I present my attempt at a Illustrious character LoRA for Cullen Rutherford from Dragon Age.<hr />V3 is my third version and attempt and to further correct the face details. The quality definitely improved as well as the face consistency. LoRA strength recommendations<ul><li>Use LoRA strength 0.7 with about CFG 3-4.</li></ul>The tags for V3 are as follows:main tags<pre><code>cullenrutherford, 1boy, mature male</code></pre>outfit tags<pre><code>fur-trimmed cloak, pauldrons, breastplate, shirt, gauntlets, gloves, belt, black pants, boots</code></pre>The settings I used for my generation testing were 3-4 CFG, 20 steps, DPM++ 2M SDE - Karras on Forge with an adetailer (specifically face_yolov8s).<hr />V2 is my second attempt and to try softening the details with the generations by adjusting the training a bit. My testing was definitely smoother and had less of that rough/3d look V1 had. I also reduced the size of the LoRA. LoRA strength recommendations<ul><li>Use LoRA strength 0.7 with about CFG 3-5.</li></ul>The tags for V2 are as follows:main tags<pre><code>cullen_rutherford, 1boy, blonde hair, brown eyes</code></pre>outfit tags<pre><code>fur-trimmed cloak, pauldrons, breastplate, shirt, gauntlets, gloves, belt, black pants, boots</code></pre>The settings I used for my generation testing were 3-5 CFG, 20 steps, DPM++ 2M SDE - Karras on Forge with an adetailer (specifically face_yolov8s).<hr />V1 is my initial attempt and produced very nice results. LoRA strength recommendations<ul><li>Use LoRA strength 0.7-0.8 with about CFG 3-5.</li></ul>The tags for V1 are as follows:main tags<pre><code>cullen_rutherford, 1boy, male, brown eyes</code></pre>outfit tags<pre><code>fur-trimmed cloak, pauldrons, breastplate, shirt, gauntlets, gloves, belt, black pants, boots</code></pre>The settings I used for my generation testing were 3-5 CFG, 20 steps, DPM++ 2M SDE - Karras on Forge with an adetailer (specifically face_yolov8s).

Dragon Age Inquisition - Cullen Rutherford - Illustrious

V2 is my second attempt and to try softening the details with the generations by adjusting the training a bit. My testing was definitely smoother and had less of that rough/3d look V1 had. I also reduced the size of the LoRA. LoRA strength recommendations<ul><li>Use LoRA strength 0.7 with about CFG 3-5.</li></ul>

Files with SHA256: 190f66d95225b5c4b4340e8a11e35428082a6946e4ef100a52403fc27631d594