<p>This is a bit of a experiment as many of the TI's I tried failed and cost several days of work. In the end I did a bit of a mix between Subject and Style, combined with prompt mixing.</p><p></p><h3>Samples generated on Protogen v2.2</h3><p></p><h2>How to use:</h2><p>It's best to use this as a mix-in with a subject. For example, with a woman you would do <code>(woman mixed with (tloustyle))</code></p><p></p><p>The model has a bias with trying to produce girls, which is odd considering there was like 2 female infected subjects trained on. Anyways, to defeat this, when doing a man, you may need to add negative prompt such as <code>((female, woman, girl))</code></p>

Files with SHA256: 7063287f37563cce985379ac300099d6fcb2477b9bd8f50074080c004bdf3633