<h1 id="floriane-is-a-spring-muse-in-her-late-20's.-p9r38jtul">Floriane is a Spring Muse in her late 20's.</h1><p>Floriane is a fictitious person. This Textual Inversion (TI) was designed to embody a timeless beauty inspired by alluring elegance and athleticism.</p><p>This is purely an experiment aimed at helping me understand vectors within TI's and v1 has many weaknesses I aim to improve (e.g., hair color consistency, face consistency). The aim of <em>Floriane</em>, in contrast to <a target="_blank" rel="ugc" href="https://civarchive.com/models/993946/snofrid">Snöfrid </a>and <a target="_blank" rel="ugc" href="https://civarchive.com/models/1003684/marsaili">Marsaili</a>, is that vectors for photography style were included and a reduced token count.</p><p>For v2, my goals are to achieve a more consistent face and body shape while maintaining flexibility in prompting. I believe I can accomplish this through image-based training once I have an established dataset spanning multiple models and styles.</p><h1 id="floriane-ti-changelog:-zfo1lb4u1">Floriane TI Changelog:</h1><h2 id="v1-initial-experimental-version.-5iws4ubx4"><strong>v1</strong> - <em>Initial experimental version.</em></h2><p>Initial experimental version that uses the <a target="_blank" rel="ugc" href="https://github.com/klimaleksus/stable-diffusion-webui-embedding-merge">embedding merger tool</a> to focus on facial features, leaving backgrounds, and art styles entirely up to the user's creative direction. There are some style tokens to push it towards <strong>glamour </strong>and <strong>fashion photography</strong>. This approach was chosen to hopefully sidestep common pitfalls seen in other embeddings, such as over-fitting to training artifacts, feature entanglement, poor dataset curation, and prompt hijacking—where a TI imposes its learned context too strongly, overriding user-specified prompts. The result is a versatile tool that maintains the core essence of <strong>Floriane's facial and body shape features</strong> while enabling dynamic adaptability across different styles, contexts, and artistic visions.</p>

Files with SHA256: 0eef09decf2089be89865f753ac47493d9ee355c270d94dba9bb27022a188363