SuperAPI AI Directory
Partner with us中文AI newsFavorites
DirectoryNews and tutorials

NewBie-image-Exp0.1 - An Image Generation Model Open-Sourced by NewBieAi Lab

Author: AI小集 · Original source: AI工具集

What Is NewBie-image-Exp0.1

NewBie-image-Exp0.1 is an experimental image generation model open-sourced by NewBieAi Lab.Image generationmodel based on the Next-DiT architecture, with 3.5B parameters. The model was pretrained on the full dan + 1m e621 dataset and trained for four months on 8×h200 hardware, totaling approximately 23,000 h200 hours. It uses Google/Gemma3-4b-it and Jina Ai/Jina Clip v2 as text encoders, paired with Flux 1 Dev-VAE. NewBie-image-Exp0.1 focuses on high-quality anime image generation and can generate anime images with rich details and striking visual effects.

NewBie-image-Exp0.1

Main Features of NewBie-image-Exp0.1

  • High-Quality Anime Image Generation: Trained on a large amount of high-quality anime data, it can generate anime-style images with rich details and striking visual effects.
  • Powerful Architecture Foundation: It adopts the Next-DiT architecture with 3.5B parameters and possesses powerful image generation capabilities.
  • Efficient Training and Optimization: It uses high-performance computing hardware for long-term training, ensuring model performance and stability.
  • Flexible Text Encoders: It combines Google/Gemma3-4b-it and Jina Ai/Jina Clip v2 to improve the alignment between text understanding and image generation.
  • Potential for Multi-Scenario Applications: It is suitable for fields such as anime creation and game design, providing powerful tool support for creative professionals.

Technical Principles of NewBie-image-Exp0.1

  • Based on the Next-DiT Architecture: The model adopts the advanced Next-DiT architecture, a deep learning architecture designed specifically for efficiently generating high-quality images and capable of handling complex image generation tasks.
  • Large-Scale Data Pretraining: It is pretrained on the full dan + 1m e621 dataset, which contains a large number of high-quality anime images, enabling the model to learn rich image features and styles.
  • High-Performance Computing Support: It was trained for four months on high-performance 8×h200 hardware, totaling approximately 23,000 h200 hours, ensuring that the model is fully trained on large-scale data and improving generation results.
  • Text Encoder Fusion: By combining Google/Gemma3-4b-it and Jina Ai/Jina Clip v2 as text encoders, it enhances the model's understanding of text descriptions and generates images that better match text requirements.
  • Optimized VAE Configuration: It uses Flux 1 Dev-VAE for image encoding and decoding, optimizing the quality and detail expression of generated images and improving the overall generation effect.
  • Deep Learning and Generative Adversarial Networks: It utilizes deep learning technology, especially the principles of generative adversarial networks (GANs), and continuously optimizes the quality of generated images through adversarial training between the generator and discriminator, making them more realistic and diverse.

Project Address for NewBie-image-Exp0.1

  • Github Repository：https://github.com/NewBieAI-Lab/NewBie-image-Exp0.1
  • Huggingface Model Library：https://huggingface.co/NewBie-AI/NewBie-image-Exp0.1

Application Scenarios of NewBie-image-Exp0.1

  • Anime Creation: It provides anime artists with a powerful image generation tool to quickly generate high-quality anime characters and scenes, accelerating the creative process.
  • Game Design: It is used for character design, scene construction, and concept art creation in game development, helping designers quickly realize their ideas.
  • Digital Art: Supports digital artists in creating unique anime-style artworks, providing rich visual materials and sources of inspiration.
  • Illustration and Comics: Assists illustrators and comic artists in generating illustrations and comic sketches, improving creative efficiency and quality.
  • Advertising and Marketing: In the fields of advertising and marketing, used to generate attractive anime-style promotional images, enhancing visual appeal.
  • Education and Training: As a teaching tool, helps students and beginners learn anime drawing techniques, providing reference images and creative inspiration.