CivArchive
    Preview 112347820
    Preview 112347837
    Preview 112347840
    Preview 112347859
    Preview 112347862
    Preview 112347878
    Preview 112347885
    Preview 112347890
    Preview 112347909
    Preview 112347940
    Preview 112347956
    Preview 112347981
    Preview 112347992
    Preview 112348139
    Preview 112348160
    Preview 112348165
    Preview 112348187

    About ZImageTurbo version

    It has been observed that due to ZImageTurbo's base model being highly photorealistic,

    whilst the training dataset is predominantly anime-style,

    excellent results tend to manifest at two distinct extremes:

    2D anime-style outputs appear between 0.8–0.6,

    and 2.5D–3D outputs appear between 0.2–0.4.

    Additionally, when using the LORA loader, please select 'only load model one'.

    Don't choose the one with 'clip'... I've no idea why either.

    發現一件事情因為ZImageTurbo底模非常寫實

    但是訓練資料輯非常的動漫

    所以好的結果會出現在兩個不同的極端

    2D動漫結果出現在0.8-0.6,2.5-3D出現在0.2-0.4

    另外lora loader請選擇only model別選含clip的...我也不知道為什麼

    See-through-body

    About this version

    triger

    see-through-body, transparent skin, machinery skeleton, cyborg, mechanical parts, science fiction,

    advanced prompts

    see-through-body, transparent skin, machinery skeleton, cyborg, mechanical parts, science fiction, mechabare, mechanical parts, solo, science fiction,sf, cyborg, fighting stance,

    with dynamic prompts as a wildcard

     see-through-body, transparent skin, machinery skeleton, cyborg, mechanical parts, science fiction 
    {indoors,| outdoors| , gradient background,|  simple background,|  two-tone background, }
    {close-up,| portrait,| upper body, | full body, }, extremely detailed eyes, fantastic details full face, mouth, 
    {from behind, |from side, |from below, |from above, }
     1girl, cyborg, { curvy, sexy | fighting stance, } { holding weapon, holding {sword,| polearm,| knife,| rifle,| axe,| pistol} {cyborg,mechanical arms, mechanical eyes, {short hair,| long hair, | braid, | bob cut,| very long hair, } , face focus | exoskeleton, breasts, mechanical arms, mechanical legs, face_focus | mecha robot, holding weapon, |}, }{1-3$$ Oil paint|Watercolor|Acrylic|Gouache|Tempera|Pastel|Charcoal|Ink|Colored pencil|Graphite| anime screenshot, | western comics \(style\), | manga style }, {multiple-tonbackground |simple background | gradient background | indoors, | outdoors, }

    V1

    mechabare_illustrious

    About this version

    train by image + leco

    triger

    mechabare, mechanical parts, science fiction , cyborg,

    advanced prompts

    mechabare, mechanical parts, solo, science fiction,sf, cyborg, fighting stance,

    with dynamic prompts as a wildcard

    mechabare, mechanical parts, solo, science fiction,sf, 1girl, cyborg, { curvy, sexy | fighting stance, } { holding weapon, holding {sword,| polearm,| knife,| rifle,| axe,| pistol} {cyborg,mechanical arms, mechanical eyes, {short hair,| long hair, | braid, | bob cut,| very long hair, } , face focus | exoskeleton, breasts, mechanical arms, mechanical legs, face_focus | mecha robot, holding weapon, | non-humanoid robot,} | { no humans, vehicle focus,{ military vehicle, | sports car | car | gun | plan |ship } }, }{1-3$$ Oil paint|Watercolor|Acrylic|Gouache|Tempera|Pastel|Charcoal|Ink|Colored pencil|Graphite| anime screenshot, | western comics \(style\), | manga style }, {multiple-tonbackground |simple background | gradient background | indoors, | outdoors, }, ,dynamic visul, limited palette theme,matel color trim,{ gradient colors, | limited palette, | vivid colors, | pastel colors, | two-tone colored } , { upper body, | full body, | Wide shot|Long shot|Full body|Three-quarter shot|Mid shot|Upper body|Portrait|Close-up|Extreme close-up}, {1-2$$ barcode tattoo, | national mark| floral print, | barcode tattoo,|gears| energy pipe,| glowing leds| mechanical wings, | high heels,| speed lines,}

    V1

    wegiht 3~0

    For Style Capture Contest trained by leco method

    sweet point 3~1.5

    如果你喜歡我的工作可以幫我買杯咖啡,謝謝

    If you like what I do, you can buy me a cup of coffee. Thanks.

    私の仕事を気に入っていただけたら、コーヒーを一杯おごってください。 ありがとうございます。

    https://ko-fi.com/vicentma

    Description

    It has been observed that due to ZImageTurbo's base model being highly photorealistic,

    whilst the training dataset is predominantly anime-style,

    excellent results tend to manifest at two distinct extremes:

    2D anime-style outputs appear between 0.8–0.6,

    and 2.5D–3D outputs appear between 0.2–0.4.

    Additionally, when using the LORA loader, please select 'only load model one'.

    Don't choose the one with 'clip'... I've no idea why either.

    發現一件事情因為ZImageTurbo底模非常寫實

    但是訓練資料輯非常的動漫

    所以好的結果會出現在兩個不同的極端

    2D動漫結果出現在0.8-0.6,2.5-3D出現在0.2-0.4

    另外lora loader請選擇only model別選含clip的...我也不知道為什麼

    FAQ

    LORA
    ZImageTurbo

    Details

    Downloads
    425
    Platform
    CivitAI
    Platform Status
    Available
    Created
    12/7/2025
    Updated
    4/23/2026
    Deleted
    -
    Trigger Words:
    mechabare, cyborg

    Files

    cyborg_Z_v2.safetensors

    Mirrors