This LoRA was trained with the goal of recreating the artwork style of characters from the mobile game Naruto Ultimate Ninja Blazing. In particular, it focuses on recreating the visual effects surrounding the characters in these artworks (some example images are included below).<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/e78a9061-4da8-4ce5-9fd1-4f930eed3a30/original=true/e78a9061-4da8-4ce5-9fd1-4f930eed3a30.jpeg" /><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/4ebc366e-bb12-4ac9-af87-13b94251833b/original=true/4ebc366e-bb12-4ac9-af87-13b94251833b.jpeg" />This LoRA was mainly designed to be combined with other character LoRAs, giving them an artistic style and finish similar to the artworks from the game. However, it can also be used to create original characters, which will tend to have a style reminiscent of Naruto characters. When combining it with other character LoRAs, you may need to adjust the weights of both LoRAs to find the optimal balance.To be honest, the LoRA does not manage to recreate 100% of the visual details of the original artworks, at least not with the level of detail I originally intended. However, it has successfully learned the concept of the dynamic poses found in the original images and does generate various effects around the character, resulting in a fairly decent approximation.This LoRA is primarily a personal experiment to explore how well certain characteristics from different visual styles can be reproduced.<h2 id="trigger-words">Trigger Words</h2><ul><li>The main trigger word is "ultimate_ninja_blazing_artwork", although it is highly recommended to add "dynamic pose" to the prompt to encourage poses that more closely resemble the original artworks.</li><li>By default, the LoRA already adds the corresponding visual effects around the character, but they can be somewhat controlled through the prompt. You can specify their color, for example "blue effects, green effects...", or their type, such as "water, flames, electricity...".</li><li>The dataset intentionally included a variety of different characters, since the goal was not for the LoRA to learn the Naruto characters themselves, but rather the visual composition of these artworks. However, the LoRA ended up learning some common traits found across many of the images, such as the typical ninja sandals. Therefore, to prevent these features from being incorrectly assigned to characters who should not have them, you may need to make use of the negative prompt. For example, in the case above, adding "ninja sandals" to the negative prompt is recommended.</li><li>As a personal recommendation, it is a good idea to add "no background", "white background", or similar terms to the prompt if you want the background to be white or a solid color, while keeping the visual effects generated by the LoRA. This makes it easier to remove the background afterwards, as it will be white or another solid color.</li></ul><h2 id="recommended-generation-settings">Recommended Generation Settings</h2><ul><li>CFG Scale: 7</li><li>Steps: 20</li><li>Sampler: DPM++ 2M</li><li>Base Model: WAI-Illustrious-SDXL (Illustrious)</li><li>Hires Upscaler: R-ESRGAN 4x</li><li>ADetailer: Face and Hands</li></ul>

Files with SHA256: 9200f7673283c31b96920be245dd1a3453073bbe3c1017fc943e41967dfcb961