<p>This was originally a response to a bounty from PipaBC for a hair style called "Wolf Cut". I have trained up 3 versions (SD1.5, SDXL, PONY)<br />The style is reminiscent of the 90s grunge movement with a rough layered cut and usually bangs. Each versions prompts slightly differently to get the best effect and I will update this description with some of these examples later.</p><p></p><h2 id="sdxl-7aqmcrwj1">SDXL</h2><p>CFG and Steps - Same as your model of choice</p><p>To get the best effect from this DORA use a strength of 0.6+ (Anything seems to work well with this one)</p><p></p><h2 id="pony-j41xy3iuf">PONY</h2><p>CFG and Steps - Same as your model of choice</p><p>To get the best effect from this DORA increase the strength to 1.2-1.4</p><p></p><h2 id="sd1.5-tyn2ehxdy">SD1.5</h2><p>CFG and Steps - Same as your model of choice</p><p>To get the best effect from this DORA increase the strength to 0.6-0.9</p>

WolfCut - HairStyle [SD15, SDXL, PONY]

<p>Fixed up the dataset and adjusted some of the training settings to improve the results. There no longer seems to be a "Good" strength but anything 0.6 and up seems good. Supporting tags like "layered hair", "bangs", "disheveled"  seem to help the generation.</p>

Files with SHA256: 8406af7de8a620712d698debd97c1149ca9ea664cedcdb845eb1f7869aea3f97