Major Update V5.0!! The overall visual quality is even more realistic, and the characters are more beautiful. Full of a girlish vibe, so we named her Qingli

We hope everyone enjoys it, use it more often!!

V5.0 is not available for download yet, please join the fan group for access!

Version V3.0 fixed compatibility issues with some filters and corrected a bug with two people appearing under certain samplers.Recommended weight values 0.3-1, feel free to test it out!

Special facial enhancement training has also been carried out to make characters more lifelike and improve the effect.qq group number: 833273550

Welcome everyone to use it! Enjoy!

Version V2: Recommended weight values 0.6-0.8

v2 added some elements to it, each version has its own characteristics

Go ahead and give it a try!

TikTok 63223961125 Looking forward to your follow

Version V1: Recommended weight values 0.6-0.8

Recommended to use the filters from Moyu Daren's humanoid series ormajicMIX realistic

The image results are quite good

Welcome everyone to try it out!

Feel free to join the Q group for discussions!

【Model Introduction】The inspiration for this model was from a very satisfying facial structure noticed during image processing. To solidify this facial appearance, the idea of creating a model called Lora emerged. A batch of photos with various angles and types were carefully selected from the generated model training images. After several version adjustments, a version that I was satisfied with finally emerged, hence the name MoLi was given to her.

【Generation Suggestions】1. If using the filters from Moyu Daren's humanoid series, sugges* ***ng the DDIM sampler, steps 28-50, keep CFG value at 7, base resolution: 512*768, scaling algorithm4x_NMKD-Superscale-SP_178000_G, redraw scale: 0.25, scaling factor: 2

2. If using other large models, sugges* ***ng the DPM++2M KARRAS sampler, steps 28-35, keep CFG value at 7, base resolution: 512*768, scaling algorithm4x_NMKD-Superscale-SP_178000_G, redraw scale: 0.3, scaling factor: 2

【Creative Overview】This model aims to give the characters a more youthful girlish feel and make them more realistic. Enhancing the photography of virtual characters with a youthful and dynamic feel. To make the characters more realistic, some girlish elements were added to the training images, enhancing the model's realism.

【Test Images】

Filter: Moyu Artificial Human V1040, LORA Weight 0.8     Filter: Moyu Artificial Human V1040, LORA Weight 0.6 Filter: Moyu Artificial Human V1040, LORA Weight 0.4

Filter: magicmixrealistV6, LORA Weight 0.8   Filter: magicmixrealistV6, LORA Weight 0.6   Filter: magicmixr LORA Weight 0.4