Game character IP enhancement | Sky: Children of the Light Rainforest Elder Lora

0
0
0
0
AnimeGirlGame Character DesignPhysical/Facial feature
Cập nhật gần đây: Đăng tải lần đầu:
Anime,Girl,Game Character Design,LoRA,SD 1.5Image info
Anime,Girl,Game Character Design,LoRA,SD 1.5Image info
Anime,Girl,Game Character Design,LoRA,SD 1.5Image info
Anime,Girl,Game Character Design,LoRA,SD 1.5Image info
Anime,Girl,Game Character Design,LoRA,SD 1.5Image info
Anime,Girl,Game Character Design,LoRA,SD 1.5Image info
Anime,Girl,Game Character Design,LoRA,SD 1.5Image info
Anime,Girl,Game Character Design,LoRA,SD 1.5Image info
Anime,Girl,Game Character Design,LoRA,SD 1.5Image info
Anime,Girl,Game Character Design,LoRA,SD 1.5Image info
Anime,Girl,Game Character Design,LoRA,SD 1.5Image info
Anime,Girl,Game Character Design,LoRA,SD 1.5Image info
Anime,Girl,Game Character Design,LoRA,SD 1.5Image info

******************* Version and Training Set Information *******************

This model uses the characteristics of the Rainforest Elder hairstyle + silver headgear as the character features. It is the second model in my Sky series and the first single-character training model in my Sky series. The training set images are from original game screenshots and some game screenshots that have undergone partial Photoshop processing.

The base model is hello25D Anime Official Version ---> https://www.liblib.art/modelinfo/c48f2af3c95e4f149954503b61ccac52

【Trigger Words】 yuma (optional)

【Training Set】 152 images | Steps: 22800 | Trainer: Autumn Leaf Integration Version lora-scripts-v1.8.3

【Disclaimer】

1. In principle, this model is intended for exchange and learning purposes only, and must not be used for commercial purposes.

2. Users should use this service in compliance with the laws and regulations of their country and the terms of service; any intellectual property disputes arising from the content generated based on this model are the responsibility of the user.

3. Without the author's authorization and consent, this model service must not be integrated into any third-party software or service.

4. When publishing generated images on social media, attribution to the model and author must be stated.


******************* Usage Suggestions *******************

【Base Model Usage Suggestions】

1. More suitable for anime and illustration-style base models, not very effective when combined with realistic hand and action styles; use a base model that matches the style of what you want to draw.

2. Well-matched models include:

hello series (helloFlatCute2D, helloFlatArt, helloFlatColorful2D_V15e, helloip3DIP, helloKid2D, helloYoung25D, helloWonderland, etc.), arthemyComics_v60, aniverse_v16Pruned, aZovyaRPGArtistTools_v4, cuteyukimixAdorable_kemiaomiao, DDJECX-1.20, pastel_anime, UIA插画 | Flat新国潮大模型, seekyou_mikan, RLhew, 甜蜜童话, 儿童插画绘本, 墨幽二次元/25D

3. Models that lean towards realism with good results include: 麦橘奇幻, 麦橘写实

* The above base models are some recommendations; you can also try based on your preferences~

【Lora Weight Suggestions】

1. Lora weights can be adjusted for darker faces. Since there is no data available for this hairstyle, a high weight setting may result in capes being more prominent.

2. Based on test images:

0.5 and below, retains facial features; 0.3-0.4 is suitable for drawing other headgear or hats, but retains the braid characteristics;

Above 0.5 results in a darker face; Avoid exceeding 0.8 as it may affect the flexibility of the character's movements;

* Different base models may have slight variations in weight compatibility; starting at 0.5 is recommended; below 0.5, it's advisable to use ADetailer for better results.

【Parameter Settings】

Image Dimensions: 768*768, 1024*1024, 768*1024, 900*1200, etc. (proportions not too long or too wide recommended)

Sampling Method: Euler a, DPM++ 2M Karras, and other commonly used samplers

Sampling Steps: 24-30

Clipping Skip: 2

Redrawing Amplitude: 0.2-0.5

CFG Ratio: 7

VAE: Use as needed

Thảo luận

Phổ biến nhất
|
Mới nhất
Gửi
Đến sớm
Tải xuống
(0.00KB)
Chi tiết
Loại
Số lần tạo hình ảnh trực tuyến
0
Tải xuống
0
Tham Số Đề Xuất
Kết Hợp Đề XuấtCheckpoint
Không
Cân nặng được khuyến nghị.
0.45
CFG
7
VAE
Không

Bộ sưu tập hình ảnh

Phổ biến nhất
|
Mới nhất