MADE BY CREATORS · LIVE ON VEYR
You don't need to know how to make it. You just need to start.
Read StoryThe live pulse of AI entertainment: creator posts, prototypes, workflows and fast-moving breakthroughs.
See all Signal
Pokeable creator @0xRhinoX used an AI coding model to build a tiny browser adventure where the landscape breathes, snores and occasionally opens one enormous eye.
Niklas Lenz built a real-time strategy experiment where language models manage armies, resources and battlefield decisions entirely on their own.
Filmmaker Dave Clark helped his eight-year-old son turn a family story into a short film using Claude, Dreamina and Seedance 2.5.
Divesh Naidoo's new Reality-Aware Rendering demonstration uses a phone performance to guide AI-generated characters inside a real environment.
The creator puts himself into AI-generated performances where the hands and musical movements matter as much as the visual style.
Chong-U combined Retro Diffusion 3D, Blender and AI coding models to create a colorful browser game built around deliveries, exploration and online multiplayer.
League of Spawn is a playable multiplayer battle arena with heroes, spells and evolving battlegrounds. Its newest map, Dragonscrest, adds an enormous new objective to the fight.
Sac combined Claude Opus 5.5, Tripo and Blender to build an interactive adaptation of the beloved story, complete with miniature planets to explore.
Mohammed Motawea built a playable bioluminescent roguelite in eight days, directing AI-assisted code and art around a wonderfully strange creature-grafting mechanic.
An indie creator is using Claude Code and Unreal Engine to turn a 2D prototype into a working 3D farming and colony simulation.
Epic's CEO describes a company-wide push for faster development and wants Unreal Engine teams to benefit from what Epic learns.
Henry Daubrez and a small team have released the third chapter of their fantasy adventure, using an evolving AI animation pipeline to bring a larger world and returning characters to life.
AI game development gets much more useful when it starts handling the work around the game—not only the art and code players see.
Making animation cheaper solves only part of the problem. Studio Jadu is trying to give small creators the pieces that normally come after production too.
Studios are starting to look for filmmakers who already know AI production tools, while schools such as NYU, USC and others are adding those tools to creative training.
The fourth annual competition is open through October 9, giving AI filmmakers a three-minute horror brief with $12,000 in cash prizes.
Lucca Cerf’s latest work is compelling because the character does not stop at a generated model. His pipeline carries it through rigging and animation until it starts to feel like a believable game character in motion.
Gemini 3.8 Live pushes conversational AI toward persistent on-screen characters that can see, listen, speak and remain visually present while tools run in the background.
For AI-built 3D, motion coherence may be a more revealing benchmark than static visual quality.
High-end agentic game creation can now produce striking interactive worlds—but today's economics can still be extreme.
Deeper production methods, multi-tool processes and creator case studies.
See all workflows
Working with Claude Opus 5.5, a fleet of coding agents and ideas from his audience, Blade built a 3D browser battle royale in three weeks—and brought the community together to play it.
YouTube is moving generative AI directly into the editing process: creators will be able to tell Gemini what changes they want in normal language and then continue editing manually.
His fantasy short combines GPT-6 Astra, Runway generation and a conventional edit. The useful lesson is choosing which tasks to hand over.
Tyler Smith's new workflow uses an AI-edited reference frame to transform existing footage without filming the scene again.
Layout to Render lets creators stage the action before using generative video to explore its finished appearance.
Mike Houser says he built the fantasy ARPG without knowing how to code, using AI to turn an idea into a substantial playable game in roughly two weeks. Now he says a publisher wants in.
View original source
Instead of releasing another AI animation demo, Hanif published the rigs, textures, rendering tools and frame checks behind it — and the films are rendered with code rather than a video model.
View original source
The producer's Higgsfield workflow turns early visual ideas into motion tests, helping collaborators react before the team finishes a shot.
View original sourceImportant new models and updates for creators.
Built from MiniMax H3, the new model combines rapid generation, up to 12 creative references and separately editable audio for more hands-on filmmaking.
Give it a rigged asset and a text prompt, and UniMate can generate motion across radically different skeletons without training a new model for each one.
Longer AI-generated shots are exciting. Keeping the characters, action and camera under control from beginning to end is what filmmakers really need.
