoRA,适合做商业广告视觉。
    An SDXL LoRA arranging Chinese herbs as styled tabletop compositions for commercial visuals.

    2024-03-20 · LoRA · 428 downloads · 131 likes

    ▍ AiARTiST 电竞座椅辅助设计 LoRA XLVersions

    Open → Liblib · Esports

    AiARTiST 的电竞椅设计 LoRA,椅背进霓虹房间,出带反射的概念图。
    AiARTiST SDXL LoRA for gaming-chair design, rendering chair backs in neon-lit rooms with reflections.

    2024-03-20 · LoRA · 72 downloads · 13 likes

    ▍ AiARTiST 简约花灯辅助设计 LoRA XLVersions

    Open → Liblib · Fancy

    专画花灯吊灯的 SDXL LoRA,锁定暗场室内,配 CADS Base model使用。
    Narrow SDXL LoRA for chandeliers and pendant lights, locked to dark indoor scenes on the CADS base.

    2024-03-19 · LoRA · 72 downloads · 17 likes

    ▍ AiARTiST 柜机冰箱广告辅助 LoRA XLVersions

    Open → Liblib · Freezer

    AiARTiST 的 SDXL 立式冷柜 LoRA,做闭门陈列与夏季室内广告。
    AiARTiST SDXL LoRA for upright commercial freezer ads: closed doors, summer rooms.

    2024-03-19 · LoRA · 55 downloads · 21 likes

    ▍ AiARTiST 旅行箱设计参考 LoRA XLVersions

    Open → Liblib · Travel case

    AiARTiST 的 SDXL 旅行箱 LoRA,做外观设计参考与户外广告图。
    AiARTiST SDXL LoRA for trolley-case design reference and outdoor luggage ad shots.

    2024-03-18 · LoRA · 50 downloads · 14 likes

    ▍ AiARTiST 电烤盘设计参考 LoRA XLVersions

    Open → Liblib · BakingTray

    AiARTiST 的电烤盘设计 LoRA,烤盘连着食材一起进厨房场景。
    AiARTiST SDXL LoRA for electric griddle design, putting tray concepts with food props into kitchens.

    2024-03-17 · LoRA · 62 downloads · 17 likes

    ▍ AiARTiST 电炒锅设计参考 LoRA XLVersions

    Open → Liblib · Cooker

    AiARTiST 的电炒锅设计 LoRA,把锅摆进厨房出图录级渲染。
    AiARTiST SDXL LoRA for electric wok design, rendering the cooker as a catalogue-style kitchen scene.

    2024-03-17 · LoRA · 55 downloads · 17 likes

    ▍ AiARTiST 简约今日穿搭 OOTD LoRA XLVersions

    Open → Liblib · OOTD

    AiARTiST 的简约穿搭 LoRA,单人全身加素底背景的每日穿搭图。
    AiARTiST SDXL LoRA for simple outfit-of-the-day shots: one full-body figure on a plain backdrop.

    2024-03-16 · LoRA · 230 downloads · 76 likes

    ▍ AiARTiST 美食调色板 Gastro-Art SDXL

    Open → Liblib · Gastro-Art

    美食 LoRA,把食材当油画颜料抹,出带绘画感的餐饮静物。
    Food LoRA that treats ingredients as oil paint, for painterly culinary still life on SDXL.

    2024-03-16 · LoRA · 658 downloads · 104 likes

    ▍ AiARTiST 直饮净水器设计 LoRA XLVersions

    Open → Liblib · Water Purifier

    AiARTiST 的直饮净水器设计 LoRA,机身带反射,落在厨房场景里出图。
    AiARTiST SDXL LoRA for direct-drinking water purifier design, rendering reflective units in kitchens.

    2024-03-15 · LoRA · 67 downloads · 19 likes

    ▍ AiARTiST 真空吸尘器设计 LoRA XLVersions

    Open → Liblib · BODY

    AiARTiST 的真空吸尘器 LoRA,这一版只出主机本体,不负责整机。
    AiARTiST SDXL LoRA for vacuum-cleaner design, covering the main unit rather than the whole machine.

    2024-03-15 · LoRA · 60 downloads · 21 likes

    ▍ AiARTiST 电热水壶设计 LoRA XLVersions

    Open → Liblib · v20240315

    AiARTiST 的电热水壶设计 LoRA,出干净的工业设计渲染,带反光。
    AiARTiST SDXL LoRA for electric kettle design, producing clean reflective industrial-design product shots.

    2024-03-15 · LoRA · 42 downloads · 10 likes

    ▍ AiARTiST 失落的宇航员 LoRA XLVersions

    Open → Liblib · v20240315

    AiARTiST 的 SDXL LoRA,骷髅化宇航员漂在星云汇成的宇宙深渊。
    AiARTiST SDXL LoRA that sets a skull-headed astronaut adrift in a nebula ocean.

    2024-03-14 · LoRA · 237 downloads · 52 likes

    ▍ AiARTiST 汉堡包 hamburger LoRA XL

    Open → Liblib · v20240315

    AiARTiST 的汉堡 LoRA:出菜单和广告里那种看着就想吃的产品图。
    AiARTiST SDXL LoRA that renders appetising hamburger product shots for menus and food ads.

    2024-03-14 · LoRA · 109 downloads · 33 likes

    ▍ AiARTiST 粗砺质感咖啡因风 LoRA XLVersions

    Open → Liblib · v20240220

    粗砺质感摄影风 SDXL LoRA,用整句自然语言出图,CADS 第二弹。
    Gritty film-texture photography LoRA for SDXL, driven by full natural-language scene sentences, second in the CADS line.

    2024-02-20 · LoRA · 594 downloads · 121 likes

    ▍ AiARTiST 专用ACG Playground SDXL

    Open → Liblib · V6-Hyper

    加速版 SDXL ACG Base model,画面更亮更工整,低显存也能快速出高清图。
    An SDXL ACG checkpoint with Hyper-SD 8-step baked in for brighter, faster hi-res anime art.

    2024-01-15 · fine-tune · 10,500 downloads · 1,229 likes


    ▍AiARTiST Architecture

    7 works · 11,090 downloads

    ▍ CADS-城市发展新篇章-Urban-development

    Open → Liblib · City

    城市发展主题海报素材 LoRA,City_Developments 触发。
    SDXL poster-material LoRA on urban growth and city development, triggered by City_Developments.

    2024-06-14 · LoRA · 202 downloads · 52 likes

    ▍ 建筑室内 · SDXL 1.0 · WhiteWall_Villa

    Open → Civitai · WhiteWall_Villa

    白墙别墅外观 LoRA,挂在 Hyper-RAY XL 建筑Base model线上。
    White-villa exterior LoRA for the Hyper-RAY XL architectural base line.

    2024-05-08 · LoRA · 1,477 downloads · 112 likes · 6 images · 19 image likes

    ▍ Hyper-RAY XL 建筑室内外家居装饰 LoRA 超市

    Open → Liblib · Furniture

    RAY XL 上的家具 LoRA,出模块沙发这类现代家具产品图。
    Furniture product-shot LoRA for modern sofas on the CADS2.0 Hyper-RAY XL interior render base.

    2024-05-05 · LoRA · 1,100 downloads · 88 likes

    ▍ Hyper-RAY XL 建筑室内外设计风格 LoRA 超市

    Open → Liblib · Light_Mansion

    RAY XL 室内渲染的绿意通透宅邸 LoRA,木色加自然光。
    Green-toned, light-filled mansion interior LoRA for the CADS2.0 Hyper-RAY XL render system.

    2024-05-03 · LoRA · 3,100 downloads · 224 likes

    ▍ 建筑室内 · SDXL 1.0 · HyperSDXL⚡

    Open → Civitai · HyperSDXL⚡

    内置 Hyper-SDXL 加速的 SDXL 室内设计大模型,秒级出效果图。
    SDXL interior-design checkpoint with Hyper-SDXL baked in for multi-second renders.

    2024-05-01 · fine-tune · 4,062 downloads · 351 likes · 120 images · 381 image likes

    ▍ CADS 室内装修 Refiner XL 重绘1.5材质光影

    Open → Liblib · REFINER

    SDXL 的室内装修精修 LoRA,放大重绘时补材质光影,保住结构。
    SDXL refiner LoRA that re-lights and re-textures interior renders at 1.5× while keeping room structure.

