Type | |
Stats | 98 0 |
Reviews | (16) |
Published | Nov 6, 2024 |
Base Model | |
Training | Steps: 1,709 Epochs: 34 |
Usage Tips | Clip Skip: 1 |
Hash | AutoV2 F13FC84688 |
The human form of Shinigami from Rain Code.
Read first
This was part of an experiment in my workflow to see how many corners I could cut to speed up the process. I didn't put nearly as much work into this as I usually do with loras. The results were adequate enough for me to upload, but poor enough that I'm not repeating it. I consider this to just sit in the 'ok' range. It's solid enough to upload, but I felt it needed a disclaimer as well.
Prompt style
This lora was trained in narrative, rather than tag, style. You can go with the narrative promps using the examples below.
Prompts
Standard outfit: "Shinigami is a young woman with purple eyes. She has two long pink and white braids held by small white bows. She has bangs that sweep to the left, attached by a black flower hairpin. She wears a black crown with pink and yellow ornaments and has two black horns on her head. She wears a form-fitting short black dress with a rose pattern. Her dress exposes a large amount of her cleavage. She has a large bow tied at the back of her dress. Shinigami wears black knee-high socks and large silver tipped boots."
Bikini: "Shinigami is a young woman with purple eyes. She has two long pink and white braids held by small white bows. She has bangs that sweep to the left, attached by a black flower hairpin. She wears a black crown with pink and yellow ornaments and has two black horns on her head. She wears a two-piece swimsuit in purple and black colors. Her swimsuit exposes a large amount of her cleavage. The top is a ruffled bandeau-style design with a sweetheart neckline, featuring light purple fabric with darker black trim. The bottom piece is purple and black with small bows on the side."
Additional prompt options
Scythe: "Her hand is holding a scythe."
Styles
The style options really didn't come out very strongly. However, you can try the following.
Game style (Mimics the shaded in-game graphics): "Rain Code game style."
Anime style (Specifically the art style of the game's artist, Rui Komatsuzaki): "The image is in the anime art style of Rui Komatsuzaki. The image's style highlights large expressive eyes and exaggerated hairstyles. The image has a vivid and diverse color palette. There is a high level of detail in the image which include intricate patterns on clothing, accessories, and background elements. The image blends cute and charming elements with darker, more sinister themes. The image has dynamic and dramatic poses. The image features a striking blend of vibrant colors and highly expressive character designs."
Flat Rui Komatsuzaki style (AKA a more Danganronpa look): "Danganronpa sprite style."
Prompts - Tags
As I mentioned, this lora was an experiment in a different (lazier) methodology. One that didn't work out as well as I'd hoped. Thankfully there's a fair chance that any model based on Illustrious will have some basic understanding of the traits associated with Shinigami which we can leverage to push things in the right direction. You can do a full tag-based prompt for her with this lora using the following.
Tags: "master_detective_archives:_rain_code, shinigami_(rain_code), highres, 1girl, black_dress, black_gloves, black_horns, black_nails, braid, breast_tattoo, breasts, cleavage, detached_sleeves, dress, fingerless_gloves, gloves, horns, long_hair, long_sleeves, nail_polish, neck_ribbon, pink_hair, purple_eyes, ribbon, symbol_in_eye, tattoo, twin_braids, white_ribbon"
Additionally, if you're using the narrative prompt and any elements aren't coming out properly you can try combining the tags "master_detective_archives:_rain_code, shinigami_(rain_code), highres" and missing element tags with the narrative prompt.
For example, if her hair wasn't coming out the right color you could try this combination of narrative and tag prompt: "master_detective_archives:_rain_code, shinigami_(rain_code), highres, (((pink_hair))), Shinigami is a young woman with purple eyes. She has two long pink and white braids held by small white bows. She has bangs that sweep to the left, attached by a black flower hairpin. She wears a black crown with pink and yellow ornaments and has two black horns on her head. She wears a form-fitting short black dress with a rose pattern. Her dress exposes a large amount of her cleavage. She has a large bow tied at the back of her dress. Shinigami wears black knee-high socks and large silver tipped boots."
Additional notes
I'm a little annoyed that I don't have her spirit form in here. However, the primary motivation for this lora was to try out a new workflow and judge its viability on the results. I wanted to keep the number of variables and complications to an absolute minimum so decided to only work with the human form of Shinigami rather than two different forms within a single lora.