Anthropic’s new model arrives with lower token prices and stronger coding claims. For creators, the useful test is what survives the next revision.
The complete VEYR news archive.
Epic is rebuilding gameplay around Verse and Scene Graph, with AI-assisted migration tools intended to help developers leave familiar systems behind.
New developer plugins let coding agents inspect projects, choose models, build media workflows and diagnose failures without repeatedly sending developers back to an API dashboard.
The Akuna Group and Curious Refuge are creating a 12-week AI filmmaking program that leads to project pitches and potential production funding.
Pixelmator Pro gets Generate Fill, Logic Pro gains an intelligent assistant, and Final Cut Pro expands its finishing toolkit.
Filmmaker Dave Clark helped his eight-year-old son turn a family story into a short film using Claude, Dreamina and Seedance 2.5.
New guidance for Claude Code's Opus 5.5 aims to separate original game development from recreating a specific existing title.
Its new on-device AI engine gives developers tools for character memory, dialogue, behavior and generated content without relying on cloud inference.
Niklas Lenz built a real-time strategy experiment where language models manage armies, resources and battlefield decisions entirely on their own.
Pokeable creator @0xRhinoX used an AI coding model to build a tiny browser adventure where the landscape breathes, snores and occasionally opens one enormous eye.
Tyler Smith's new workflow uses an AI-edited reference frame to transform existing footage without filming the scene again.
Divesh Naidoo's new Reality-Aware Rendering demonstration uses a phone performance to guide AI-generated characters inside a real environment.
EditAgent uses specialized editing methods and visual evidence to improve complex image edits—and professional retouchers preferred its approach to repeated refinement.
Built from MiniMax H3, the new model combines rapid generation, up to 12 creative references and separately editable audio for more hands-on filmmaking.
Google DeepMind's new world model generates interactive environments, sounds and characters that respond as people explore—and lets them change the scene along the way.
The creator puts himself into AI-generated performances where the hands and musical movements matter as much as the visual style.
The 1 Billion Followers Summit is offering a major prize for a five-to-seven-minute film, with specific rules about AI production, originality and rights.
With Bonn Park and Sarika Hemi Lakhani, the Run Lola Run director is transforming live performances into the larger world of Volcano High using Seedance 2.5.
You don't need to know how to make it. You just need to start.
Chong-U combined Retro Diffusion 3D, Blender and AI coding models to create a colorful browser game built around deliveries, exploration and online multiplayer.
The 2026 Epic AI MegaJam invites developers to build playable Unreal Engine and Fortnite experiences, with unlimited AI agents and continuation grants of up to $50,000.
Alaya Lab's video-to-video research takes primitive gameplay footage and reimagines its appearance, while measuring whether characters and objects remain where the game put them.
Digital Sun wants players to help shape its new Dracula game. But concerns about AI-assisted copycats have convinced the studio to replace a broad Steam test with a smaller community event.
Drunken Goddess Reflux was heading into Steam Next Fest. Its creators have replaced the public demo with a closed playtest, worried that rapid imitation could expose their distinctive game before launch.
The new Directional Sprites tool tackles a familiar headache for 2D game developers: keeping a character's appearance consistent as it turns, walks and fights.
Working with Claude Opus 5.5, a fleet of coding agents and ideas from his audience, Blade built a 3D browser battle royale in three weeks—and brought the community together to play it.
Google DeepMind researchers gave four Gemini 4 Argon agents a shared goal, no assigned roles and a survival world full of setbacks. They eventually won, after writing 2,783 programs and dying 605 times.
Muse Spark Open brings an unusually powerful AI coding model to developer-controlled hardware, with a demonstration of how an agent can examine footage and help shape an edit.
Japan’s Actors Union has proposed new rules that would give performers control over AI training, synthetic dialogue and multilingual dubbing, while opening the door to paid voice licensing.
League of Spawn is a playable multiplayer battle arena with heroes, spells and evolving battlegrounds. Its newest map, Dragonscrest, adds an enormous new objective to the fight.
A voice recorded in an ordinary room can now be adapted for a beach, a moving car or a noisy factory. For filmmakers, the interesting part is how the picture helps shape the sound.