    2024-04-18 · LoRA · 1,100 downloads · 273 likes

    ▍ AiARTiST 奇异建筑设计师 LoRA XLVersions

    Open → Liblib · v20240217

    AiARTiST 的 SDXL 概念建筑 LoRA,风格词定版式,自行加元素。
    AiARTiST SDXL LoRA for concept architecture: one style trigger sets the layout.

    2024-02-17 · LoRA · 49 downloads · 77 likes


    ▍AiARTiST Chinese Art

    13 works · 7,813 downloads

    ▍ AiARTiST-武林XL-SwordmanXL-LoRA

    Open → Liblib

    2024-04-21 · LoRA · 407 downloads · 140 likes

    ▍ AiARTiST 儿童古装肖像摄影XL LoRA

    Open → Liblib · AncientCostume

    拍儿童古装国风肖像的 SDXL LoRA,出汉服造型、道具与雪景。
    An SDXL LoRA for Chinese ancient-costume children's portraits — hanfu styling, props, wintry scenery.

    2024-04-01 · LoRA · 829 downloads · 173 likes

    ▍ AiARTiST 中国刺绣图样生成XL LoRA

    Open → Liblib · PlanarView

    生成中国刺绣图样的 SDXL LoRA,出圆形纹样与亭台,适合图案设计。
    An SDXL LoRA making Chinese embroidery patterns — circular motifs, pavilions, stitched texture — for surface design.

    2024-03-29 · LoRA · 179 downloads · 71 likes

    ▍ AiARTiST 赛博国潮 XL CCC LoRA

    Open → Liblib · CCC

    赛博国潮 SDXL LoRA,给国风角色与街景压上霓虹、雨夜和科技细节。
    Cyber-China-chic SDXL LoRA, layering neon, rain and tech detail onto Chinese-styled characters and streets.

    2024-03-29 · LoRA · 215 downloads · 50 likes

    ▍ AiARTiST 植物大爱僵尸 PFZ LoRA XL

    Open → Liblib · PFZ

    AiARTiST 的 SDXL LoRA,把植物打僵尸的角色做成电影级游戏美术。
    AiARTiST SDXL LoRA that restages plant-versus-zombie game characters as cinematic game art.

    2024-03-28 · LoRA · 308 downloads · 41 likes

    ▍ AiARTiST 2142废土 Wasteland LoRA

    Open → Liblib · 2142

    SDXL 废土风 LoRA,做废墟场景里的电影感角色与动作场面。
    SDXL wasteland LoRA for cinematic ruined-scene characters and action shots, one trigger word.

    2024-03-27 · LoRA · 711 downloads · 88 likes

    ▍ 国风 · SDXL 1.0 · AiARTiST 绝尘 Ancient Chines

    Open → Civitai · v1.0

    中国古代武将 SDXL LoRA,画铠甲、长兵器与战场人像。
    SDXL LoRA for ancient Chinese generals: armour, polearms and battlefield portraiture.

    2024-03-24 · LoRA · 679 downloads · 74 likes · 54 images · 48 image likes

    ▍ AiARTiST 如是我闻 Tathagata LoRAXL

    Open → Liblib · Tathagata

    AiARTiST 的 SDXL 佛教题材 LoRA,如来面容配悬浮花瓣与霓虹。
    AiARTiST SDXL LoRA for Chinese Buddhist imagery: Tathagata faces, petals, smoke, neon.

    2024-03-23 · LoRA · 374 downloads · 61 likes

    ▍ 国风 · SDXL 1.0 · AiARTiST 如是我闻 Tathagata

    Open → Civitai · v1.0

    国风佛教题材 SDXL LoRA,加造像、寺院与经卷氛围。
    SDXL LoRA for Chinese Buddhist deity, temple and sutra scenes on any SDXL base.

    2024-03-21 · LoRA · 380 downloads · 43 likes · 38 images · 51 image likes

    ▍ 国风 · SDXL 1.0 · AiARTiST Golden east 东方黑神话

    Open → Civitai · v20240101

    东方神话插画 SDXL LoRA,把人像重绘成神话手绘绘卷。
    Multi-style Chinese-fantasy LoRA that repaints portraits as mythic hand-drawn illustration.

    2024-03-12 · LoRA · 843 downloads · 97 likes · 60 images · 74 image likes

    ▍ 国风 · SDXL 1.0 · AiARTiST Fully Bloomed 梦想绽

    Open → Civitai · v20240310

    SDXL 风格化艺术摄影 LoRA,挂在写实Base model上改造人像观感。
    SDXL LoRA giving portraits a stylized art-photography look on realistic checkpoints.

    2024-03-09 · LoRA · 197 downloads · 28 likes · 43 images · 101 image likes

    ▍ AiARTiST-IP Design 吉哆GEEDOO XL

    Open → Liblib

    2024-03-02 · LoRA · 91 downloads · 51 likes

    ▍ miracle-东方黑神话 国风插画多态LORA模型SDXL

    Open → Liblib · v20240101

    国风插画多态 LoRA,加 Goldeneast 出手绘,不加走写实。
    Multi-look Chinese-fantasy SDXL LoRA: hand-drawn ink with Goldeneast, or realistic with no trigger.

    2023-12-31 · LoRA · 2,600 downloads · 596 likes


    ▍NSFW

    1 works · 6,217 downloads

    ▍ 红潮 · SDXL 1.0 · Relustion XL [FP8]

    Open → Civitai · Relustion XL [FP8]

    基于 CADS3.0 XL 的 SDXL 写实 NSFW Base model,兼做精修。
    SDXL realistic NSFW base built on CADS3.0 XL, mainly used as a material refiner for Pony and Flux.

    2024-12-07 · fine-tune · 6,217 downloads · 170 likes · 86 images · 967 image likes


    ▍Tooling · Acceleration · Workflows

    9 works · 4,496 downloads

    ▍ CADSWorkflows-爱马时尚风手提包设计参考-lightflow

    Open → Liblib · 会员LightflowWorkflows

    LightFlow Workflows,做「爱马时尚风」手提包配色参考。
    LightFlow workflow for handbag colourway reference in the 爱马 fashion register, on CADS2.0XL/UNIT1 Hyper.

    2024-06-11 · workflow · 77 downloads · 25 likes

    ▍ CADSWorkflows-发泡EVA软拖鞋设计参考-lightflow

    Open → Liblib · 会员LightflowWorkflows

    发泡 EVA 软拖鞋设计参考的 LightFlow Workflows,参数已预置。
    LightFlow workflow for EVA foam slipper concept design, with Hyper-era parameters and control maps preloaded.

    2024-06-03 · workflow · 27 downloads · 5 likes

    ▍ CADSWorkflows-彩绘及车衣设计参考-lightflow

    Open → Liblib · 会员LightflowWorkflows

    车衣彩绘设计参考的 LightFlow Workflows,认两个 Hyper Base model。
    LightFlow workflow for car livery and paint-wrap design, wired for CADS2.0 XL and ACG 6 Hyper bases.

    2024-06-02 · workflow · 109 downloads · 22 likes

    ▍ CNWorkflows-图标LOGO设计参考ACG6-lightflow

    Open → Liblib · 粉丝专享lightflow

    粉丝专享的图标与 LOGO Workflows,ACG6 Base model,不用挂 LoRA。
    Fan-tier LightFlow workflow for app icon and logo grids on ACG Playground SDXL V6, no LoRA needed.

    2024-06-02 · workflow · 328 downloads · 58 likes

    ▍ CADSWorkflows-爆款老爹鞋设计参考-lightflow

    Open → Liblib · 会员LightflowWorkflows

    CADS2.0XL 上的 LightFlow Workflows,做老爹鞋鞋型参考。
    LightFlow workflow for chunky dad-shoe design reference on the CADS2.0XL design system.

    2024-06-01 · workflow · 34 downloads · 8 likes

    ▍ CNWorkflows-618电商购物节场景ACG6-lightflow

    Open → Liblib · lightflowWorkflows

    618 购物节电商场景的 ControlNet Workflows,跑 ACG6 Base model。
    LightFlow ControlNet workflow for 618 shopping-festival e-commerce scenes on ACG Playground SDXL V6.

    2024-06-01 · workflow · 177 downloads · 22 likes

    ▍ CNWorkflows-61儿童节卡通场景ACG6-lightflow

    Open → Liblib · lightflowWorkflows

    六一儿童节卡通场景Workflows,ACG6 Base model加配套 LoRA。
    LightFlow workflow for Children's Day cartoon scenes on ACG Playground SDXL V6 with a paired LoRA.

