DanDan is daydreaming: ink art style | Chinese painting | Tencent Huyuan-DiT

0
0
0
0
Chinese StyleAnime StyleGirlStyle Boost
Actualizado recientemente: Publicado por primera vez:
Chinese Style,Anime Style,Girl,LoRA,HunyuanDiT v1.2Image info
Chinese Style,Anime Style,Girl,LoRA,HunyuanDiT v1.2Image info
Chinese Style,Anime Style,Girl,LoRA,HunyuanDiT v1.2Image info
Chinese Style,Anime Style,Girl,LoRA,HunyuanDiT v1.2Image info
Chinese Style,Anime Style,Girl,LoRA,HunyuanDiT v1.2Image info
Chinese Style,Anime Style,Girl,LoRA,HunyuanDiT v1.2Image info
Chinese Style,Anime Style,Girl,LoRA,HunyuanDiT v1.2Image info
Chinese Style,Anime Style,Girl,LoRA,HunyuanDiT v1.2Image info
Chinese Style,Anime Style,Girl,LoRA,HunyuanDiT v1.2Image info

Dandan in a daze: ink art style enhancement | Chinese style black and white painting style | ink splashing rendering enhancement | Tencent DiT_V1.2 hybrid model

Model introduction: ink art style, black and white Chinese style, traditional Chinese ink splashing.

Creative overview: The creator has been very obsessed with Chinese painting style, ink rendering, etc. The more rich Chinese elements he sees, the more flavorful he feels.

Model usage instructions: The checkpoint film is based on the training of the Tencent DiT_v1.2 hybrid model.

When testing this small lora model, other two-dimensional large models, three-dimensional large models, and realistic human figure large models also test images, and the ink effect is very good.

Prompt word usage precautions: Trigger words are dandan, ink_art

However, the effect of scenes dominated by male characters has not been fully tested yet.

Other related parameter settings recommendations:

1. Sampling method: Recommended DPM++2M Karras, Eula, Eula A, SDE, with iteration steps of 20-30. Other sampling methods can also try LCM with iteration steps of ≤15, etc.

2. Enable high-resolution repair: For realistic human figure large models, it is recommended to use 8x-NMKD-Superscale, 4x-UltraSharp magnification algorithms; for two-dimensional, 3D large models, you can use R-ESRGAN_4x+ Anime6B, R-ESRGAN_4x+, and other magnification algorithms.

3. Test image resolution dimensions: 1:1 or 9:16 or 16:9 or 3:4 or 4:3, 512*512, 1024*1024; or 512*768; or 768*1024, 600*800, etc.

4. It is recommended to enable: After Detailer face retouch plug-ins, such as face_yolov8m, face_yolov8n, face_yolov8n_v2, etc.

Welcome all creators to use, provide feedback, and engage in friendly exchanges.

Discusión

Más popular
|
Más nuevo
Enviar
Próximamente
Descargar
(0.00KB)
Detalles
Tipo
Recuento de generación en línea
0
Descargas
0
Parámetros Recomendados
Combinación Recomendada de Checkpoint
Ninguno
Peso recomendado
0.8
CFG
7
VAE
Ninguno

Galería

Más popular
|
Más nuevo