A fantasy chess demonstration from Gemini 4 Argon gives game creators something more interesting to look at than another AI benchmark: an interactive world.
Sac combined Claude Opus 5.5, Tripo and Blender to build an interactive adaptation of the beloved story, complete with miniature planets to explore.
Microsoft is assembling a team to investigate AI-assisted game ports, browser versions and remixes—and what those experiments could mean for the future of play.
An indie creator is using Claude Code and Unreal Engine to turn a 2D prototype into a working 3D farming and colony simulation.
Mohammed Motawea built a playable bioluminescent roguelite in eight days, directing AI-assisted code and art around a wonderfully strange creature-grafting mechanic.
Bloomberg reports a sweeping change to code and art production on an unannounced Marvel and Star Wars shooter. Tim Sweeney says Epic is reorganizing around AI-accelerated development.
Epic's CEO describes a company-wide push for faster development and wants Unreal Engine teams to benefit from what Epic learns.
Henry Daubrez and a small team have released the third chapter of their fantasy adventure, using an evolving AI animation pipeline to bring a larger world and returning characters to life.
Reallusion and Escape.AI are bringing 3D animation, generative video and film distribution together, giving independent creators more control over the stories they make.
Anthropic's new animation tool turns presentations, charts and ideas into editable motion graphics, with Runway available for more advanced video finishing.
The Return of Illunes runs an AI game master directly on the player's computer, interpreting unexpected actions while traditional game systems handle combat and progression.
Playground turns everyday descriptions into playable games, giving people a way to create, experiment and share without learning to code.
Reverie Movies has introduced Kubrick, a conversational filmmaking assistant that can review scenes, reorganize shots and help creators turn generated footage into a finished film.
Unity's new AI creation platform combines natural-language direction, a professional game engine and artist-created assets to help more people build ambitious interactive experiences.
Easy Fox's voice-controlled driving game lets players coach an unpredictable little sister. The studio says its free demo now costs more than $1,000 a day in AI services.
Quadify tackles the mesh cleanup and asset preparation that sit between an attractive 3D object and a usable game asset.
Kero: The Loud One turns neighborhood favors into a small 3D adventure, ending with a party.
The 3D creation service connects a character’s shape, rig and movement, giving artists an inspectable route from an image to an animated asset.
The filmmaker has acknowledged the technology while declining to identify its specific applications.
Codex and ModRetro connect AI-assisted game creation to Game Boy Color-compatible hardware.
Nikolay Shestak’s unsettling premise won the $450,000 Grand Prix at the inaugural Astana AI Film Festival.
Layout to Render lets creators stage the action before using generative video to explore its finished appearance.
The REX project aims to modernize RE Engine as games become larger and AI enters development workflows.
Neilson Koerner-Safrata’s open-source project gives Claude Code several specialized generation tools to coordinate.
An ambitious fan project demonstrated agent-driven development, then went offline after its creator reported a legal demand.
An experimental desktop turns real-time video into lessons, visualizations and worlds that respond to interaction.
Maxon’s new connection turns conversational requests into operations artists can inspect, adjust and undo.
Kimodo isn’t new. What is interesting now is seeing creators turn NVIDIA’s text-and-constraint motion model into something that sits inside an everyday 3D workflow.
Mike Houser says he built the fantasy ARPG without knowing how to code, using AI to turn an idea into a substantial playable game in roughly two weeks. Now he says a publisher wants in.
Give it a rigged asset and a text prompt, and UniMate can generate motion across radically different skeletons without training a new model for each one.
A new $500 million investment in CMU’s computer-science school includes GPUs, servers, storage and expanded access to AI tools. For student creators, that matters more than the size of the donation.
Making animation cheaper solves only part of the problem. Studio Jadu is trying to give small creators the pieces that normally come after production too.
AI game development gets much more useful when it starts handling the work around the game—not only the art and code players see.
The next useful AI filmmaking tool may be the one that manages the space between individual generations.
The company’s growth announcement arrives alongside a mobile app that puts more of its 3D workflow on a phone.
The interesting thing about an always-on AI isn’t that it can complete one more task. It’s that it can stay responsible for the project.