    2024-05-30 · workflow · 212 downloads · 30 likes

    ▍ CNWorkflows-端午节划龙舟微缩景观ACG6-lightflow

    Open → Liblib · lightflowWorkflows

    端午划龙舟微缩景观的 LightFlow Workflows,跑 ACG6 Base model。
    LightFlow workflow for a Dragon Boat Festival miniature diorama on ACG Playground SDXL V6.

    2024-05-29 · workflow · 232 downloads · 27 likes

    ▍ UNIT3XL-基础单元-Workflows兼容版-AiARTiST

    Open → Liblib

    2024-04-10 · fine-tune · 3,300 downloads · 542 likes


    ▍Dark Beast

    1 works · 3,920 downloads

    ▍ 黑兽 · SDXL 1.0 · HyperBeast🟩SDXXL

    Open → Civitai · HyperBeast🟩SDXXL

    黑兽系列的 SDXL 1.0 分支,主打业余随手拍的写实感。
    SDXL 1.0 entry in the Dark Beast line, built for candid amateur photo realism with optional triggers.

    2026-02-22 · fine-tune · 3,920 downloads · 117 likes · 22 images · 209 image likes


    ▍UNIT Base Reworks 会员模型

    9 works · 3,106 downloads

    ▍ Universal-SD3 魔改新策略利用UNET替换DiT

    Open → Liblib · UNET-SD3

    实验性 SDXL 大模型,借 SD3 的 CLIP/VAE 提画质。
    Experimental SDXL checkpoint that borrows SD3's CLIP and VAE for higher image quality.

    2024-08-15 · fine-tune · 247 downloads · 8 likes

    ▍ F.1模拟器-UNIT3SDXL可用-Simulator

    Open → Liblib · UNIT3Enhancer

    给 UNIT3 加 FLUX 那种对比与锐化的 SDXL LoRA,专攻特写。
    SDXL LoRA lending UNIT3 FLUX.1 Dev contrast and sharpening for photo, portrait and product close-ups.

    2024-08-11 · LoRA · 244 downloads · 26 likes

    ▍ V3模拟器-SDXL用V3-Simulator-LoRA

    Open → Liblib · XLSD3

    给任意 SDXL 模型加 SD3 式对比度与锐度,适合肖像、产品特写。
    LoRA that lends any SDXL model SD3-style contrast and sharpness, aimed at close-ups.

    2024-06-14 · LoRA · 283 downloads · 58 likes

    ▍ UNIT2XL基础单元2-图形设计与IP形象AiARTiST

    Open → Liblib · UNIT2HD

    SDXL 大模型,做 2D/3D 图形设计与卡通 IP 形象,内置加速器。
    SDXL checkpoint for 2D/3D graphic design and cartoon IP characters, with a built-in Hyper accelerator.

    2024-06-08 · fine-tune · 707 downloads · 177 likes

    ▍ XL生鲜电商3DKV视觉设计-CADS2HYPER&UNIT

    Open → Liblib · 会员专属

    SDXL 上的生鲜电商 3D 场景 KV LoRA,配 ACG V6。
    SDXL LoRA for fresh-produce e-commerce 3D scenes, wired for ACG V6 with CADS2/UNIT1 presets.

    2024-05-28 · LoRA · 241 downloads · 82 likes

    ▍ XL可爱3D动物KV视觉设计-CADS2HYPER&UNIT

    Open → Liblib · 开放Downloads

    SDXL 上的可爱 3D 动物 KV LoRA,配 ACG V6 出图。
    SDXL LoRA for cute 3D animal key visuals, tuned for ACG Playground V6 at Hyper speed.

    2024-05-28 · LoRA · 25 downloads · 44 likes

    ▍ XL电商3D数字人场景KV-CADS2 HYPER&UNIT

    Open → Liblib · 会员专属

    电商 3D 数字人场景 KV LoRA,触发词 3d_posterkv。
    SDXL LoRA for 3D digital-human e-commerce scenes, triggered by setting_poster and 3d_posterkv.

    2024-05-26 · LoRA · 308 downloads · 85 likes

    ▍ XL电商3D场景KV视觉设计-CADS2HYPER&UNIT

    Open → Liblib · 会员专享

    电商 3D 场景主视觉 LoRA,触发词 3D_Playground。
    SDXL LoRA for 3D e-commerce key visuals, triggered by 3D_Playground and run on CADS2.0 Extra Large.

    2024-05-20 · LoRA · 115 downloads · 379 likes

    ▍ XL咖啡奶茶饮品KV视觉设计-CADS2HYPER&UNIT

    Open → Liblib · 粉丝专享

    SDXL 上的咖啡奶茶饮品 KV LoRA,触发词 cup milk tea。
    SDXL LoRA for coffee and milk-tea drink key visuals, triggered with cup milk tea.

    2024-05-18 · LoRA · 936 downloads · 102 likes


    ▍MIRACLE Imaging

    1 works · 494 downloads

    ▍ MIRACLE-造梦空间 大师胶片电影风格模拟

    Open → Liblib

    2023-12-31 · LoRA · 494 downloads · 181 likes

    ▍ Miracle-蜜雪冰城 雪王 MIXUE LoRA XL

    Open → Liblib

    2024-02-28 · LoRA · 3 downloads · 13 likes

    ▍ AiARTiS-大码青春 Oversized Younger

    Open → Liblib

    2024-02-25 · LoRA · 106 downloads · 62 likes


    SD 1.5 Fine-tune / LoRas

    First used 2023-12-09 · 6 brands · 21 works · 36,094 downloads

    ▍AiARTiST Industrial

    4 works · 12,269 downloads

    ▍ AiARTiST-珠宝金饰 黄金营销设计辅助LoRA 1.5

    Open → Liblib · v20240301

    SD1.5 的珠宝金饰营销版式 LoRA,Jewellerygold 触发。
    SD 1.5 LoRA for gold jewellery store marketing layouts, triggered by Jewellerygold plus style words.

    2024-01-17 · LoRA · 165 downloads · 167 likes

    ▍ AiARTiST-舌尖中国 美食素材LORA模型1.5Versions

    Open → Liblib · v20240112

    SD1.5 的中华美食素材 LoRA,BitingChina 触发。
    SD 1.5 food-material LoRA for Chinese cuisine posters, launched by the BitingChina trigger.

    2024-01-12 · LoRA · 2,300 downloads · 357 likes

    ▍ AiARTiST-直播间简单背景生成器LORA模型 1.5

    Open → Liblib

    2024-01-08 · LoRA · 204 downloads · 175 likes

    ▍ AiARTiST 艺术海报构图设计大模型 1.5Versions

    Open → Liblib · SD15

    自训 SD 1.5 整模,把构图辅助 LoRA 融进去,主攻海报版式预演。
    Self-trained SD 1.5 checkpoint with a composition LoRA merged in, aimed at poster and layout drafts.

    2024-01-05 · fine-tune · 9,600 downloads · 1,594 likes


    ▍NSFW

    1 works · 10,841 downloads

    ▍ 红潮 · SD 1.5 · Relustion 1.5 [FP32]

    Open → Civitai · Relustion 1.5 [FP32]

    复古路线的 SD 1.5 高清重铸版,内置 VAE,NSFW 解锁。
    Retro-minded SD 1.5 HD rebuild from RETRO SD material and HD4K, built-in VAE, NSFW unlocked.

    2025-01-18 · fine-tune · 10,841 downloads · 297 likes · 99 images · 430 image likes


    ▍AiARTiST Chinese Art

    4 works · 7,231 downloads

    ▍ AiARTiST 梦想绽放 LoRA 1.5Versions

    Open → Liblib · v20240313

    AiARTiST 的 SD 1.5 画意人像 LoRA,给写实Base model添花开感。
    AiARTiST SD 1.5 LoRA for bloom-themed stylized portrait photography over realistic models.

    2024-03-12 · LoRA · 193 downloads · 36 likes

    ▍ AiARTiST-古装剧场景大模型SD1.5版

    Open → Liblib

    2024-03-05 · fine-tune · 338 downloads · 60 likes

    ▍ MIRACLE-书法字体精炼器 LORA 1.5 Versions

    Open → Liblib · v20240102

    书法字体 LoRA,ControlNet 控笔势,出海报与 LOGO 字形。
    SD1.5 calligraphy-lettering LoRA: ControlNet depth steers the brush stroke for poster and logo type.

