Search AI Products and News

CSGO is a brand new style transfer model developed by the InstantX team in collaboration with Nanjing University of Science and Technology, Beihang University, and Peking University. It aims to enhance image generation technology, particularly emphasizing the combination of content and style. The model supports three modes: 1) Synthesizing styled images from content images and style reference images; 2) Generating corresponding content-style images from style reference images and text prompts; 3) Editing specified objects in images through text. The core of CSGO lies in its unique data construction process, featuring designed data generation and automated cleaning.

Clapper is an AI generation and video editing tool developed independently by an individual developer, integrating various AI technologies including image generation, video creation, and audio. It offers APIs for image, audio, music, and video generation, allowing users to complete the entire process from creative conception to final production on a single platform. Clapper features a timeline function that supports content generation and editing orchestration, providing both a traditional interface and an infinite canvas mode to meet the needs of divergent thinking and precision editing. As an open-source tool, Clapper...

A study from the University of California, Berkeley, reveals that large language models significantly decrease the quality of images generated by DALL-E 3 after automatic prompt modifications, with a decline of 58%. Through online experiments comparing DALL-E 2, DALL-E 3, and DALL-E 3 with automatic prompt revisions, it was found that DALL-E 3 outperformed DALL-E 2 in technical capability and adaptability to user prompt strategies. The study noted that DALL-E 3 users achieved better image matching through longer, semantically similar, and more descriptive prompts.

PanoFree is an innovative multi-view image generation technology that can produce various panoramic images, including long images, 360-degree panoramas, and skyboxes, without the need for fine-tuning. Its core lies in addressing consistency issues and artifacts during the generation process through an iterative deformation and repair process while enhancing cross-view awareness. PanoFree employs techniques such as cross-view guidance, risk area estimation, erasure, and symmetric bidirectional guidance, combined with guided semantic and density control to ensure the preservation of scene structure. Experiments show that PanoFree achieves ...

MoeThread's AI creation platform 'MoBi MaLiang' announces a complete upgrade, providing users with a more efficient and personalized image generation experience. The upgraded MoBi MaLiang has particularly enhanced its understanding of Chinese language and culture, making creative portrait generation more precise. Users can generate personalized creative portraits quickly in just three simple steps: upload a photo, enter a description, and generate an image.

ViPer is a personalized image generation tool designed to interact with users to learn and apply their artistic preferences, similar to Midjourney. By analyzing user feedback on images, ViPer identifies preferences such as 'soft colors', 'diffused lighting', and 'dreamy atmosphere' and uses these as guiding principles when generating new images, avoiding elements that the user dislikes. This allows ViPer to create images that align with the user's aesthetic and reflect a personalized style, providing a unique artistic creation experience. The online experience can be found at: h

The KOALA AI model has been released and can run on 8GB memory, generating high-quality images in just 2 seconds. Developed by a South Korean scientific team, this new artificial intelligence image generation model uses 'knowledge distillation' technology, reducing parameters to 700 million, enabling image generation with only 8GB of memory. The generation time has been shortened to 2 seconds, making the KOALA model faster than OpenAI's DALL-E model.

Google Bard announced a global update today, expanding the Gemini Pro large language model capabilities to 40 languages and adding an image generation feature. This update is a significant step in Bard's development, making it easier to use and more competitive. Gemini Pro is Google's most advanced LLM to date. Bard ranks second in the Chatbots Arena leaderboard.

The aMUSEd model launched by Hugging Face can generate images in seconds, faster than competitors like Stable Diffusion. aMUSEd utilizes a lightweight text-to-image model based on Google's MUSE model. It adopts a Masked Image Model (MIM) architecture, reducing inference steps and enhancing generation speed and interpretability. The model is available in a demo on Hugging Face.

The Alibaba Tongyi Laboratory open sources the AI image generation fine-tuning framework SCEdit. The SCEdit framework supports fine-tuning for image generation tasks and demonstrates high efficiency in generation tasks. It achieves precise control in controllable generation tasks while saving training memory costs.

