Archive, page 10

Restaurants are increasingly turning to generative AI to produce food imagery for their menus, but customers are picking up on something being off - even if they can't always articulate what. The homogenizing tendencies of AI image models mean that dishes end up looking idealized and interchangeable, stripping away the character that makes food photography feel honest. The gap between expectation and reality can leave diners disappointed before they've even ordered.

Google has expanded Gemini Spark's capabilities to include direct integration with Google Photos, giving AI Pro and Ultra subscribers a way to manage their photo libraries through natural language prompts. The update covers editing, organizing, and sharing photos without needing to navigate the app manually. It marks a notable step in bringing conversational AI into everyday personal media management.

Midjourney is asking its user community to submit ideas and feedback to help shape what the team works on next. A formal voting session is planned within the next week or two, giving users a direct say in the platform's development priorities.

ByteDance has released Seedream 5.0 Lite, a lighter edition of its text-to-image model that incorporates deeper reasoning to improve generation accuracy. The update focuses on more precise prompt following and compositional fidelity without the full resource footprint of its larger counterpart. It represents a continued push by ByteDance's Seed team to make capable image generation more accessible and efficient.

ByteDance's Seed team has introduced Seedream 5.0 Pro, a new image generation model that positions itself around design comprehension rather than pure visual output. The model aims to bridge the gap between generating images and understanding the underlying intent of design tasks. It represents an evolution in how the team is framing the role of generative models in creative and professional workflows.

Adobe has integrated more than 70 tools from its creative suite - including Photoshop, Illustrator, and Premiere Pro - directly into Slack. The integration lets users invoke creative actions through Slackbot without leaving their workflow. It marks a notable step toward embedding professional creative tooling into everyday collaboration software.

Eddie AI has unveiled an Agent Marketplace featuring editing agents modeled on the body of work of notable filmmakers and editors, including Walter Murch, Thelma Schoonmaker, Roger Deakins, and more than two dozen others. Users can also build their own custom agents to enforce brand guidelines, legal checklists, or personal editing rules. The initial rollout includes 29 agents available through the Eddie AI platform.

AI-generated content is no longer confined to social media feeds - it's turning up in job applications, product reviews, and insurance claims, raising serious questions about authenticity online. Pangram Labs CEO argues we are approaching a tipping point where the web's foundational trust is at risk. A wave of startups is now working to build detection tools capable of separating human-created content from machine-generated output.

Camera Intelligence has introduced Caira, a Micro Four Thirds mirrorless camera that incorporates generative video editing directly into the hardware. It is the first mirrorless system to bring this kind of AI-native capability on-board, removing the need for post-production software to apply generative edits. The camera positions itself as an end-to-end production tool for videographers.

Runway has launched Runway Dev MCP, a hosted Model Context Protocol server designed to integrate Runway's AI capabilities directly into developer workflows. It allows coding agents to research models, configure routing, and troubleshoot failed generations without switching out of the editor environment.

The Republican nominee for New York governor has released an AI-generated campaign video depicting Democratic figures Zohran Mamdani and Kathy Hochul in a way that critics are calling unsettling. The clip reportedly mimics the soft, warm aesthetic of a prescription drug advertisement. It is one of the more notable recent examples of AI-generated imagery entering state-level political advertising.

World Labs has introduced Atlas, a single AI model capable of generating, reconstructing, and simulating 3D scenes from only a handful of input images. Unlike specialist models that treat inputs as flat sequences, Atlas anchors everything in 3D space from the start. The company also highlights its potential for producing synthetic training data for robotics.

ByteDance's Seed team has unveiled Seedance 2.5, a video generation model focused on one-take creation and flexible referencing capabilities. The update aims to give creators more control over how source material - such as characters, styles, or objects - is carried through generated video. It represents a continued push by ByteDance to mature its generative video offerings for practical production use.

Google is rolling out agent-based video analysis across several Gemini Flash models, letting the model choose which parts of a video to examine rather than processing every frame at a fixed rate. The approach cuts token consumption by up to 88 percent while reportedly improving accuracy, particularly on longer recordings. It marks a meaningful shift in how multimodal models handle video input at scale.

Google has been quietly approaching major Hollywood studios about licensing deals that would let it train AI models on copyrighted film and television content, offering substantial payments in return. While the arrangement looks straightforward on paper, the risks and rewards are not evenly distributed between the two sides. The studios may have more to lose from these agreements than the upfront cash suggests.

Google has added agentic video understanding to its latest Gemini models, enabling more accurate, multi-step reasoning over video content at lower token cost. The capability allows Gemini to handle longer, more complex video tasks autonomously rather than requiring repeated user prompting. Google says the update also reduces overall token usage.

Miro's brand team used Runway to produce the keynote intro video for its Canvas26 event, localizing the content across four global markets without any live-action shooting or stock footage. The project cut production time roughly in half compared to traditional methods. It offers a concrete look at how an in-house creative team is integrating generative video into a real production workflow.

Google has introduced Google Pics, a prompt-driven design tool that positions itself as an AI-first alternative to platforms like Canva and Adobe. Rather than working with templates and drag-and-drop elements, users describe what they want and let the AI handle the visual composition. The move signals Google's intent to carve out space in the consumer and professional creative software market.

Google DeepMind has introduced agentic video understanding capabilities within Gemini, allowing the model to actively analyze and reason over video content rather than passively responding to prompts. The update positions Gemini as a more autonomous tool for extracting meaning from visual and temporal information in video. This marks a step toward AI systems that can take initiative when working with complex, time-based media.

Visko has introduced Orbis, a video generation model capable of producing 4K output in real time. Unlike most generative video tools that render clips in full before delivering them, Orbis accepts prompt changes while the stream is still running. The combination of high resolution and live interactivity marks a notable step in how AI video generation can be used in practice.

Google has introduced Pics, a suite of creative design tools built into Workspace that lets business users edit and generate images using AI. Built around Google's Gemini and the Nano Banana generative model, Pics offers more targeted control over image creation than typical chatbot interfaces. The tool is aimed at filling a gap where AI image generation has struggled to deliver consistent, usable results for professional and marketing contexts.

Google is expanding access to its Pics image editor, making it available to subscribers of its AI Pro and Ultra plans. The move brings a dedicated AI-powered editing tool into Google's premium subscription tier, giving paying users another reason to stay within the Google ecosystem for creative work.

Google has launched Pics, a new image creation and editing tool integrated into Google Workspace. Built on the company's latest Nano Banana model, the tool aims to make generating and modifying images straightforward for everyday users within the apps they already use. Here is what you need to know about the new offering.

DxO has released PhotoLab 10, an update to its flagship photo-editing application that places a notable emphasis on AI-assisted masking and RAW image processing. The new version also brings color grading refinements aimed at tightening the overall editing workflow. Key additions address shortcomings from earlier releases while introducing new tools for more precise image control.