    2024-01-02 · LoRA · 4,900 downloads · 1,113 likes

    ▍ MIRACLE-诡夜奇谈 国风插画手绘风格 Lora

    Open → Liblib · v20231228

    SD1.5 国风手绘 LoRA,画水墨工笔的志怪插画:妖、龙、麒麟、鲸。
    SD1.5 guofeng hand-drawn LoRA for ink-and-gongbi folklore illustration of beasts and spirits.

    2023-12-28 · LoRA · 1,800 downloads · 393 likes


    ▍MIRACLE Imaging

    10 works · 5,028 downloads

    ▍ MIRACLE · SD 1.5 · 上海照相馆 Portrait Studio

    Open → Civitai · v20240122

    SD 1.5 复古影楼人像风格 LoRA,给写实Base model补拍老照片。
    SD 1.5 style LoRA for vintage Shanghai photo-studio portraits, triggered by SHPortraitStudio.

    2024-03-09 · LoRA · 447 downloads · 37 likes · 32 images · 44 image likes

    ▍ MIRACLE-上海照相馆 LORA模型 1.5 Versions

    Open → Liblib · v20240122

    SD1.5 风格 LoRA,把写实Base model的人像拍成老上海照相馆式肖像。
    SD1.5 style LoRA that shoots portraits like an old Shanghai studio, high-key or low-key.

    2024-01-22 · LoRA · 1,100 downloads · 295 likes

    ▍ MIRACLE-WorldCraft 电影风格模拟(竖屏)

    Open → Liblib · v20231224

    SD1.5 竖屏电影感 LoRA,把画面调成电影剧照的影调,适合手机壁纸。
    Vertical SD1.5 LoRA that grades a render toward a WorldCraft cinematic film look for phone wallpapers.

    2023-12-24 · LoRA · 392 downloads · 34 likes

    ▍ MIRACLE-不惑仔之龙征虎战 复古胶片电影模拟(竖屏)

    Open → Liblib · v20231218

    竖屏 SD1.5 复古胶片 LoRA,出港片味剧照,故意保留胶片底噪。
    Vertical SD1.5 LoRA for retro Hong Kong film stills, keeping the analogue grain on purpose.

    2023-12-18 · LoRA · 253 downloads · 12 likes

    ▍ MIRACLE-Planet 科幻电影风格模拟(竖屏)

    Open → Liblib · v20231216

    SD1.5 竖屏科幻电影调 LoRA,出手机壁纸比例的画面。
    SD1.5 vertical LoRA for sci-fi film colour, tuned for phone-wallpaper portrait crops.

    2023-12-15 · LoRA · 749 downloads · 54 likes

    ▍ MIRACLE-HELL`oBoY 怀旧电影风格模拟(竖屏)

    Open → Liblib · v20231213

    SD1.5 竖屏怀旧电影感 LoRA,出手机壁纸尺寸的胶片色调。
    SD1.5 vertical LoRA simulating nostalgic film-movie colour for phone-wallpaper crops.

    2023-12-13 · LoRA · 154 downloads · 13 likes

    ▍ MIRACLE-ViceCity 胶片电影风格模拟(竖屏)

    Open → Liblib

    2023-12-12 · LoRA · 644 downloads · 58 likes

    ▍ MIRACLE-春光年华 胶片电影风格模拟(竖屏)

    Open → Liblib · v20231211

    竖屏 SD1.5 复古胶片 LoRA,触发词 WKW,出 90 年代港片味剧照。
    Vertical SD1.5 LoRA with a WKW tag, giving renders a 90s nostalgic film-still palette.

    2023-12-11 · LoRA · 408 downloads · 53 likes

    ▍ MIRACLE-SHELL of the GHOST 电影风

    Open → Liblib · v20231210

    SD1.5 竖屏胶片电影调 LoRA,走「SHELL of GHOST」那挂。
    SD1.5 vertical film-grade LoRA in the SHELL of the GHOST register, for wallpaper crops.

    2023-12-09 · LoRA · 268 downloads · 24 likes

    ▍ MIRACLE-70mm IMAX胶片电影风滤镜 改善感光度

    Open → Liblib

    2023-12-09 · LoRA · 613 downloads · 92 likes


    ▍Tooling · Acceleration · Workflows

    1 works · 572 downloads

    ▍ 工具 · SD 1.5 · Boogu + ideogram 4

    Open → Civitai · Boogu + ideogram 4

    Boogu 配 Ideogram 4 的 ComfyUI Workflows,文字控版式。
    ComfyUI workflow combining Boogu-Image-0.1 with Ideogram 4 for dense text and layout control.

    2026-06-20 · workflow · 572 downloads · 34 likes · 29 images · 217 image likes


    ▍UNIT Base Reworks 会员模型

    1 works · 153 downloads

    ▍ F.1模拟器-SD15-Miracle-Simulator

    Open → Liblib

    2024-08-11 · fine-tune · 153 downloads · 37 likes


    Ernie Fine-tune

    First used 2026-04-27 · 1 brands · 1 works · 8,151 downloads

    ▍RedCraft

    1 works · 8,151 downloads

    ▍ 红潮 · Ernie · ERNIE♥RedMIX

    Open → Civitai · ERNIE♥RedMIX

    红潮基于 ERNIE 的混合精度 SFT 微调,放宽限制、拓宽风格。
    ERNIE-based RedCraft fine-tune using mixed-precision SFT to loosen generation limits and widen the style range.

    2026-04-27 · fine-tune · 8,151 downloads · 236 likes · 143 images · 2,190 image likes


    HiDream Fine-tune

    First used 2025-04-28 · 2 brands · 2 works · 6,803 downloads

    ▍RedCraft

    1 works · 6,726 downloads

    ▍ 红潮 · HiDream · 赩梦|REDiDream(NSFW i1)

    Open → Civitai · 赩梦|REDiDream(NSFW i1)

    红潮系列里基于 HiDream 的检查点,标注 NSFW,本版未附参数。
    HiDream-based RedCraft checkpoint tagged NSFW, with no sampling recipe published for this version.

    2025-04-28 · fine-tune · 6,726 downloads · 467 likes · 109 images · 5,633 image likes

    ▍Others (Liblib)

    1 works · 77 downloads

    ▍ 赩梦|REDiDream Pro via HiDreami1

    Open → Liblib

    2025-04-28 · fine-tune · 77 downloads · 9 likes


    Kolors Fine-tune / LoRas

    First used 2024-07-19 · 2 brands · 5 works · 8,104 downloads

    ▍UNIT Base Reworks

    4 works · 7,080 downloads

    ▍ KOLORS元影内测(非公开)_非公开模型

    Open → Liblib · 城市地标v1

    Kolors Base model的城市地标 LoRA,挂在非公开内测Entries下。
    Non-public Kolors LoRA for city-landmark subjects, listed as an internal beta entry.

    2024-09-25 · LoRA · 0 downloads · 0 likes

    ▍ UNIT-可丽(Kolors Girl)女装Base model中英文加速版

    Open → Liblib

    2024-08-08 · fine-tune · 6,300 downloads · 480 likes

    ▍ UNIT-可影(Kolors Film)加速版大电影LoRA

    Open → Liblib

    2024-08-04 · LoRA · 68 downloads · 32 likes

    ▍ UNIT-Kolors(可图)加速版6-8步文生图调优模型

    Open → Liblib · HYPER

    Kolors 加速版,6-8 步在线文生图,中文提示词吃得准。
    Kolors checkpoint tuned for 6-7 step online text-to-image, strong on Chinese prompts.

    2024-07-23 · fine-tune · 712 downloads · 41 likes

    ▍Tooling · Acceleration · Workflows

    1 works · 1,024 downloads

    ▍ 工具 · Kolors · accelerated

    Open → Civitai · accelerated

    Kolors 文生图的 8 步加速版权重,认中文提示词。
    An 8-step accelerated Kolors text-to-image checkpoint that understands Chinese prompts.

    2024-07-19 · fine-tune · 1,024 downloads · 80 likes · 32 images · 150 image likes


    HunyuanDiT Fine-tune / LoRa

    First used 2024-07-28 · 1 brands · 2 works · 200 downloads

    ▍UNIT Base Reworks

    2 works · 200 downloads

    ▍ UNIT-混元(Hunyuan Girl)中文DiT人像Base model

    Open → Liblib

    2024-08-12 · fine-tune · 174 downloads · 22 likes

    ▍ UNIT-混元(Hunyuan)DiT扁平艺术幻想LoRA

    Open → Liblib · DEMO

    混元 DiT 上的扁平幻想插画 LoRA,DEMO 版训练偏低。
    HunyuanDiT flat-illustration fantasy LoRA, a lightly trained DEMO build run on Chinese prompts.

