CivArchive
    Vision Vanguard 24 (Krea 2) - V2026.1
    NSFW
    Preview 135269431
    Preview 135269494
    Preview 135269496
    Preview 135269500
    Preview 135269502
    Preview 135269507
    Preview 135269508
    Preview 135269514
    Preview 135269516
    Preview 135269518
    Preview 135269519
    Preview 135269523
    Preview 135269524
    Preview 135269527
    Preview 135269526
    Preview 135269532
    Preview 135269534
    Preview 135269541
    Preview 135269540
    Preview 135269544

    Vision Vanguard 24

    LoRA expertly trained by Divergent AI

    Film-grade realism. A masterly command of natural light and analogue texture.

    The Aesthetic

    Vision Vanguard 24 is the definitive "Documentary Gaze" model. It is designed for creators who demand photographic conviction across an expansive range of subjects. From the gritty, high-contrast intensity of a monochromatic street procession to the hyper-saturated, surreal brilliance of a mid-century still life, this model grounds every pixel in the physical world. It obsesses over the textures that digital models often forget: the beading of water on a whale's skin, the intricate cracks in sun-bleached earth, and the tactile warmth of a knit sweater in shadow.

    Key Features

    • Tactile Surface Fidelity: Renders micro-textures with clinical precision—rust, weathered wood, skin pores, and mineral deposits.

    • The Chiaroscuro Engine: Advanced handling of dramatic light-to-dark transitions, creating depth and weight through directional natural light.

    • Versatile Chromatic Range: Masterfully transitions between grainy, high-contrast black-and-white and hyper-saturated, complementary color palettes.

    • Atmospheric Scale: Exceptional rendering of vast landscapes, receded horizons, and the subtle "haze" of overcast environments.

    Ideal Use Cases

    • Documentary Photography: Capture raw, high-stakes human moments with the weight of traditional film.

    • Abstract Geological Art: Generate "God's eye" aerial views of salt flats, mineral pools, and desert dunes.

    • Industrial & Rural Decay: Explore the beauty of rust, ruins, and overgrown remnants of human industry.

    • Macro Still Life: Create intense, clinical observations of small objects, wildlife, and biological textures.

    Getting Started

    To unlock the "Documentary Gaze," use the primary trigger word documentary photograph. Structure your prompts with camera language first, followed by subject, environment, and finally, specific texture or lighting details.

    Primary Trigger Words: documentary photograph, eye-level medium shot

    Secondary Modifiers: high-contrast black and white, saturated color palette, film grain, shallow depth of field


    Created by Divergent AI


    Description

    Technical Specifications

    •   Base Model: Krea 2

    •   Version: 2016.1

    •   Recommended Weight: 0.8-1.0

    •   Resolution: 512 (Trained) 1024-1536 (Output)

    •   CFG Scale: 3.5 (Base) 1.0 (Turbo)

    •   Steps: 35-50 (Base) 8 (Turbo)

    FAQ

    Comments (4)

    TheEventHorizonJun 30, 2026
    CivitAI

    Should be really useful for cinematic images, im a bit confused, why is it named vanguard 24, if you uploaded it after vanguard 25?

    thedivergentai
    Author
    Jul 1, 2026

    They use a dataset based on images related to different years. So each vision vanguard is all based on photography. But they all have completely different datasets.

    Vision Vanguard 25 = images from 2025
    24 = images from 2024 etc.

    I actually have datasets heading all the way back to 2012. Maybe I'll release them one day.

    They should all be very good at generalization and any photographic content. Their databases are all at least 1000 hi-res images covering large diverse subject matter so they're really more just like an overall realism style if you want to view it like that.

    TheEventHorizonJul 1, 2026· 1 reaction

    @thedivergentai Speaking of years, Krea desperately needs some 80s/90s/early 2000s loras to get retro styles, can I make a request for them?

    thedivergentai
    Author
    Jul 3, 2026· 1 reaction

    @TheEventHorizon Hey so I heard you on the decades themed stuff. I'm working on it. Trying to build up a nice large dataset. I'll start with the 80's.

    Makeup of dataset will be something like:

    Documentary: ~40% | Street, home, transit, candid life

    Retail ephemera: ~30% | Ads, catalogs, packaging, mall signage, magazine pages

    Fashion/music/events: ~20% | Concerts, sports, youth culture, neon fashion

    Technology/gaming: ~10% | Consoles, arcades, computers, VHS-era objects

    LORA
    Krea 2

    Details

    Downloads
    588
    Platform
    CivitAI
    Platform Status
    Available
    Created
    6/29/2026
    Updated
    7/7/2026
    Deleted
    -

    Files

    (Krea 2) Vision Vanguard 24 V2026.1.safetensors