CivArchive
    Detailer TA+Z-Image-Turbo - PRO
    Preview 1
    Preview 2
    Preview 3
    Preview 4
    Preview 5
    Preview 6
    Preview 7
    Preview 8
    Preview 9
    Preview 10

    One LoRA to Rule Them All.

    Stop juggling a dozen different plugins. You just found the Ultimate Upgrade. Meet Detailer TA. We took the engine of Z-Image-Turbo-PRO and tuned it until it broke, then tuned it back up to perfection. Whether you want Raw Realism, Nostalgic Polaroid, Sleek CGI, 3D Render, or even Anime, this LoRA adapts instantly. It’s not just a detailer; it’s the master key for every style.

    Why you need this in your toolkit:

    * The Universal Translator: Works seamlessly on Editorial photos, gritty Polaroids, shiny 3D renders, and Anime line art. It fixes them all.

    * No More "Plastic" Faces: We crushed the CGI look. You get real pores, grain, and authentic texture.

    * Anatomy Police: It keeps hands, fingers, and faces in check so you don't get nightmare fuel.

    * Lazy Genius: Works at low settings, meaning you get high-end results without the headache.

    ⚠️ Warning: You might forget what "bad generation" looks like.

    ---

    ### 📋 Usage Guide (The "Swiss Army Knife" Setup)

    The Trigger Rule:

    > Don't use trigger words. Keep your prompt clean and let the Detailer do the heavy lifting.

    Recommended Strength:

    > 0.6 to 0.8

    > Adjust slightly based on style—lower for Anime, higher for gritty Realism.

    > Mixing It Up: If you’re stacking this with other LoRAs, keep the strength low (0.3–0.5).

    The "Vibe" Menu (Pick your flavor):

    Want a specific look? Add these to your prompt:

    * Nostalgia: `(SHOT ON CRT SOFT VHS:1.4)` — Warm, analog glow.

    * Authentic: `(SHOT ON IPHONE6:1.6)` — Raw, realistic, snapshot style.

    Pro Settings:

    * CFG: Keep it low (~1).

    * Sampler: DPM++ 2M SGM Uniform or heunpp2 for crisp edges.

    Perfect for creators who switch styles daily and refuse to compromise on quality.

    (Created by Immortal — Because "Good Enough" isn't in our vocabulary.)

    Description

    "This model's training material does not include images of celebrities or copyrighted content. Please do not use this model to generate images of celebrities." Add these in your prompts ( SHOT ON CRT SOFT VHS:1.4), ( SHOT ON IPHONE6:1.6) for effects. Try vae-ft-mse-840000-ema-pruned.ckpt for editorial quality. Try ae.sft , None , automatic as VAE for realistic quality. This is sampler i use with steps dpmpp_2m sgm_uniform/ heunpp2 linear_quadratic with 9 steps and CFG Scale 1. You can try upscaler , it works best with 4x-UltraSharp , R-ESRGAN 4x+ , depend on what type of image you want. Keep Denoising Strength to 0.3 Use trigger words ae, model1 or ae, model2 if want to try something more add these ae-muse, ae-editorial, ae-ugc, ae-swim
    LORA
    Z-Image-Turbo

    Details

    Downloads
    71,162
    Platform
    TensorHub
    Platform Status
    Available
    Created
    12/7/2025
    Updated
    2/12/2026
    Deleted
    -
    Trigger Words:
    imrt

    Files

    6e96d2d9-897c-4721-a1a8-811c6adb36cb.TA_trained.safetensors