The interesting part isn’t that AI generated some shots. It’s that one agent was trusted with enough of the production process to turn a loose creative brief into an edited film.
AI is supposed to give individuals the production power of a studio. But as creative workflows shift toward metered agents, tokens and generation credits, the people with the least capital risk being priced out.
The next model war may not be won by the smartest AI. It may be won by the model that gives creators the most intelligence they can actually afford to use.
Instead of releasing another AI animation demo, Hanif published the rigs, textures, rendering tools and frame checks behind it — and the films are rendered with code rather than a video model.
The new motion-design workflow can generate assets, build and animate a composition, render the result and leave creators with a normal After Effects project they can keep changing.
Better AI-video prompting sometimes starts with something less futuristic: knowing what the shot you want is actually called.
Longer AI-generated shots are exciting. Keeping the characters, action and camera under control from beginning to end is what filmmakers really need.
The important idea here is editability. Higgsfield is trying to let an agent do production work inside the tools creators already use—and leave behind a project humans can keep working on.
ElevenLabs is pushing text-to-speech beyond simply reading lines correctly. V4 is designed to understand how those lines should actually be performed.
Sparknify selected 32 shorts from more than 3,000 submissions, then put human-made and AI-generated films into a blind final judging stage focused on cinematic quality and emotional impact.
Gemini 3.8 Live pushes conversational AI toward persistent on-screen characters that can see, listen, speak and remain visually present while tools run in the background.
AI is sharply lowering the cost of video production in China, but exploding supply is making discovery and audience attention more valuable.
Particle 6 is bringing veteran filmmaking leadership into an AI-assisted feature pipeline, with Rob Minkoff joining early enough to shape story, characters and screenplay rather than merely endorsing a finished experiment.
Danse Macabre shows a more controlled use of generative AI: Hisko Hulsing’s team used AI texture and stylization passes to support live-action reference, oil painting, 3D animation and manual finishing rather than asking a model to author the film.
Studios are starting to look for filmmakers who already know AI production tools, while schools such as NYU, USC and others are adding those tools to creative training.
Seedream now combines image creation, reference blending and editing inside one model in Higgsfield.
Varkos uses AI as part of the game itself rather than simply using AI to create the game's art.
A 51-minute AI-created horror film is attracting a real audience, which is more meaningful for entertainment than another viral ten-second clip.
The Dor Brothers are known for visually ambitious AI filmmaking, but their production process isn't simply about generating spectacular images. The team describes creating many candidate shots, keeping a small selection and doing the creative work needed to make those pieces belong together.
AI filmmaker Nawod Madhuwantha is tackling a familiar frustration: generating a nearly perfect shot, spotting one mistake and having to recreate everything to fix it.
Combos is trying to turn AI game generation into a social ecosystem where users create, play and remix one another's games.
Claude is beginning to perform production animation tasks inside Blender instead of merely explaining how to do them.
Creator interest around Astra video workflows appears unusually high.
YouTube is expanding AI beyond video creation and into the decisions that happen before and after publishing. New tools can review drafts, suggest improvements and help creators experiment with how their work reaches an audience.
Creator @ushangul has shared a production approach that combines AI-assisted writing, video generation and editing into a repeatable process for YouTube Shorts.
CineMe is developing a workspace where filmmakers can turn a written screenplay into a shared visual plan, connecting creative decisions with the practical work of preparing a production.
Inkitt is shifting the AI-video problem from generating individual clips to managing scenes, reusable assets, revisions and longer-form assembly.
Lucca Cerf’s latest work is compelling because the character does not stop at a generated model. His pipeline carries it through rigging and animation until it starts to feel like a believable game character in motion.
Peter Yang used Claude Opus 5.5 to create a recognizable 3D scene of the Golden Gate Bridge entirely through code. In his own experiments, he found the model competitive with GPT-6 Astra for building 3D environments.
AI models can already generate impressive 3D scenes. The more important question now is whether those scenes can survive production: closer inspection, new camera angles, material changes, object edits, and repeated iteration without forcing an artist to rebuild the work.
Kevin Ngo used Claude Opus 5 to create animation frames through JavaScript, demonstrating another way individual creators can experiment with motion without relying entirely on traditional animation tools.