    2024-07-28 · LoRA · 26 downloads · 9 likes


    HunyuanVideo Fine-tune / LoRa

    First used 2024-12-23 · 1 brands · 3 works · 10,555 downloads

    ▍Tooling · Acceleration · Workflows

    3 works · 10,555 downloads

    ▍ 工具 · Hunyuan Video · 首尾及关键帧(kayframes)

    Open → Civitai · 首尾及关键帧(kayframes)

    FramePack 关键帧测试版,可设尾帧、给关键帧分权重、分段写提示词。
    Keyframe test build of the FramePack Hunyuan Video wrapper, with end frame and weighted keyframes.

    2025-04-20 · fine-tune · 632 downloads · 33 likes · 9 images · 110 image likes

    ▍ 工具 · Hunyuan Video · FramePackI2V_fp8_e4m3fn

    Open → Civitai · FramePackI2V_fp8_e4m3fn

    FramePack 图生视频 FP8 权重,接 Hunyuan Video。
    FP8 and bf16 FramePack image-to-video weights for Hunyuan Video, light enough for small GPUs.

    2025-04-19 · fine-tune · 3,873 downloads · 336 likes · 12 images · 960 image likes

    ▍ 工具 · Hunyuan Video · Hunyuan FASTVIDEO 2024

    Open → Civitai · Hunyuan FASTVIDEO 2024

    HunyuanVideo 加速 LoRA:6-10 步出片。
    FastVideo's HunyuanVideo accelerator LoRA, sampling good video in 6-10 steps instead of 50.

    2024-12-23 · LoRA · 6,050 downloads · 221 likes


    Other

    First used 2024-05-16 · 1 brands · 3 works · 4,106 downloads

    ▍Tooling · Acceleration · Workflows

    3 works · 4,106 downloads

    ▍ 工具 · Other · Ideogram4 2026.06

    Open → Civitai · Ideogram4 2026.06

    ideogram4 Workflows,提示词写带边界框和色板的 JSON。
    ideogram4 workflow in ComfyUI taking structured JSON prompts with bounding boxes and hex palettes.

    2026-06-03 · workflow · 1,767 downloads · 82 likes · 24 images · 141 image likes

    ▍ 工具 · Other · ACESTEP workflow🔥

    Open → Civitai · ACESTEP workflow🔥

    ACE-Step 音乐生成的 ComfyUI 节点包与示例Workflows,含音频到音频。
    ComfyUI node pack and sample workflow for ACE-Step music generation, plus audio-to-audio flows.

    2025-05-08 · workflow · 1,371 downloads · 75 likes · 1 images · 40 image likes

    ▍ 工具 · Other · Kolors 2024.1

    Open → Civitai · Kolors 2024.1

    Kolors DiT 文生图Workflows,Hyper 8 步加速加 SDXL 精修。
    Kolors DiT text-to-image workflow with Hyper 8-step acceleration, SDXL refinement and an IPA Plus branch.

    2024-07-15 · workflow · 968 downloads · 48 likes · 45 images · 227 image likes



    做好工具人 服务艺术家

    Description

    Finished uploading REDZimage15_ComfyUI_workflows_example.zip

    19.11 GB - AIO checkpoint
    11.46 GB - none AIO bf16(single transformers)
    5.73 GB - none AIO fp8 (single transformers)
    670.08 MB - Lora

    • You can download the LoRA version later on this page

    (⬆️Pruned Model fp16 (670.08 MB) is LoRA Safetensors⬆️)

    • Single file transformers have been uploaded BF16/FP8 e4m3fn

    ---

    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): https://huggingface.co/datasets/lrzjason/ZImageTurboGen

    ---

    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

    FAQ

    Comments (107)

    wzr905636Dec 8, 2025· 1 reaction
    CivitAI

    BF16是融合文本编码和VAE才19G内容和FP16相同吧?

    AiMetatron
    Author
    Dec 8, 2025· 1 reaction

    FP16是抽出来的lora,强度比完整模型略低一点,好处是可以调整权重跟其他lora混合用。

    推荐下载BF16或者FP8的独立 Transformers 文件,后缀也是 safetensors :Pruned Model bf16 (11.46 GB) 放在 ComfyUI 的 diffusion_models 目录下

    KINGOALDec 8, 2025· 2 reactions
    CivitAI

    Not sure why, I can not use 5G and 11G version on any workflow, it always warn me that lack of clip, such as CLIPTextEncode ERROR: clip input is invalid: None. But the 19G version works very well.

    Btw, I didn't find the workflow.zip you mentioned finishing upload? The workflow in your pictures can not run smaller versions too.

    AiMetatron
    Author
    Dec 8, 2025· 1 reaction

    你需要单独下载 text_encoders 模型,并放置在你使用的推理环境的正确位置(比如 ComfyUI 的 models\text_encoders )

    You need to download the text_decoders model separately and place it in the correct location in the inference environment you are using (such as ComfyUI's models \ text_decoders)

    text_encoders links:

    Z Image - Text Encoder | ZImageTurbo Checkpoint | Civitai

    KINGOALDec 8, 2025

    @AiMetatron Thanks, can use in the comfyui z-image example workflow.

    henserluDec 8, 2025· 1 reaction
    CivitAI

    目前这个ZTurbo蒸馏模型***差点意思,微调LoRA也有很多问题,看来得反蒸馏才行

    AiMetatron
    Author
    Dec 8, 2025· 2 reactions

    现在已经有三个训练环境提供支持了,实现路径也不完全相同,期待CivitAI的训练器支持!

    stardvDec 9, 2025· 3 reactions

    是的,几乎所有的LORA都会降低模型的画质,不用lora会更好生成,用了LORA必须加大步数来修正,而且LORA一般不能超过0.4

    henserluDec 9, 2025

    @stardv 我的测试跟你说得很相同,而且这个模型串联LoRA问题更大,因为每个权重密度都不能超过0.4,太高了画质降低严重,太低效果不佳,头疼的很。另外LoRA对输出文字一致性影响非常大,估计在训练LoRA数据集里还要蒸馏一些模型中已有的良好文字输出图片

    tubelube33Dec 8, 2025· 1 reaction
    CivitAI

    I get None type error for clip text encoder... for all versions.. and used your workflow 0_o

    tubelube33Dec 8, 2025· 1 reaction

    Also i had qwen_3_b_.safetensor as text encoder from the original as text encoder for z image , thank u for the work !

    wakarimasenDec 9, 2025· 2 reactions

    Update ComfyUI bro

    AiMetatron
    Author
    Dec 9, 2025· 1 reaction

    Thz

    wyjzpDec 9, 2025· 2 reactions
    CivitAI

    大佬,我2070显卡8G显存,是不是就下fp8版本的?

    xiaowuzxc442Dec 9, 2025· 2 reactions

    有32g内存的话可以fp16

    AiMetatron
    Author
    Dec 9, 2025· 1 reaction

    对!佬~bf16和FP8都能玩,看心情,速度差不多❤

    wyjzpDec 10, 2025

    @AiMetatron 话说我出了一张图之后,再点运行就一直卡在K采样器那个节点不动了,这是咋回事,前面明明都出了一张,1024*1024的,500+秒才出来的🤣

    wyjzpDec 10, 2025

    @xiaowuzxc442 话说我出了一张图之后,再点运行就一直卡在K采样器那个节点不动了,这是咋回事,前面明明都出了一张,1024*1024的,500+秒才出来的🤣

    obiqeqop737578Dec 9, 2025· 2 reactions
    CivitAI

    您好,我在使用您的工作流+模型时有一大串报错:加载 NextDiT 的 state_dict 时出错参数大小不匹配;请问是什么原因,感谢解惑

    AiMetatron
    Author
    Dec 9, 2025· 1 reaction

    升级一下ComfyUI,如果不是这个问题的话,检查下是不是TE(Clip)或者VAE没选对。

    obiqeqop737578Dec 11, 2025

    谢谢!更新cheatpoint节点后解决了,感谢,您的模型质量非常高

    ImbriumDec 9, 2025· 2 reactions
    CivitAI

    Hi, workflows and examples show REDZ15_DetailDaemonZ_lora_v1.1.safetensors as a LORA. Where can we get this LORA?

