<h1 id="palworld-fenglope-lora-for-qwensdxlflux">Palworld - Fenglope LORA for Qwen/SDXL/Flux</h1>This LORA is for the Fenglope species from the game Palworld.<h2 id="qwen-version">Qwen version</h2>Needs natural language prompting like Flux. The triggers that needs to be integrated into your prompt is<pre><code>fenglope creature</code></pre><h2 id="sdxl-version">SDXL version</h2>Please note: SDXL version was trained on a model that was deleted by the model creator so it might not be of much use anymore. The LORA is intended for use with highly realistic SDXL/pony models. I recommend the model CalypsoRealism (<a target="_blank" rel="ugc" href="https://civarchive.com/models/644327?modelVersionId=1078326">https://civarchive.com/models/644327?modelVersionId=1078326</a>). All example images (including the cartoony ones) were created with this model.I can not guarantee that it works well with other checkpoints. Usage:Trigger word is only<pre><code>Fenglope</code></pre>but some tags (it wasn’t actually trained on them) also have a positive impact on the result:<pre><code>blue fur, yellow horns, hooves, body markings</code></pre>The LORA is slightly overtrained for maximum likeliness. At full strength it may create some minor artifacts due to the overtraining. Personally it is not a problem for me, because I inpaint everything anyway. But if you want to avoid that just lower the strength a bit.The dataset was SFW only. But if you use it with the suggested checkpoint you can also create NSFW images with it.<h1 id="palworld-fenglope-lora-for-flux-dev">Palworld – Fenglope LORA for flux-dev</h1>I made a version of my Fenglope LORA for flux-dev!Usage:The trigger is<pre><code>f3ngl0pe creature</code></pre>but adding the following to your prompt can help to improve the results:<pre><code>has a cloud-like mane and branched golden horns, a horned f3ngl0pe creature with blue fur, the horned f3ngl0pe creature has black hooves</code></pre>Tested with flux-dev Q8Know issues: The form of the horns is sometimes not accurate

Files with SHA256: b71f57a9e17fb3eda2c71bdd5a3fdc12d084af5197be0d14d3bb22cfdaa8876e