The WHEE mobile app from Meitu has officially launched, providing one-stop AI visual creation services. Powered by the Miracle Vision 4.0 model, WHEE offers AI art and image generation features. Users can easily use natural language to describe their needs. The WHEE mobile app brings together works from creators in various fields, serving as a source of inspiration for creativity. Additionally, WHEE provides creation guides and model training instructions to assist users.

Blogger @ZhanMeiLi has sparked the Douyin platform with photos generated using AI image expansion. In the realm of AI image expansion, internet users have unleashed their creativity, producing an array of innovative works. While AI image expansion has garnered attention, there are still limitations regarding the authenticity of images. Social media has seen a surge of creative works that blend character depictions with imaginative scenes. In the future, as technology continues to develop, AI image expansion is expected to excel in a wider range of application scenarios.

At the re:Invent conference, Amazon introduced the Titan Image Generator, a significant advancement that offers customers legal protection options. The Titan Image Generator is part of Amazon AWS, designed for enterprise developers, supporting text generation and image editing with built-in protection mechanisms. Unlike other consumer-oriented image generators, Amazon focuses on the enterprise market to mitigate bias and incorporates invisible watermark identification for AI-generated content. Amazon's AI strategy enters a new phase, providing legal safeguards, with expectations in the fields of image generation and application.

AI-generated white faces are now so realistic that they appear 'more real than real', yet they exhibit bias in the representation of other ethnicities. The research found that biases in the algorithm's training data lead AI to favor white features when depicting other ethnic groups, potentially exacerbating online racial bias. The researchers urge public education to prevent misinformation and online scams, highlighting the potential issues brought by AI's 'hyper-realism'.

The research team has proposed a novel training method to enhance the image generation capabilities of DALL-E3. This method integrates synthetic and real titles to address spatial perception and text presentation issues. Advanced language models like GPT-4 play a key role in improving text quality and depth. The research shows significant improvements in the quality and accuracy of DALL-E3's image generation. This strategy lays the foundation for the continued development of future text-to-image generation technologies.

After hard work, Midjourney has launched a new version of its website, improving image search speed and functionality. The new site offers faster speed and stability, eliminating lag issues. A new lighting mode allows users to easily create realistic scenes. The added lightbox feature presents works more intuitively, promoting appreciation of details and creative exchange. A personal search feature enables users to conveniently find works they are interested in.

Tencent Hunyuan announces the launch of its image generation feature, which has strong Chinese understanding capabilities and can create images in various styles, including landscapes, characters, and animations. It has been integrated by over 180 businesses and will play a role in advertising, gaming, and content creation.

👀 Exploring Applications of Large Models: The article explores the differences in image understanding and generation between GPT-4 and DALL-E 3 through generative adversarial tests. 💡 Limitations of Large Models: Cyclical testing reveals that large models have limitations in making definitive judgments when encountering ambiguous images. 🔎 New Methods for Testing Large Models: Using cyclical testing methods allows for an in-depth exploration of large models' abilities in image-text understanding.

Tencent Cloud showcased its new AI painting product in Shanghai, which is based on an in-house developed AI painting model. It now supports over 25 generative styles that accommodate Chinese elements in image generation, such as the understanding of ancient poetry and traditional ink wash paper-cutting. The product performs excellently, with API response times reduced by 50%. Users can flexibly adjust the image characteristics and styles according to their needs.

📷 Image Filters: OpenAI trained a proprietary image classifier to detect suspicious patterns in images to prevent the generation of disturbing content. 📷 Cultural Bias: DALL-E 3 still has cultural biases, often leaning towards Western culture, especially in ambiguous prompts. ⛔ Copyright Issues: OpenAI acknowledges that despite all safety measures, it is impossible to predict every combination, and copyrighted material might appear in generated images.