    ImbriumDec 10, 2025

    @AiMetatron Thank you! What a great LORA :)

    balbal2Dec 10, 2025· 3 reactions
    CivitAI

    NSFW效果异常的差,还是无法正常生成关键部位,对比原版只是略微有一点点进步,但还是无法正常生成,概率不高

    AiMetatron
    Author
    Dec 10, 2025· 2 reactions

    我一般喜欢配合 LoRAs 来一起使用,效果良好😉:

    [ZIT] Mystic XXX - v1.0 | ZImageTurbo LoRA | Civitai

    ThisStanDec 10, 2025· 1 reaction
    CivitAI

    为什么结合自己的lora效果骤降(我自己的lora只是想改变五官长相)

    AiMetatron
    Author
    Dec 10, 2025· 2 reactions

    很多人,包括我自己在测试利用ZiTurbo版本训练lora的时候会遇到细节丢失的问题。当然,多个lora相互影响也会干涉细节权重,为此我们准备了一个细节守护滑块lora,可以用于更细致的调节画面丰富度:DetailDeamonZ-SliderLoRA For ZImageTurbo & RedZ1.5 - DetailDeamonZ-SliderLoRA | ZImageTurbo LoRA | Civitai

    AiMetatron
    Author
    Dec 10, 2025· 2 reactions

    Many people, including myself, encounter the problem of losing details when testing the use of ZiTurbo version to train Lora. Of course, the mutual influence of multiple loras can also interfere with the weight of details. Therefore, we have prepared a detail guarding slider lora, which can be used to adjust the richness of the image in more detail:DetailDeamonZ-SliderLoRA For ZImageTurbo & RedZ1.5 - DetailDeamonZ-SliderLoRA | ZImageTurbo LoRA | Civitai

    ThisStanDec 11, 2025

    @AiMetatron 我艹。。。。。效果极佳。。。。。我用的是ZIT的AIO版本,换成UNET版本的话,那个600多M的LORA需要叠加这个细节LORA吗

    wyxzddsjj919Dec 10, 2025· 2 reactions
    CivitAI

    大佬,请问下这个LORA 对调度器和采样器有要求吗?推荐强度是0.5~0.75吗,激活LORA的关键词是什么?谢谢

    AiMetatron
    Author
    Dec 10, 2025· 1 reaction

    您好,对调度器和采样没有改变,一切还是跟原来的 ZimageTurbo 版本一样。

    由于LoRA是在AIO模型中抽出的,所以没有特定的提示词,推荐强度要根据其他lora的匹配来测试出最好的结果。

    同时,对于多个lora相互干涉导致的细节丢失等问题,我们发布了一个细节守护器滑块lora:

    DetailDeamonZ-SliderLoRA For ZImageTurbo & RedZ1.5 - DetailDeamonZ-SliderLoRA | ZImageTurbo LoRA | Civitai

    可以用来加强或者减弱画面的细节丰富程度。

    AiMetatron
    Author
    Dec 10, 2025· 1 reaction

    Hi, there have been no changes to the scheduler and sampling, everything remains the same as the original Zimage Turbo version.

    RZ15 LoRA is extracted from the AIO model, there are no specific prompt words, and the recommendation strength needs to be tested for the best results based on matches with other loras.At the same time, for issues such as loss of details caused by interference between multiple loras, we have released a detail guardian slider lora:

    DetailDeamonZ-SliderLoRA For ZImageTurbo & RedZ1.5 - DetailDeamonZ-SliderLoRA | ZImageTurbo LoRA | Civitai

    AiMetatron
    Author
    Dec 10, 2025· 1 reaction

    DetailDeamonZ Slider LoRA can be used to enhance or reduce the richness of details in a picture.

    fox23vang226Dec 10, 2025· 4 reactions
    CivitAI

    I tried red-zimage but I feel like it lost some prompt accuracy. I'm not getting the things I want with the same prompt used with regular Z-image turbo. Plus I get more deformities than usual. I understand these are alpha models and will take many months to a year to tweak, so look forward to the finished product.

    iriscoltDec 10, 2025· 2 reactions

    Ditto, the images have artifacts, weird non-sequiturs, and an obvious “AI-generated” feel that Z Image avoids.

    AiMetatron
    Author
    Dec 10, 2025· 2 reactions

    Thank you so much for the detailed and honest comparison!

    We really appreciate you taking the time to test both models side-by-side.

    You're absolutely right: the current RedZimage v1.5 was intentionally tuned to remove the heavy “internet-aesthetic/influencer” look and push toward greater realism and naturalness. In that process, we did sacrifice some of the polished, instantly-pleasing beauty that Z Image Turbo has out of the box, so right now Turbo still looks more “perfect” .

    That trade-off was deliberate, but we agree the aesthetic side still has room to improve. We are already training the next version at full speed, keeping the realistic foundation while bringing back much stronger artistic appeal and refinement. We’re very confident the upcoming release will close (and hopefully reverse) that gap.

    Your feedback is incredibly valuable and directly helps us prioritize the right things. Thank you again, and we hope the next version will win you over! 🚀

    AiMetatron
    Author
    Dec 10, 2025· 1 reaction

    非常感谢您详细而诚实的比较!

    我们非常感谢您花时间并排测试这两个模型。

    完全正确:当前的 RedZimage v1.5 经过精心调整,去除了沉重的“互联网美学/影响者”外观,并朝着更真实和自然的方向发展。在这个过程中,我们确实牺牲了Z Image Turbo开箱即用的一些抛光、即时愉悦的美感,所以现在Turbo看起来仍然更“完美”。

    这种权衡是经过深思熟虑的,但我们一致认为,美学方面仍有改进的空间。我们已经在全速训练下一个版本,在保持现实基础的同时,带回更强的艺术吸引力和精致感。我们非常有信心即将发布的版本将缩小(并希望扭转)这一差距。

    您的反馈非常有价值,直接帮助我们优先考虑正确的事情。再次感谢您,我们希望下一个版本能赢得您的青睐! 🚀

    iriscoltDec 10, 2025· 7 reactions
    CivitAI

    I ran the same prompts through Z Image Turbo and RedCraft, and honestly the original clearly outperforms RedCraft... its images are much less likely to be identified as AI-generated by a trained eye.

    AiMetatron
    Author
    Dec 10, 2025· 4 reactions

    Thank you so much for the detailed and honest comparison!

    We really appreciate you taking the time to test both models side-by-side.

    You're absolutely right: the current RedZimage v1.5 was intentionally tuned to remove the heavy “internet-aesthetic/influencer” look and push toward greater realism and naturalness. In that process, we did sacrifice some of the polished, instantly-pleasing beauty that Z Image Turbo has out of the box, so right now Turbo still looks more “perfect” .

    That trade-off was deliberate, but we agree the aesthetic side still has room to improve. We are already training the next version at full speed, keeping the realistic foundation while bringing back much stronger artistic appeal and refinement. We’re very confident the upcoming release will close (and hopefully reverse) that gap.

    Your feedback is incredibly valuable and directly helps us prioritize the right things. Thank you again, and we hope the next version will win you over! 🚀

    AiMetatron
    Author
    Dec 10, 2025· 2 reactions

    非常感谢您详细而诚实的比较!

    我们非常感谢您花时间并排测试这两个模型。

    完全正确:当前的 RedZimage v1.5 经过精心调整,去除了沉重的“互联网美学/影响者”外观,并朝着更真实和自然的方向发展。在这个过程中,我们确实牺牲了Z Image Turbo开箱即用的一些抛光、即时愉悦的美感,所以现在Turbo看起来仍然更“完美”。

    这种权衡是经过深思熟虑的,但我们一致认为,美学方面仍有改进的空间。我们已经在全速训练下一个版本,在保持现实基础的同时,带回更强的艺术吸引力和精致感。我们非常有信心即将发布的版本将缩小(并希望扭转)这一差距。

    您的反馈非常有价值,直接帮助我们优先考虑正确的事情。再次感谢您,我们希望下一个版本能赢得您的青睐! 🚀

    BeBoYoDec 10, 2025· 1 reaction
    CivitAI

    請問Krita的Krita AI 裡能用嗎? 如果我要用Pruned Model fp8, 除了下載VAE還要下載什麼?

    alexbird33114Dec 12, 2025· 4 reactions
    CivitAI

    生图质量和细节风格比原版好多了,感谢付出。

    请教fp8版本是不是只有模型,没有vae和clip,我用checkpoints loader加载出的vae和clip都报错,还得再加节点加载vae和clip模型

    AiMetatron
    Author
    Dec 12, 2025

    你好,确实没有上传FP8的AIO版本,不过现在新已经在制作中了,下一个版本再补全。

    alexbird33114Dec 12, 2025

    @AiMetatron 感谢感谢,我发现另一个现象是红潮模型生成的图片亮度要比原版高,尤其是哪怕用强提示词要求生成low-key暗调的画面,结果出来的也还是中间调正常亮度的,不知道是否是我使用不当,使用示范图片的工作流

    hugopnabais659Dec 12, 2025· 6 reactions
    CivitAI

    Very confusing filenames in the downloads!

