z-image / 2214055 /2492710 /README.md
Sentinel7's picture
Upload 2214055/2492710/README.md with huggingface_hub
24568d6 verified
|
raw
history blame
4.86 kB
metadata
license: other
tags:
  - photorealistic
  - portrait
  - storytelling
  - style
  - vintage

"Delicate Mood" Google Nano Banana Pro Moody Portrait Style [Z-Image-Turbo] by AutoPastel - V1 - 2492710

Model Type: LORA

Base Model: ZImageTurbo

Trigger Words: None

Tags: photorealistic, portrait, storytelling, style, vintage

Gallery

Description

LoRA Bounty: https://civitai.com/bounties/10858/8000-buzz-z-image-turbo-moody-portrait-challenge-bounty

End Date: Jan 9, 2026


This is a style LoRA trained to bring the Google Nano Banana Pro portrait vibe into Z-Image-Turbo—that soft, emotional, analog-film feeling that makes faces look like they’re carrying a story.

Ever since Nano Banana Pro released, I kept coming back to it for one reason: even with the same prompt, it tends to produce portraits that feel more moody and human than most models, with a beautifully imperfect film texture. I wanted that same atmosphere—without switching ecosystems—so I distilled the look into a LoRA for Z-Image-Turbo.

What it’s trained on

  • Hand-picked 30+ of my own Nano Banana Pro generations
  • Curated specifically for moody portrait / emotional figure aesthetics

Sample Images

What it adds to your images

This LoRA is designed to enhance mood and introduce analog/optical character, including:

  • Analog film texture + noticeable film grain
  • Shallow depth of field and soft falloff (portrait-friendly separation)
  • Occasional ghosting / double exposure for a dreamy, imperfect feel
  • Subtle bloom/halation, flare streaks, and "glass" artifacts
  • A characteristic red edge / red margin light leak (edge burn / light leak)

Also, Z-Image-Turbo can sometimes feel a bit too consistent across random seeds. This LoRA helps increase visual variety and adds controlled imperfections that make portraits feel less “generated” and more photographed.

Recommended settings

  • LoRA weight: 0.7 – 1.0 (sweet spot)

    • 0.7–0.85: subtle mood + texture
    • 0.9–1.0: stronger film character, more artifacts (ghosting/light leaks may appear)
  • No trigger words needed but "moody portrait" might be of help.


Author: AutoPastel

Model: CivitAI Model Page

Archive: CivArchive Page