CivArchive
    Preview 2090138Preview 2087702Preview 2039930Preview 2039913Preview 2040524Preview 2039909Preview 2041455Preview 2041411Preview 2040110Preview 2040155Preview 2047688Preview 2048149Preview 2048209Preview 2048548Preview 2048724Preview 2048795Preview 2048862Preview 2054437Preview 2054344Preview 2054472

    All Sample image are made Base SDXL_1.0 with Hi-res fix only - (No refiner)

    This LoRas enhance/improve image to get more aesthetic/artistic/creative, detailled.

    xl_more_art-full_v1 this version is little more photorealistic than betas but you can also do anime/illustrations.

    Weight 0.8-1 is recommended but you ca push more than 1 on less detailled render. ^^

    This LoRa also contains a training for aerial photos (use "Aerial" keywords), more details here : https://civarchive.com/articles/2030/how-to-create-realistic-aerial-land-art

    Please help me :

    Add to your collection and share your creations here !

    Description

    Get images more photorealistic, add details

    This Lora make images more creative, aesthetic, detailled.

    • weight 0.7 is a good value for cfg scale >= 7

    • weight 0.8-1.0 are a good values for cfg scale 5.5-7.0

    • weight 0.1-0.3 is a good value for small changes

    • Trained with 9 epochs on 6k steps

    I use realistic images to train this lora

    LORA
    SDXL 1.0

    Details

    Downloads
    5,417
    Platform
    CivitAI
    Platform Status
    Available
    Created
    8/15/2023
    Updated
    9/27/2025
    Deleted
    -

    Files

    xl_real_beta1.safetensors

    Mirrors

    Huggingface (1 mirrors)
    CivitAI (1 mirrors)