AWPainting 1.4

This iteration mainly focuses on enhancing scene diversity and picture tension. There has been a certain improvement in picture details and color saturation compared to 1.3, with a more diverse range of character facial types. Controlling the lens with vocabulary similar to Top view, perspective composition can lead to a better understanding and visual effect compared to 1.3.

Recommended resolutions: 512*768, 576*864, 640*960

For generating low-resolution images, you can use hire high-definition restoration to increase resolution, and the recommended algorithm is ESRGAN_4x.

AWPainting 1.3

Building on 1.2, this version improves the diversity of character facial features, supporting male, female, and various skin tones. It enhances the variety of character movements and coordination, optimizes lighting and overall shading effects, and has good natural language responsiveness (you can use GPT to write prompts for natural language).

AWPainting 1.2

Building on 1.1, this version enhances lighting and picture expression finesse, creating a more lifelike image and softer facial lighting. It has slightly improved responsiveness to prompts. For more test results comparing 1.2 and 1.1, please refer to X:@dynamicwangs.

AWPainting 1.1

It enhances image lighting effects and the variety of character movements, making facial features more distinct and recognizable. The t2i direct output accuracy has seen some improvement.

Paired with the newly released LoRA DynamicKungfu, it can effectively reproduce the style of some of my previous works.

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

AWPainting 1.0

This model has a wide applicability, suitable for illustrations, 2D, and generally 2.5D images.

Notably, it excels in reproducing the niji style, with excellent compatibility with LoRA nijiexpress and different LoRA combinations, resulting in high-quality direct output images, as shown in the example images, all t2i direct output + high-definition restoration (no controlnet).

Recommended with LoRA: Nijiexpress V1 V2, Xiaorenshu, Moxin, Qincai

Recommended Prompts: Fisheyes, masterpiece, top quality, best quality, official art, beautiful and aesthetic, animation

Neg_Prompts: EasyNegativeV2, ng_deepnegative_v1_75t, (low_quality:1.4), (worst_quality:1.4), (badhandv4:1.1), collage, artist_name, signature, artist_logo, watermark

Disclaimer:

The text content generated by this model service is maintained and used by you based on your independent judgment. You must use it in compliance with applicable laws and regulations and the terms of this service. You are solely responsible for any intellectual property issues arising from the content generated by this service. You are responsible for the generated content, and I am not liable for any losses incurred. Except as otherwise provided by laws and regulations, without my written authorization, you may not integrate this model service into any third-party software or services, nor may you disclose the association between the generated content and this service to any third party in any direct or indirect, explicit, or implicit manner. Without my written consent, non-individual users (including but not limited to enterprises, profit-making organizations) may not use the model for commercial purposes (including but not limited to text-to-image, image-to-image, face swapping, etc.). Should you violate this provision, I reserve the right to seek compensation for any losses incurred.