3D Character - SD1.5

<<<Version 1.0>>>

Welcome back! Sample images are all generated using the SD1.5 large model <AWPortrait CG_v1.1> + 4090 local maps (N), local generation results may differ from online generation!

Recommended large models:AWPortrait CG_v1.1

https://www.liblib.art/modelinfo/2d0f035c95b74afc882dd805dc779066

Recommended LoRa weight:0.6-0.8 (recommended 0.7)

Trigger words: Not required

Recommended prompt words:

The content of the image you want (recommended to use natural language) + other supplementary element tags

For example: A female character wearing exquisite armor and a crown. Her armor is adorned with complex patterns and designs, primarily in silver, embellished with green gems. She has black hair and her eyes are a unique orange-yellow. Xilin holds a white lotus, standing in a sunlit forest surrounded by lush trees and green vegetation. Sunlight filters through the leaves, adding a mysterious and tranquil atmosphere to the scene.

(A female character wearing exquisite armor and a crown. Her armor is adorned with complex patterns and designs, primarily in silver, embellished with green gems. She has black hair and her eyes are a unique orange-yellow. Xilin holds a white lotus, standing in a sunlit forest surrounded by lush trees and green vegetation. Sunlight filters through the leaves, adding a mysterious and tranquil atmosphere to the scene.)

Welcome to add elements on your own or use cherished prompt words~

1. Its characteristic is that compared to AWPortrait CG's hyper-realistic style, it doesn't have as much skin texture and light details, belonging to a semi-realistic style;

2. When using, it is best to use a solid color background effect. (The training purpose is to produce characters that are easy to cut out, so the training set consists of solid/color gradient backgrounds, with no scenes. Further iteration will be made according to the situation~)

3. For specific LoRA effects, please refer to the example images and the sample return area!

Negative prompt word recommendations:

ng_deepnegative_v1_75t, (badhandv4:1.2), (worst quality:2), (low quality:2), (normal quality:2), lowres, bad anatomy, bad hands, ((monochrome)), ((grayscale)) watermark, moles

The above negative prompt words produce the best results! The example images are all generated using this negative prompt word.

Sampling method: DPM++ 2M Karras (feels like this will result in better hands)

Iterations: 30 (depending on your mood)

Recommended size: 3:4 (other sizes are welcome for experimentation)

Prompt word guidance coefficient (CFG): 7.0

High-definition resolution restoration: 4x-UltraSharp If there are images that meet the requirements, use high-definition enlargement, redraw amplitude0.3-0.4is sufficient, or directly redraw with an amplitude above 0.5 for card extraction~

<Attention>

There is a chance of problems with hands, body, and limbs during generation, multiple redraws may be necessary!

- Recommended direct use of 2x high-definition enlargement + 0.5 redraw extraction~

- When the character is full-body or smaller in proportion, facial collapse and blurring may occur. Please enable ADetailer and selectface_yolov8nfor restoration!

——————————————————————————————————————

Model usage statement:

1. You are not allowed to host this model and its derivative versions (such as fusion model versions) on websites/applications intended to earn revenue or donations. If you need to use this model and its derivative versions for commercial purposes (such as generative services, selling images, publishing images publicly on short video platforms or publications, etc.), please contact me through the platform's internal contact (for members, follow the membership rights of the Lib platform);

2. You are not allowed to directly sell images generated by this model unless you have made enough manual modifications to the image so that it can be fully considered your personal work in legal terms. If you violate this provision, you will be solely responsible for all legal consequences, and I shall not be held responsible (for members, follow the membership rights of the Lib platform);

3. You cannot intentionally use this model to create or share illegal or harmful content for dissemination and output. Please abide by public order and good morals and use this model for positive and constructive purposes.