I present my attempt at a Illustrious character LoRA for Ardbert Hylfyst from FFXIV.<hr />V4 is my fourth attempt and to fix his face because I didn't like how his face shape and facial hair looked on V3 the more I looked at it. It definitely looks better but you may need to add beard to negative prompt because it occasionally likes to give him more facial hair than I trained for him to have. Also, If you leave out goatee tag and add facial hair to negative prompt and he'll have a smooth face since I decided to tag his facial hair (unlike V3). LoRA strength recommendations<ul><li>Use LoRA strength 0.7 with about CFG 3-4. <span style="background-color:rgb(26, 27, 30);color:rgb(193, 194, 197);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px">Also, I use an adetailer <span style="background-color:rgb(26, 27, 30);color:rgb(193, 194, 197);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px">in my generations for this LoRA.</li></ul>The tags for V4 are as follows:main tags<pre><code>ardbertffxiv, 1boy, hyur, very short hair, goatee</code></pre>note:outfit tags<pre><code>fur-trimmed collar, fur-trimmed pauldrons, armor, fur-trimmed gloves, tassets, fur-trimmed boots</code></pre>optional tag:<pre><code>axe</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 />V3 is my third attempt and to refine the quality in addition to reducing the training tag count a bit. LoRA strength recommendations<ul><li>Use LoRA strength 0.7 with about CFG 3-4. <span style="background-color:rgb(26, 27, 30);color:rgb(193, 194, 197);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px">Also, I use an adetailer <span style="background-color:rgb(26, 27, 30);color:rgb(193, 194, 197);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px">in my generations for this LoRA.</li></ul>The tags for V3 are as follows:main tags<pre><code>ardbertffxiv, 1boy, hyur</code></pre>outfit tags<pre><code>fur-trimmed collar, fur-trimmed pauldrons, armor, fur-trimmed gloves, tassets, fur-trimmed boots</code></pre>optional tag:<pre><code>axe</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 /><hr />V2 is my second attempt and to try refining the training so the results look a bit cleaner. I did see an improvement with my test generations, especially NSFW. LoRA strength recommendations<ul><li>Use LoRA strength 0.7 for SFW / 0.6 for NSFW with about CFG 3-5.</li></ul>The tags for V2 are as follows:main tags<pre><code>ardbert_ffxiv, 1boy, mature male, short hair, brown hair, blue eyes, goatee</code></pre>outfit tags<pre><code>fur-trimmed collar, fur-trimmed pauldrons, armor, fur-trimmed gloves, tassets, fur-trimmed boots</code></pre>optional tag:<pre><code>axe</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 /><hr />V1 is my initial attempt and produces rather nice results. LoRA strength recommendations<ul><li>Use LoRA strength 0.7 with about CFG 3-5.</li></ul>The tags for V1 are as follows:main tags<pre><code>ardbert_ffxiv, 1boy, mature male, short hair, brown hair, blue eyes, goatee</code></pre>outfit tags<pre><code>fur-trimmed collar, fur-trimmed pauldrons, armor, fur-trimmed gloves, tassets, fur-trimmed boots</code></pre>optional tag:<pre><code>axe</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).

FFXIV (Final Fantasy 14) - Ardbert Hylfyst - Illustrious

V2 is my second attempt and to try refining the training so the results look a bit cleaner. I did see an improvement with my test generations, especially NSFW. LoRA strength recommendations<ul><li>Use LoRA strength 0.7 for SFW / 0.6 for NSFW with about CFG 3-5.</li></ul>

Files with SHA256: a60359aa3db4b158c0a4b77d3f8b3e58d589f9884ba906c631f02456eeccbfeb