This is a 4-vector embedding of Billie Eilish made with a dataset of 90 images and 10000 steps.I tested the embedding with different models. The sample images were generated with the following models.<a target="_blank" rel="ugc" href="https://civarchive.com/models/4201/realistic-vision-v12">Realistic Vision V1.2</a><a target="_blank" rel="ugc" href="https://civarchive.com/models/1173/hassanblend-1512-and-previous-versions">HassanBlend 1.5</a><a target="_blank" rel="ugc" href="https://civarchive.com/models/3450/moistmix-v1">MoistMix V1</a><a target="_blank" rel="ugc" href="https://civarchive.com/models/3627/protogen-v22-anime-official-release">Protogen v2.2</a><a target="_blank" rel="ugc" href="https://civarchive.com/models/3666/protogen-x34-photorealism-official-release">Protogen x3.4</a><a target="_blank" rel="ugc" href="https://civarchive.com/models/3816/protogen-x53-photorealism-official-release">Protogen x5.3</a><a target="_blank" rel="ugc" href="https://civarchive.com/models/3867/protogen-x58-rebuilt-scifianime-official-release">Protogen x5.8</a><a target="_blank" rel="ugc" href="https://civarchive.com/models/4331/protogen-nova-experimental-model-official-release">Protogen Nova</a><a target="_blank" rel="ugc" href="https://civarchive.com/models/3811/dreamlike-photoreal-20">Dreamlike Photoreal 2.0</a><a target="_blank" rel="ugc" href="https://civarchive.com/models/3950/art-and-eros-aeros-a-tribute-to-beauty">Art &amp; Eros (aEros)</a><a target="_blank" rel="ugc" href="https://civarchive.com/models/1265/analog-diffusion">Analog Diffusion</a><a target="_blank" rel="ugc" href="https://civarchive.com/models/1274/dreamlike-diffusion-10">Dreamlike Diffusion 1.0</a><a target="_blank" rel="ugc" href="https://civarchive.com/models/1366/elldreths-lucid-mix">Elldreth's Lucid Mix</a><a target="_blank" rel="ugc" href="https://civarchive.com/models/4384/dreamshaper">DreamShaper</a>It seems to perform particularly well with DreamShaper , Realistic Vision 1.2 and the different Protogen models.It performs worse with Dreamlike Diffusion 1.0 and Analog Diffusionat least according to the tests I've done.

Files with SHA256: 134a81ff2ab88e59fe27f7310fa46e34eef128643bb3990c0f59e3c1d136ac69