Miracle Plus Size Youth Oversized Younger

Miracle - Plus Size Youth Oversized Younger LoRa XL Version File Name: Miracle - Plus Size Youth Oversized Younger LoRa XL_v20240225.SafeTensors

“LoRa Vol.3,” which is used in conjunction with the “AiArtist-CADS Commercial Advertisement Assistance System

XL Version”, is here.

Bottom model link: https://www.liblib.ai/modelinfo/a56ebacdba7d4e30b97bb124bc3fc28f Very used

Simple. Select a youth clothing style according to your needs, use style words to trigger a color scheme, and add elements in the picture to create a refreshing youth clothing style. I'm looking forward to your ideas!

Trigger words: No trigger word

style color scheme

(few references):

black oversized short sleeve t-shirt and white shorts black oversized short-sleeved tshirt and white

shorts

beige oversized short sleeve shirt and brown pants beige oversized short sleeved shirt

and brown pants

dark blue oversized short sleeve shirt and beige pants dark blue oversized short-sleeved shirt and beige pants

white oversized short sleeve t-shirt and jeans white oversized short sleeve tshirt

and jeans

pink oversized short sleeve t-shirt and white shorts pink oversized short-sleeved tshirt and white

shorts

“It is recommended to use the sample image hint directly, then add modified details as needed”

Recommended Lora weight: 0.8 to 1, recommended number of generation steps: 25 to 35 CFG

scale: 5 to 8 Recommended sampling method: DPM++

2M Karras,

Welcome to try other sampling methods. FP8 inference can be plotted normally. It supports LCM/Turbo/Lightning Lightning accelerated plotting with DPM++ SDE /CFG:2/generation steps 4-8 >

Click Load <

Recommended Resolution: Any resolution greater than or equal to 1024 on the long side, It is recommended to set the long edge locally to 1536. It is recommended that the result be straight

out; there is no need to select high-definition magnification in the inference interface.

Image post-processing high-definition settings: 8x_nmkd-superscale_150000_g can be superimposed

; 4x-UltraSharp or R-ESRGAN 4x+ Anime6B accelerated generation:

1. Using Liblib online generation, you can obtain more stable image quality, reduce the complexity of deployment and download, and efficiently match the platform base model.

2. sdxl-Lightning is a lightning-fast text-to-image generation model, Lib online generation is available > Click Load <

It can produce high-quality, high-resolution images in 4-8 steps.

For more information, see ByteDance's research paper:

SDXL-Lightning

: Progressive Adversarial Diffusion Distillation Progressive Anti-Diffusion Distillation link

https://www.liblib.ai/modelinfo/33b8201621f94809965b684d12ba5c47

3. This LoRa is suitable for

SDXL large model: AiArtist-CADS commercial advertising assistance system XL version or other SDXL large model links

https://www.liblib.ai/modelinfo/a56ebacdba7d4e30b97bb124bc3fc28f

————————————————

—

——————————————————————————— If you have any questions about testing, please leave a message on the Lib model page, business cooperation+V, please see the homepage Non-human | Metaverse Digital Human/AIgC/Metaverse Direction