<a target="_blank" rel="ugc" href="https://patreon.com/psoft"><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/8dcbc56a-15e9-429e-9b36-c2bc22594678/width=525/8dcbc56a-15e9-429e-9b36-c2bc22594678.jpeg" /></a> Support my work on <a target="_blank" rel="ugc" href="https://www.patreon.com/psoft">Patreon</a> and <a target="_blank" rel="ugc" href="https://ko-fi.com/psoft">Ko-Fi</a> and get access to tutorials and exclusive models. Rate and leave a like if you enjoyed it, and follow for new models!<hr />LoRA trained with images of character Miyamoto Rei (High School of the Dead)<h3 id="heading-3">v1.5 (NEW)</h3>Increased dataset, modified tags, cleaned up tags, used new and improved training parameters. This results in a both more consistent and flexible LoRA. It works with more models and can be combined with other LoRAs. The example images combine with another character LoRA.Additional infoI make these models for fun and for the challenge and release them for free for everyone to enjoy, for non-commercial use (see license). The one thing I ask of you is to provide rating and feedback! That helps me understand where I can improve and helps me stay motivated to produce more content!I accept suggestions or requests here or on Civitai discord server.You can also check my more polished images and upcoming models on my <a target="_blank" rel="ugc" href="https://www.pixiv.net/en/users/25545070">pixiv </a>profile.-- previous versions --v1.5New training parameters, tags and dataset. Trained on NAI. Tested on <a target="_blank" rel="ugc" href="https://civarchive.com/models/4451/abyssorangemix2-hardcore">AOM2</a>, <a target="_blank" rel="ugc" href="https://civarchive.com/models/8281/perfect-world">Perfect World</a>, <a target="_blank" rel="ugc" href="https://civarchive.com/models/10028/neverending-dream-ned">NeverEnding Dream</a>, and other models. Works with 1.0 weight and 0.3-0.7 CFG, but depends on the model.You don't need a trigger word to get the character. For the basic uniform you can simply use the trigger "serafuku". You can also add to the positive or negative prompt to get variations from the tagged training data: shirt, skirt, thighhighs, elbow pads, knee pads, sneakers, rifle, stick, blood, torn clothes, shoulder holster.See the information in the image examples, including for tested models.Works with multiple models and LoRAs.v1.0Better results with 0.5-0.7 weight and 0.3-0.6 CFG (v1.0), but it may depend on the model used.Works with realistic and anime models and to work with other characters and styles (see examples on how to achieve this). Clothes can be changed by adding, removing or reordering these tags:"green sailor collar, black bow, green skirt, black thighhighs, long sleeves, brown hair, brown eyes, hair between eyes, ponytail, sneakers, white footwear"(tip: don't describe the shoes if you don't want a full body image)See the version information for more on the LoRA training. Please post your images and let me know your feedback.

Miyamoto Rei (HOTD) - Character

Trained with NAI.54 images, not cropped. Manually tagged.10 repeats, 20 epochs, batch size 30.LR: 0.0002, Unet LR: 0.0002, Text encoder LR: 0.0002Net dim/alpha: 8

Files with SHA256: 2e68dc457e8f3525a8107dbf5551af73f9c2021bb0134cded0d7e821ed1e139c