Archive, page 9

Amazon Prime Video has introduced an AI-assisted lip sync feature that adjusts an actor's mouth movements to match dubbed audio. The tool is currently limited to the English dub of the German series Maxton Hall, with plans to roll it out to more titles over time.

ChatGPT's image generator has received an update aimed at producing results that more closely reflect the people and details present in source photos. PetaPixel put the model to the test by using it to transform a real person into a medieval knight. The results offer a practical look at how well the updated model handles likeness fidelity.

OpenAI is updating its image generation offering with ChatGPT Images 2.5, introducing two distinct models - Flare for faster outputs and Sunburst for more precise editing. The rollout appears uneven, with different users receiving different model access at different times. Early testing suggests the improvements are real but not universally available yet.

OpenAI has released ChatGPT Images 2.5, an updated image generation system that promises faster output and more accurate multi-turn editing. The update also introduces a Sketch feature, template support, and two new API models for developers. The changes build on the image capabilities OpenAI has been expanding within ChatGPT over the past year.

Lightricks has released LTXV, a video generation model designed for controllable, high-quality output. The model supports long-form clips, keyframe-based control, and advanced conditioning options. It positions itself as a tool for creators who need precise influence over the generation process.

Lightricks has released LTX-2, an AI video generation model aimed at professional and production workflows. It supports native 4K output, high frame rates, and extended long-form content generation. The model is designed with precise creative control in mind, targeting studios and creators who need reliable, consistent results.

Lightricks has quietly shifted LTX-2.3 into legacy status as LTX-2.5 takes over as the current recommended model. LTX-2.3 remains fully supported and still offers open weights, synchronized audio, and native portrait video generation. For teams already building on it, nothing breaks - but new projects are now steered toward the newer release.

Lightricks has released LTX-2.5, an open-weight video generation model capable of producing multi-shot scenes in a single pass. It also supports editing of real footage and exports in cinema-grade EXR format, making it a notable option for production-oriented workflows. The weights are available for fine-tuning and self-hosting.

Apple's Clean Up tool, introduced in iOS 18, gets a notable upgrade in iOS 27, making it easier to remove unwanted objects from photos directly on iPhone. The feature uses on-device AI to identify and erase distracting elements while filling in the background naturally. Engadget walks through exactly how to use the updated tool in the iPhone Photos app.

Runway has launched native plugins for Adobe Premiere Pro and After Effects, bringing its generative image and video tools directly into two of the most widely used editing applications. Users can generate media, restyle timeline footage, and place results into their compositions or sequences without leaving the Adobe environment.

OpenAI has introduced a drawing-based input tool called Sketch alongside the release of ChatGPT Images 2.5. Users can now draw a rough doodle directly inside the ChatGPT interface and use it as a visual reference for image generation. The feature is activated by typing @Sketch into the chat box.

Adobe is bringing generative video and audio capabilities directly into Premiere Pro's timeline with a new Generative Media Tool, announced at IBC 2026. Editors can describe what they need and generate context-aware video clips or sound effects without leaving their project, with support for multiple AI models including Google Veo, Runway, and Luma. The update also includes a range of smarter audio tools and new AI features for After Effects.

Blackmagic Design has released DaVinci Resolve 21.1, a substantial update that brings AI assistant integration to the editing workflow alongside more than 100 new tools and controls. The update also expands the software's photo editing capabilities and adds support for a range of new camera formats and log profiles. Python scripting, previously available in the free version, has been moved exclusively to the Studio tier.

Runway has announced that the team behind Kinetix, a Paris-based research lab specializing in 3D human motion and physically-grounded video generation, is joining the company. The move is aimed at advancing Runway's work on world models for robotics and expanding its research footprint in Europe.

Adobe is reworking how video editors access AI generation tools inside Premiere Pro, bringing video, audio, and music generation directly into the editing timeline. The update centers on a new Generative Media tool that lets editors fill gaps in their projects without switching apps or breaking their flow. The underlying generators are not entirely new, but their integration into the timeline is a meaningful shift in how the tools are used in practice.

OpenAI has introduced ChatGPT Images 2.5, an updated image generation model designed to produce more personalized and polished results from user inputs like text prompts, sketches, and reference photos. The update focuses on better translating a user's intent into the final output. It represents an incremental but meaningful step forward in how ChatGPT handles visual creation tasks.

Google's Gemini app is testing a new visual feature called Immersive View within its Students section, offering an image-driven way to explore topics through interconnected nodes. The interface appears to let learners navigate subjects - from dinosaurs to molecules - in a branching, non-linear format. The feature has been spotted in early testing by TestingCatalog.

When a restaurant's Facebook post featuring obviously AI-generated food images drew criticism in the comments, a familiar counter-argument surfaced: traditional photography is manipulated too, so what's the difference? The claim deserves more scrutiny than it usually gets.

IndieWire's Brian Welk sat on the jury for the Reply AI Film Festival, screening six hours of generative AI short films ahead of the event's showcase in Venice. His reflections offer a grounded look at where AI filmmaking stands today - its recurring patterns, its limitations, and the creative choices that actually stand out.

Google has introduced an agentic video understanding capability for its Gemini Flash models that selectively loads only the video segments relevant to a given prompt, rather than processing entire videos at a flat rate of one frame per second. The approach reduces video token consumption by as much as 88%, making long-form video analysis significantly more efficient. This marks a shift in how large language models interact with video content - from passive ingestion to active, query-driven navig

Runway has launched a Team plan aimed at small creative groups, offering shared workspaces, pooled generation credits, and Brand Kits for two to nine seats at $55 per seat per month. The new tier sits between individual subscriptions and larger enterprise arrangements, making collaborative AI video work more practical for studios and agencies working at modest scale.

Google has expanded Gemini Spark's capabilities to include direct management of Google Photos libraries, allowing the AI assistant to edit albums, create shared collections, and convert photos into calendar events. The feature is currently limited to AI Pro and Ultra subscribers. It marks a deeper integration between Google's AI assistant layer and its core photo storage product.

Instagram's "AI Content" label, designed to flag synthetically generated images, has been misfiring in noticeable ways. Users report that the tag is appearing on photos edited with ordinary tools like Canva's background remover, while genuinely AI-generated imagery continues to go unlabeled. The inconsistency is undermining trust in the system rather than building it.

Restaurants and food brands are increasingly using AI-generated images to promote their products, with results that range from odd to outright unsettling. Distorted noodles, lumpy burgers, and ice cream that looks more like brain matter are becoming a familiar sight. The underlying reasons come down to how diffusion models actually learn to represent food.