CivArchive
    Preview 116988823
    Preview 116995746
    Preview 116991169
    Preview 116998181
    Preview 116998293

    🟢THE IMAGES HERE IS MADE USING THE MAX REALISM KIT,

    💎 MAX REALISM KIT — $49.99, 8 products, 4 comfyui workflows, 19-page guide on AI influencer setup, updates free forever
    👉 https://monkeyforever.gumroad.com/l/max-realism-kit👈

    💬 Discord (community, 2.300+ members): discord.gg/69nQp4F9bq

    WHAT'S IN IT

    🧩 Krea 2 txt2img + img2img, custom sampler, pre-tuned

    🔬 Skin detailing pass — pores that survive the final pass

    🎨 ChromaGrade — inherits the exact grade of any reference

    🧠 Multi-LoRA stacking that actually works

    🏋️ Full LoRA training toolkit — train your own character

    🔞 Full NSFW support with dedicated detailing

    Running an ai influencer account? the MAX REALISM KIT fixes this.
    You need max realism to get more traffic.

    🚀 NEW: Krea 2 v1.0

    A full retrain of the Krea 2 version, built for stability and control.

    • Revamped training settings for much more stable results

    • Higher-quality dataset with fewer outliers

    • Precise captions, which give much better prompt adherence

    • Handles higher strength. Up to 2.0 is recommended. 1.5 is the sweet spot. 1.0 does little, and 2.0 improves the overall style without breaking the image. Lower it if the effect is too strong.

    The dataset is real-life smartphone photography: mirror selfies, beach and pool shots, bedrooms, bathrooms, car interiors, city streets at night, plus cars, landscapes and everyday scenes. SFW and NSFW are both covered.

    ✍️ How to prompt Krea 2 v1.0

    It was trained on detailed natural-language captions, so write prompts the same way. Describe the:

    • Subject: build, hair, skin details (freckles, tan lines, tattoos, piercings, jewelry)

    • Outfit and pose: what they're wearing and doing

    • Setting: where the photo is taken and what's in the background

    • Framing: mirror selfie, high-angle, low-angle, close-up, "framed from the upper thighs up"

    • Lighting: direct sunlight, diffuse daylight, warm indoor lighting, flash

    Example:

    A slim young woman takes a mirror selfie in a bathroom, holding a smartphone in one hand. She has long brown hair and freckles across her nose, and wears a grey ribbed tank top with light-wash jeans. White tiled walls are behind her under warm indoor lighting, framed from the upper thighs up.

    Old tag-style prompts and Z-Image workflows won't carry over.


    📜 Version History

    Krea 2 v0.5 (Experimental): The first Krea 2 port. It was trained on 600 images of modern iPhone photos with a larger NSFW spread, replacing the older early-2000s data.

    Z-Image-Turbo versions:

    • v5.0 (Real Life): Largest Z-Image update, with 100+ new images. It has fewer artifacts, better anatomy and better lighting physics.

    • v3.5 Photorealism: Optical realism: subsurface scattering, specular highlights, chromatic aberration. Strong on mirror selfies.

    • v3.0: Expanded dataset that balances the raw amateur vibe with high-fidelity detail.

    • v2.0: Experimental "camera roll" aesthetic. It's gritty and unpolished, and sometimes temperamental.

    • v1.0: The foundation, a lightweight base layer of realism and texture.

    💡 Z-Image-Turbo tips: Use strength 0.60–0.70. Optional tags:

    • Camera roll: amateur digital snapshot, candid, smartphone capture, high ISO noise, direct on-camera flash

    • Detail: visible pores, visible vellus hair, subsurface scattering, detailed skin texture

    • Optics: wide-angle lens, barrel distortion, chromatic aberration, depth of field

    Description

    Images are much more detailed now. More stable. Doesn't force amateur low quality images. Changed dataset around, added more subjects = giving better and more different faces and etc.

    FAQ