This is a collection containing multiple variants for a Character LoRA. See Description for more information. This version was previously called SangoV3Rev2. If you like my work and would like to leave a review, I mean.. feel free to. I do this for purely for fun - nothing else.<hr />Trained on <a target="_blank" rel="ugc" href="https://civarchive.com/models/221751?modelVersionId=634653">The Deep Dark v6.1</a> Dataset is from Inuyasha: The Final Act. Backgrounds removed. Trained on site. 4440 steps (148 images @ 20 epochs, 6 repeats, 4 batch size). Potentially overtrained. Start at lower weights and make your way up. By this I mean a lora weight of 0.5, cfg scale of 5. See trained model page and gallery for others generally.<hr /><h3 id="prompt-usage-dbqczd1en">Prompt Usage</h3>The trigger word is Sango. I didn't use unique tags for clothing. Here are two prompt examples. Hair color, breast size, eye color, hair style can be modified. Try 'giant boomerang' for Hiraikotsu. Untested. Kimono Outfit<pre><code>Sango, 1girl, solo, brown eyes, brown hair, black hair, long hair, low-tied long hair, makeup, bangs, kimono, pink kimono, skirt, green skirt, anime coloring, retro artstyle</code></pre>Armor Outfit<pre><code>Sango, 1girl, solo, brown eyes, brown hair, black hair, ponytail, makeup, bangs, black bodysuit, shoulder pads, anime coloring, retro artstyle</code></pre>Be sure to make use of quality tags, body tags, expressions. Have some decent negatives. You can reference my gallery for examples. Anything else I didn't mention is probably not worth mentioning.

Sango (Inuyasha: The Final Act) | Pony | The Deep Dark v6.1 | SangoV3

This variant is the milestone where I tried to improve my datasets and stuff. Trained on The Deep Dark v6.1. Much less steps. Backgrounds replaced with simple backgrounds. Trigger Words are basic examples. You will need to make use of body angles, backgrounds and more to make them look better. See Description for more information.

SangoV3Rev2.safetensors

SangoV3.safetensors

SangoV3

This is a collection containing multiple variants for a Character LoRA. See Description for more information. This version was previously called SangoV2.5. If you like my work and would like to leave a review, I mean.. feel free to. I do this for purely for fun - nothing else.<hr />Trained on <a target="_blank" rel="ugc" href="https://civarchive.com/models/221751?modelVersionId=469243">The Deep Dark v6.0</a> Dataset is from Inuyasha: The Final Act. Tagged backgrounds. Dataset consists of 188 images. Trained on site, used 25 epochs, 7 repeats, 4 batch size or 8225 steps. It is overtrained. As a result of overtraining, bath scenes are dark. Weight/Setting Suggestions (Estimate)<ul><li>Lora Weight: 0.5 ~ 0.8</li><li>CFG Scale: 5 ~ 10</li><li>Sampler: Euler A, DPM++ series (unsure)</li><li>Steps: 20 ~ 40</li><li>Denoise: 0.5 ~ 0.75</li><li>Resolution: 512x512 (use upscaler/hires)</li></ul> See the gallery for prompt guidance. I have provided some x/y/z plots in the gallery, should give you an idea.<hr /><h3 id="prompt-usage-0jik04z7m">Prompt Usage</h3>The trigger word is Sango. I didn't use unique tags for clothing. Here are two prompt examples. Hair color, breast size, eye color, hair style can be modified. Hiraikotsu wasn't tagged. Kimono Outfit<pre><code>Sango, 1girl, solo, brown eyes, brown hair, black hair, long hair, low-tied long hair, makeup, bangs, kimono, skirt, green skirt, anime coloring, retro artstyle</code></pre>Armor Outfit<pre><code>Sango, 1girl, solo, brown eyes, brown hair, black hair, ponytail, makeup, bangs, armor, shoulder pads, anime coloring, retro artstyle</code></pre>Be sure to make use of quality tags, body tags, expressions. Have some decent negatives. You can reference my gallery for examples. Anything else I didn't mention is probably not worth mentioning.

Files with SHA256: 82b5dab2d3c39a317e933b0a4c0f4e88cc367e5b595dae3b1e165609418044e0