    qekDec 12, 2025

    It's due to civitai renaming files if a model is a checkpoint or a VAE

    qekDec 12, 2025· 3 reactions
    CivitAI

    It seems it creates better images than the base and others

    4423025858Dec 13, 2025· 2 reactions
    CivitAI

    我必须为作者道歉,前几天我只看到一个600M的AIO模型,感觉到莫名其妙,现在换了个浏览器,我终于看到主模型了

    EasonnLiDec 23, 2025· 1 reaction

    大家不要加入这个作者的飞机群,我进去了后互相致意,分享了我的作品。有两三个人夸了我的作品后,这个作者可能心里不平衡,就群里启用禁言,并把我踢出了群。我对这样心胸狭窄的人感到很震惊,我从头到尾没有想抢他风头的意思,只是发了自己的模型地址,分享一下自己的作品。如果这样不让别人分享,那群有什么意义,都是用来崇拜他一个人的吗?

    4423025858Dec 13, 2025· 2 reactions
    CivitAI

    有没有带diffusionfe的目录的fp8的仓库呢,ai tookit必要要带diffusionfe结构的目录才能训练

    2675664993642Dec 13, 2025· 3 reactions
    CivitAI

    不知道是不是我工作流的问题,我感觉它生成的图片没有z-image原始模型的少女感了

    MrSmith2025Dec 13, 2025· 2 reactions
    CivitAI

    Thanks for the AIO ZIT Model. Works good so far, but it loses the ability to spell words correct. But other than that a good model.

    student6688Dec 14, 2025· 10 reactions
    CivitAI

    请问工作流的链接在哪里?

    AiMetatron
    Author
    Dec 18, 2025· 1 reaction

    右边有一个File菜单,点开里面有文件清单。

    TatsumiKojirouDec 14, 2025· 5 reactions
    CivitAI

    El modelo está bien, pero algunas loras que funcionan bien con el modelo original hacen cosas raras en este

    LYNX233Dec 14, 2025· 2 reactions
    CivitAI

    11.46GB - 无 AIO bf16
    5.73 GB - 无 AIO FP8

    这两种模型是从19G的大模型里面拆分出来的吗,因为它们看起来不像是量化版本,我下载了较小的模型放入工作流,发现一直报错

    SimonartDec 19, 2025· 1 reaction

    没错,我也发现了

    AiMetatron
    Author
    Dec 19, 2025

    这两个是独立的 diffusion_models, 放到 diffusion_models目录下,用unet加载器加载

    Van978Dec 16, 2025· 5 reactions
    CivitAI

    I’ve noticed that every image I post that was generated using this model gets flagged for review, whereas images generated with the Zimage-turbo model are never flagged. Could you explain why this happens?

    AiMetatron
    Author
    Dec 18, 2025· 3 reactions

    😭

    4423025858Dec 16, 2025· 2 reactions
    CivitAI

    现在我理解了这玩意的难度,我在尝试训练它的lora补全它的解剖错误,但是我发现它特别容易拟合成一个奇怪的形状。虽然它正确的出现在了那个正确的位置,但在我有限的经验和记忆中,,它不是那个形状,也许是我的训练集问题,现在我在补充更多的照片进入训练集,希望练成一个正常的生殖器该有的样子。。。。。

    h3x33421Dec 16, 2025

    会不会是标记做得不对。。

    4423025858Dec 16, 2025

    @h3x33421 应该是过拟合了,我找了它的中间步数,好多了,至少在我有限的记忆中,哦,对就是哪个玩意,只是我不知道能否上传这个lora

    @4423025858 务必上传呀大佬

    YSCyscDec 24, 2025· 2 reactions

    这个参见Pony作者写的那篇文章,差不多就可以理解为什么会这样。

    例如,Pony是基于SDXL的,SDXL具有有限的NSFW功能,各种性爱姿势它是不懂的,那么训练集就要足够多让它来学习这个概念,同时数据集要多样化来防止过拟合,包括后来的Illustrious XL模型也是。百万级以上数据集,在服务器上面训练了几个月。

    顺便分享下,我以前训练LorA的一个过程。很早以前SD1.5时代,麦橘是个很流行的模型,也是有限的NSFW功能,例如能画裸露图,但是不能做乳交的图,搭配LoRA也不行,因为它压根就不明白paizuri这个概念。我当时用了数百张包含paizuri性爱的图,既充足又多样化,让其学会了paizuri这个概念,最终满足了我的需求。SD1.5时代虽然当时有很多性交姿势的LoRA,但是几乎都是二次元基于NAI的,无法很好用于麦橘这种写实模型,当时自己炼才满足。至于后来的SDXL以后的模型,训练LoRA更容易了,包括后来的Flux.

    对于大模型已经理解的概念(例如人物LoRA 只需要20-30张效果就很不错),少量数据集少量步数就有很好的效果,特别是Flux,这种训练是对已有概念的微调;对于

    大模型不能理解的概念,需要大量的数据集和很大的训练步数来让模型理解这个概念,训练的时候还要同时训练text_encoder让它来理解这个新概念。

    总结一句话:

    已有知识 → 少量数据即可偏移:已经会“骑自行车”的人学电动车 → 只需少量练习;

    缺失知识 → 需要大量数据来“建模”:完全没见过“车”的人学电动车 → 必须先学车的概念,难度和样本量都大。

    Difused_TerroristDec 17, 2025· 4 reactions
    CivitAI

    here take some buzz my lord

    guyuan55555Dec 18, 2025· 4 reactions
    CivitAI

    这个是否支持生成NSFW图片?原版zit对nsfw比较保守。

    AiMetatron
    Author
    Dec 18, 2025· 1 reaction

    请结合一下MysticXXX Z V2,或者其他NSFW解锁LoRA

    https://civitai.com/models/2206377/zit-mystic-xxx

    qekDec 18, 2025

    @AiMetatron 那个LoRA我用起来效果不太好 :(

    AiMetatron
    Author
    Dec 18, 2025· 7 reactions
    CivitAI

    2026.1.1 VX 新规执行,请移步TG https://t.me/+Ka2NhoUF5W1iMDRl

    qekDec 18, 2025· 2 reactions

    Sad

    CivitaiNonsenseDec 18, 2025· 1 reaction
    CivitAI

    Draw Things app by Liuliu Compatibility not offered. How come.

    qekDec 18, 2025

    It may be meh then. Unless a possible better performance for some users is real. Doesn't support Z at all?

    Aerospace0076Dec 20, 2025· 1 reaction

    Dec 15, 2025 release works fine. First, install the official DrawThings Z Image 1.0 model (6-bit version is smaller if you don’t plan to play with that model). That will install the proper z-image text encoder and vae as well. Then you can install any another z image checkpoint (I installed the 5.7GB fp8 REDZimage) and it should work.

    CivitaiNonsenseDec 21, 2025

    @Aerospace0076 there are so many variations, so you imported the pruned model fp8?...i thought pruned models don't import into draw things ...but so cool have to try right away.. thanks so much

    Aerospace0076Dec 21, 2025· 1 reaction

    @CivitaiNonsense  Pruned models seem to work just fine in my experience. Enjoy.