Adobe has completed its acquisition of Topaz Labs, the company known for improving video and images through AI. The deal brings more attention to an essential part of creative production: making footage look good enough to deliver.
YouTube is moving generative AI directly into the editing process: creators will be able to tell Gemini what changes they want in normal language and then continue editing manually.
High-end agentic game creation can now produce striking interactive worlds—but today's economics can still be extreme.
Smart Video shows AI video moving from isolated generations toward agent-driven production, revision and longer-form assembly.
Katzenberg thinks AI could put serious production capability into the hands of far more creators — if they help shape the rules around how it is used.
OpenAI’s shutdown reaches the Videos API and Sora 2 models. Exporting old work and replacing a live integration need separate attention.
Danny Limanseta is experimenting with a wonderfully simple game idea: begin as a small monster, grow stronger and eventually become the biggest threat in the city.
Curious Refuge is exploring a filmmaking approach that begins with an actor rather than an AI prompt. The team films the performance first, then uses image and video models to transform the world around it.
Anthropic’s new model arrives with lower token prices and stronger coding claims. For creators, the useful test is what survives the next revision.
Vee Camallere is using reusable characters, locations and color treatments to keep an AI-assisted series consistent.
Mirage's Tesseract is designed to let creators direct a video edit through an AI agent while keeping the underlying project editable.
R2 turns generative video into a running interactive system that can preserve state and react to player input in real time.
The September 28–30 online jam requires a portrait browser game on itch.io. Test the export before the theme arrives.
Five finalists emerged from more than 2,500 submissions pairing a trailer with a written treatment. They screen September 25 in Los Angeles.
Submissions close November 20. Entrants can use other tools, but AKOOL must play a meaningful part in the finished work.
Theoretically Media demonstrates how Dreamina's AI assistant can help filmmakers prepare a production instead of jumping straight into generating the finished video.
Agora-2 shows world models moving beyond generated scenes toward shared, persistent systems where people and AI agents act inside the same evolving simulation.
TypeSafe AI's Jev points toward a different role for AI in games: deciding what to do next from predefined options instead of generating open-ended text.
Searchable footage, a studio-trained model and licensed character voices solve different problems. For creators, the first question is which job each tool can help them finish.
Former DeepMind researchers are reportedly seeking a large seed round. Creators still need a product they can test.
Flow Studio’s 3D Editor and Canvas demonstration preserved scene elements that artists could revise individually.
System 3 says it read Antstream’s description of human-finished artwork as a denial of generative AI use.
The association reports that 85.8% of surveyed game developers use generative AI at work. A separate company survey asks how that use is managed.
Particle6 and Xicoia are exploring what happens when an AI-created character can respond to people directly instead of appearing only in pre-rendered scenes.
Independent developer freesoulapps is experimenting with something more ambitious than generating individual 3D models: building a functioning science-fiction RPG where players can explore, meet characters and complete quests.
LIVE 404 imagines a horror stream where the audience doesn't simply watch events unfold. Their comments guide a virtual streamer exploring frightening locations.
One Solution Games describes how a two-person team used AI assistance to build a Unity game, then had to become more disciplined as the project grew.
Qwen combines generation and editing with transparent output and more reference control. Its downloadable weights carry a research-only license; commercial use requires separate permission.
A filmmaking tutorial explores how a short, carefully chosen source clip can produce more useful alternate angles than trying to regenerate an entire scene.
Pocket FM is developing Sherpa as a creative assistant for serialized fiction, helping writers plan episodes and keep track of the characters, relationships and events that make a long-running story work.
A Variety review of Bahman Ghobadi's film praised its performances but argued that some generated imagery disrupted the experience.
Karandeep Anand is due to start October 2. His brief spans data and AI platforms, infrastructure, product and engineering.
At its September event, TheGrill brought together creators working across filmed performances and generated imagery.
Kunchevsky says AI took routine code off his hands and helped him try more versions of the game he had in mind.
A 90-second AKOOL film-challenge entry uses a sword fight to explore betrayal, with an emphasis on physical action rather than spectacle.
His fantasy short combines GPT-6 Astra, Runway generation and a conventional edit. The useful lesson is choosing which tasks to hand over.
