Sensemirage 194

SenseTime's "SenseMirage" is a creative platform that includes SenseTime's self-developed AI GC large model and convenient LoRA training capabilities, and provides accelerated inference for third-party community open-source models, offering creators more convenient and complete content production and creation tools. SenseMirage not only integrates SenseTime's self-developed text-to-image generation model with over 1 billion parameters and supports inference acceleration optimization for non-self-developed models, but also, combined with self-developed models and training capabilities, eliminates the need for localized deployment processes. Based on self-trained LoRA models, it efficiently generates more diverse content.

SenseTime SenseMirage is an AI large model-based text-to-image creative platform that enables users to quickly generate high-quality images by inputting text descriptions. SenseTime SenseMirage not only provides SenseTime's self-developed AIGC model, but also supports importing open-source models from third-party communities, allowing users to freely choose and switch between different styles and effects.

The SenseTime SenseMirage platform is an AI content creation community platform for various creators, aiming to enhance people's creativity through AI technology and build a diverse platform and community ecosystem. Users can browse, favorite, share, and comment on image works created by other users or teams on the platform, as well as participate in various themes and activities to communicate and collaborate with other creators.

SenseMirage Features:

1. Prompt completion feature: Users only need to combine a small number of prompts to leverage the product's powerful text understanding capabilities and extensive image generation styles to produce richly detailed image content.

2. Self-developed painting large model + community open-source models: Equipped with SenseTime's self-developed painting large model Artist, it also supports one-click import of open-source models from third-party communities such as HuggingFace, Civitai, and GitHub, eliminating the high costs and cumbersome processes associated with user-localized deployment. With diversified and convenient content production tools, it meets the painting needs of various styles.

3. LoRA training capability: Supports users to upload images and combine them with SenseTime's self-developed models or third-party models to train customized LoRA models. Users can quickly complete model fine-tuning through drag-and-drop operations to create their personal exclusive generative AI model and craft their own unique style.

SenseMirage Updates:

Starting September 14, 2023, users can register and use the platform by visiting https://miaohua.sensetime.com/. On October 8, 2023, the mobile version of SenseTime SenseMirage was launched, bringing users a more convenient experience, allowing them to create anytime, anywhere via their smartphones.

On October 8, 2023, SenseTime SenseMirage was officially upgraded to version 0.3.5, featuring a newly upgraded SenseTime self-developed painting large model Artist 0.3.5, which enhances the understanding of painting prompts. Additionally, the Aesthetic Guidance Algorithm 1.0 was introduced, capable of generating more artistic paintings with delicate content, rich layering, and better presentation of aesthetic texture details. Furthermore, the new version has been significantly optimized for anime styles and Asian portraits, improving image quality for a more comfortable viewing experience.

On November 7, 2023, SenseMirage's "Inspiration Square" released an update. The new version is divided into two sections: "Featured Images" and "Model Square," showcasing exquisite community artworks, and supports keyword search functionality to enhance inspiration and creation efficiency.