PonyXL version of a previously released NAI LoRA. It's trained on the same six shots of <a target="_blank" rel="ugc" href="https://imgur.com/a/eDL0f7c">official character art</a> but much like its predecessor it can generate other outputs good enough to play around with.How to use Just prompt away, any fantasy/RPG theme generally work. Unless otherwise specified it's usually "bottom centric" focusing on thighs. As the dataset shows the faces are smaller than normal compared and can come out less-than-great. Prompting specific facial features or using an extension like <a target="_blank" rel="ugc" href="https://github.com/wkpark/uddetailer">µ Detection Detailer</a> can help with this if you don't want to do your touching finishes in img2img/outpainting. The samples images are a mix with it on and off. Note that sampler will change the style output, Euler a is good for more clean outputs and DPM++ 2M Karras brings out the "crayon" look better.For more prompt tips please see the sample images or the uploaded zip with training tags.Misc Trained on Pony SDXL at 1024 resolution with a dataset of six images. All images are untouched txt2img save for a few using uDetailer, no other inpainting or helper embeddings used. E.g. - things should look better if you're running more than the bare minimum. The training was done on a small subset of images, as such the output of this LoRA is not an accurate representation neither in quality or style of the source material. Consider buying the official merchandise if you want the good stuff.

Files with SHA256: 76bf35bb3ac27f9d45176e9d005b2b7cee0032a0485592218cbc514964417ff9