<h1 id="overview">Overview</h1>Update: use the all-in-one model instead!A helper model to create giant women, mainly to fix shadows, perspective, and size relations; not a one-word-prompt wonder!This LoRA can roughly produce these sizes:<ul><li>Two-story building (~10m)</li><li>Multi-story building (~30m)</li><li>High-rise building (~100m)</li><li>Skyscraper (~300m)</li></ul>The trigger is <code>gntss</code> or <code>giantess</code>. It's important to describe her size in relation to her surroundings. See the example prompts.An additional trigger <code>{10|30|100|300}-meter-tall</code> can help to anchor her size.<h3 id="zib-v1">ZIB - V1</h3>Z-Image can already sort of create giantesses, but they are cartoonish and have distorted anatomies. These models aims to fix these issues.I recommend a dual-stage workflow (split sigmas) with Z-Image doing 25-50% of the steps and Z-Image Turbo doing the rest. This will greatly improve fidelity and realism.<ul><li>Training model: <a target="_blank" rel="ugc" href="https://huggingface.co/Tongyi-MAI/Z-Image">Z-Image</a></li><li>LoRA strength: 1.0 (up to 2.0 on Turbo)</li><li>Sampler: res_2s+beta57 / euler</li><li>Steps: 40+ (8+ on Turbo)</li><li>CFG: 3.0-5.0 (1.0 on Turbo)</li></ul><pre><code>A photograph of a 10-meter-tall giantess with dark hair in a ponytail standing on a city street. Her left hand rests gently against the second-story exterior wall of a residential building as she smiles warmly at a tiny woman leaning out of a window. Below her, cars travel along the road and people walk along the sidewalk.</code></pre>

💁🏼‍♀️ Giantess (Helpers for Z-Image)

Updated version to create all four sizes. Have a look at the example prompts and trigger words.

Files with SHA256: 9b8dae3ed0144414ecbdd82ea78eae689595eb4633490f620367c7cec3332ade