    GetcontactDec 20, 2025· 3 reactions
    CivitAI

    Have new Z-version nsfw content? Female

    AiMetatron
    Author
    Dec 21, 2025· 4 reactions

    Here we go : https://civitai.com/models/2242173/dark-beast-z-or-dbz-or-updated-dec21-or-latest-dbzv1 Dark Beast Z-造相·黑-DBZ⚡️image

    Dark Beast Z Super Fusion – 模型介绍

    Dark Beast Z Super Fusion 是一款专为顶级创作者打造的融合模型,旨在实现超越现实边界的极致超写实情色影像。通过精妙融合 MysticXXX ZIT v2 的强烈诱惑写实感、Kawaii Kinky[NE] v1 的大胆挑逗能量,以及 Detail DeamonZ 无与伦比的精细细节掌控能力,并以突破性的超摄影写实底模 BEYOND REALITY sft ZiTurbo 为基础,该模型可生成令人惊艳的 2M 像素输出,具备完美无瑕的皮肤纹理、栩栩如生的面部表情,以及超越真实摄影的超精细光影表现——即使在最复杂且暗示性的构图中亦游刃有余。

    最终成果是一个极具多功能性的检查点,擅长渲染超越现实的视觉效果:想象皮肤渲染得如此精湛,能清晰呈现每一个毛孔与细微血管;眼神深邃仿佛能直刺观者灵魂;身体兼具解剖学完美与自然瑕疵,搭配挑逗姿势、精致内衣、复杂束缚元素,以及传达纯粹、未经滤镜的欲望与强烈情感的表情。

    GetcontactDec 23, 2025

    @AiMetatron thanks!

    mariedoll123Dec 20, 2025· 5 reactions
    CivitAI

    Absolute monster of a model that easily wipes the floor with all the " pretenders " out there.

    if you want to try zimage this is the one for you .. Many thanks 👍👍👍

    76543Dec 21, 2025· 1 reaction
    CivitAI

    Can you upload a workflow for the fp8 version? I keep getting errors. Thanks

    qekDec 21, 2025

    Already posted

    Mrpopo714Dec 23, 2025· 6 reactions
    CivitAI

    @AiMetatron it seems that the lora for redcraftRedzimageUpdatedDEC03_redzimage15AIO is the exact same name as the model. As a result, civitai detects it as a "checkpoint" not "lora" will there be a fix to that?

    123bbtDec 23, 2025· 1 reaction
    CivitAI

    FP16或Q8plz

    overnomDec 23, 2025· 4 reactions
    CivitAI

    character lora traned with original ziamge doesn't seem really work with this model?

    qekDec 24, 2025· 1 reaction

    Wrong

    RPxGoonDec 25, 2025· 7 reactions
    CivitAI

    what is this model supposed to do differnt than zit base model? very unclear how this is any different or what its advantages are. can someone sell me on a reason to download this?

    qekDec 25, 2025

    Can try to get better porn images, but needs loras anyway

    qekDec 26, 2025· 1 reaction

    @xGoon I don't get Asian faces even if don't provide such details, but Z is made in China~

    studiokevinabanto112Dec 25, 2025· 8 reactions
    CivitAI

    ERROR: Could not detect model type of: /workspace/ComfyUI/models/checkpoints/z_image/redcraftRedzimageUpdatedDEC03_redzimage15AIO.safetensors

    Prompt executed in 0.98 seconds

    FETCH ComfyRegistry Data: 65/115

    FETCH ComfyRegistry Data: 70/115

    FETCH ComfyRegistry Data: 75/115

    FETCH ComfyRegistry Data: 80/115

    FETCH ComfyRegistry Data: 85/115

    FETCH ComfyRegistry Data: 90/115

    FETCH ComfyRegistry Data: 95/115

    FETCH ComfyRegistry Data: 100/115

    FETCH ComfyRegistry Data: 105/115

    got prompt

    Failed to validate prompt for output 80:

    * CheckpointLoaderSimple 62:

    - Value not in list: ckpt_name: 'redcraftRedzimageUpdatedDEC03_redzimage15AIO.safetensors' not in ['qwen_image_edit_2509/Qwen-Rapid-AIO-NSFW-v14.1.safetensors', 'z_image/redcraftRedzimageUpdatedDEC03_redzimage15AIO.safetensors']

    * LoraLoaderModelOnly 71:

    - Value not in list: lora_name: 'maya z image.safetensors' not in (list of length 67)

    Output will be ignored

    invalid prompt: {'type': 'prompt_outputs_failed_validation', 'message': 'Prompt outputs failed validation', 'details': '', 'extra_info': {}}

    FETCH ComfyRegistry Data: 110/115

    FETCH ComfyRegistry Data: 115/115

    FETCH ComfyRegistry Data [DONE]

    [ComfyUI-Manager] default cache updated: https://api.comfy.org/nodes

    FETCH DATA from: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/custom-node-list.json [DONE]

    [ComfyUI-Manager] All startup tasks have been completed.

    got prompt

    Warning, This is not a checkpoint file, trying to load it as a diffusion model only.

    !!! Exception during processing !!! ERROR: Could not detect model type of: /workspace/ComfyUI/models/checkpoints/z_image/redcraftRedzimageUpdatedDEC03_redzimage15AIO.safetensors

    Traceback (most recent call last):

    File "/workspace/ComfyUI/execution.py", line 516, in execute

    output_data, output_ui, has_subgraph, has_pending_tasks = await get_output_data(prompt_id, unique_id, obj, input_data_all, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb, v3_data=v3_data)

    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

    File "/workspace/ComfyUI/execution.py", line 330, in get_output_data

    return_values = await asyncmap_node_over_list(prompt_id, unique_id, obj, input_data_all, obj.FUNCTION, allow_interrupt=True, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb, v3_data=v3_data)

    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

    File "/workspace/ComfyUI/execution.py", line 304, in asyncmap_node_over_list

    await process_inputs(input_dict, i)

    File "/workspace/ComfyUI/execution.py", line 292, in process_inputs

    result = f(**inputs)

    ^^^^^^^^^^^

    File "/workspace/ComfyUI/nodes.py", line 581, in load_checkpoint

    out = comfy.sd.load_checkpoint_guess_config(ckpt_path, output_vae=True, output_clip=True, embedding_directory=folder_paths.get_folder_paths("embeddings"))

    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

    File "/workspace/ComfyUI/comfy/sd.py", line 1344, in load_checkpoint_guess_config

    raise RuntimeError("ERROR: Could not detect model type of: {}\n{}".format(ckpt_path, model_detection_error_hint(ckpt_path, sd)))

    RuntimeError: ERROR: Could not detect model type of: /workspace/ComfyUI/models/checkpoints/z_image/redcraftRedzimageUpdatedDEC03_redzimage15AIO.safetensors

    Prompt executed in 0.02 seconds

    TienXDec 26, 2025· 3 reactions

    It's a Lora, but they have it listed as a checkpoint, you need to go down a bit and download the 19 gb version, it's so dumb, they have them all named the same thing...or just use it with zimage as a lora

    sbili5726430Jan 6, 2026· 2 reactions
    CivitAI

    你好,我下载了AIO的 GGUF文件,不知道用什么节点加载此模型,用普通的UNET load(gguf)加载器无法加载,运行后显示报错 ,谢谢

    rickie3stef979Jan 7, 2026

    同样问题

    shuidaiJan 13, 2026

    +one

    kivitaiJan 28, 2026

    The same issue here.

    1636676617214Jan 29, 2026

    蹲

    PdidiJan 7, 2026· 3 reactions
    CivitAI

    Im confused af, is the 670 mb checkpoint merge better than the full GIGA size models listed below it? Why not have examples for them too? Again, I'm confused and learning these things.

    qekJan 7, 2026· 1 reaction

    ?
    It's a LoRA

    PdidiJan 7, 2026

    @qek damn, I didnt know checkpoint merge is a Lora, thank you :D

    giorgioosn367Jan 8, 2026

    @Pdidi no way this is a lora...

    SSSONGGGJan 8, 2026· 3 reactions
    CivitAI

    请问11GB的修剪版本也是AIO模型吗?

    AiMetatron
    Author
    Jan 9, 2026· 2 reactions

    不是的,是独立的 diffusion model

    6vidit9Jan 13, 2026· 4 reactions
    CivitAI

    Is this the ultimate ZIT banger??, ABSOLUTELY.

    Cinematic af, kudos to the author🤘🥂

    AiMetatron
    Author
    Jan 14, 2026· 4 reactions

    Like @Mrpopo714 @Tozi_White, and so many other artistic masters, your work has greatly expanded our horizons and elevated the use of the Z-image and others model to new heights. Salute to you! 🥂💗

    6vidit9Jan 14, 2026· 1 reaction

    @AiMetatron Thank you for your amazing work, dude. This is a fantastic model — the realism and composition are fantastic. This shows how much effort and time you invested.

    Mrpopo714Jan 20, 2026

    @AiMetatron Very much appreciated!

    Checkpoint
    ZImageTurbo

    Details

    Downloads
    38,994
    Platform
    CivitAI
    Platform Status
    Available
    Created
    12/7/2025
    Updated
    10/27/2026
    Deleted
    -

    Files

    redcraftHybridH3Krea2t_redzit15AIO.gguf

    redcraftHybridH3Krea2t_redzit15AIO_trainingData.zip

    Available On (1 platform)

    Same model published on other platforms. May have additional downloads or version variants.