Developers can convert clips up to five minutes long to a selected frame rate, including 120fps.
After criticism of VISION 2026 II, the studio promised corrections and limits on AI in videos shown to players. It still sees a role for AI behind the scenes.
A Shanghai court ordered an AI voice service to stop infringing conduct and pay 750,000 yuan. The voices, game imagery and way the packs were sold all mattered.
The event explored what happens when image, voice, video, world-building and agent systems have to hand work to one another.
Veteran Nintendo artist Takaya Imamura is using AI to turn visual game concepts into a prototype he can test and revise.
Its rules make AI disclosure mandatory and say it will not lower a film’s score. The free regular deadline is September 30.
Tripo highlights creators pairing its 3D asset tools with AI coding and scene-building assistants to build interactive environments.
Its launch announcement describes separate tracks and adjustable graphics. The useful test comes when a client asks for a small change.
The developer shared a browser prototype after repeated feature additions, bug fixes and performance work.
Most AI models are trying to make something. Jev is built for a different job: making quick decisions inside games, software and creative pipelines.
Franklin AI is demonstrating one of the most useful applications of AI-generated 3D assets: taking the models out of a preview window and putting them to work inside a game.
The fourth annual competition is open through October 9, giving AI filmmakers a three-minute horror brief with $12,000 in cash prizes.
WordFusion is an element-combination puzzle that uses Google Gemini when players try combinations the developer hasn't written in advance.
Independent creator az9713 used Claude through Higgsfield's Blender Bridge to construct a short 3D animation. The most interesting part wasn't the finished scene, but the mistakes identified and corrected along the way.
The monthly Hollywood festival advertises a filmmaker Q&A after every screening. Its next listed date is October 18.
The September 17–19 conference dates have passed. Its published schedule remains a guide to the people and subjects on the program.
CTO DongKi Lee says the studio takes the criticism seriously, while still seeing a role for AI in development.
Applications stay open year-round. The company offers tool access and possible paid projects, with selection based on existing work.
The San Francisco event pairs creative-production sessions with world-model research and robotics. For filmmakers, availability and control are the details to listen for.
Tamar AI launched a streaming destination for AI-made films and series, connecting its TapNow creation tools with distribution.
The producer's Higgsfield workflow turns early visual ideas into motion tests, helping collaborators react before the team finishes a shot.
Adobe’s Generative Media tool brings video and sound creation into the timeline, where the result has to fit.
Robert Gaudette revisited a story he had never been able to film and used AI production tools to make it for the Reply AI Film Festival.
The Featherweight creator used CapCut's reusable grading tools to preserve decisions that episodic filmmaking otherwise has to make again and again.
An OpenAI case study shows how one playable foundation became a test bed for several visual directions.
Google says Beast Industries is using Gemini for brainstorming and planning challenges that stretch well beyond writing a video script.
Google’s agentic mode revisits relevant passages instead of sampling the whole video at one fixed rate.
Cinema Studio 4.0 lets creators choose how often a generated sequence cuts, alongside its camera, lighting and reference settings.
Wonder Studios' second Beyond the Loop season pairs six filmmakers with support to turn ambitious concepts into finished shorts.
In a collaborative Reddit game, players propose new buildings and vote on which generated preview becomes part of their shared town.
Seedance 2.5 introduces targeted video edits. The useful test is what happens to the parts you wanted to keep.
For AI-built 3D, motion coherence may be a more revealing benchmark than static visual quality.
Roblox says many people using its simpler Build experience had never worked in Studio, suggesting a different path into game creation.
.webp)
ElevenLabs has released a free audiobook of Homer's epic with an authorized synthetic Michael Caine narration, character voices, music and effects.
ElevenLabs is building approved character voices with performers and specialists, while Hasbro retains review of their use.
ElevenLabs’ model supports more than 90 languages, but its in-app output does not have an edit button.
The ElevenLabs partnership offers personal likeness templates and a separate route for commercial projects.
Ask, Plan and Agent modes give developers different ways to use project context in the editor.
Roblox's Cube research is exploring 3D objects whose separate parts can behave together in a game.
Made something exceptional with AI? We want to see it.
Submit