AltNext - Interactive Experience Summit

@Soulmotionlabs @runwayml Makes sense, but in traditional pipelines control comes from separating stages like storyboard, layout, animation. Here it’s all in one system. Curious if that means small changes force you to rework everything instead of tweaking one layer?

@AltnextSummit @runwayml Yes essentially. The beauty of these new models is you can still create and use traditional storyboards/animatics etc. We need as much control as we can get for ai to be a proper tool for animation. It's definitely pretty much there.

Seedance is absolutely insane for animation. Made in @runwayml Full Prompt: 15-second cinematic animated short, premium stylized 3D feature-film quality, charming rounded character design, expressive faces, snappy timing, fluid body motion, polished bedroom environment, warm natural window light, soft bounce lighting, realistic cloth and hair motion, shallow depth of field, cinematic lensing, playful magical tone, clean composition, strong silhouette posing. character: @hero boy SHOT LIST WITH TIMING 0.0 - 3.0 sec Medium shot. Interior bedroom. 8-year-old boy sitting on bed, arm stretched forward, palm open toward a single pen on desk in front of him. He strains hard trying telekinesis. Face tense, eyes squinted, body shaking with effort, subtle bed bounce. Audible grunts. Camera slowly pushes in on boys face. 3.0 - 5.0 sec Cut to wide shot of full room. Bed on one side, desk with a pen on it. Posters, books, scattered toys, lived-in kid bedroom. Boy continues forcing powers toward pen. Curtains flutter slightly. Camera locked with slight handheld energy. Pen in foreground with the boy out of focus in the background. 5.0 - 7.0 sec Cut to close-up of boy’s face. Extreme concentration. Teeth clenched, cheeks puffing, eyebrows compressed. He gives one final powerful grunt. Tiny facial tremble, comedic over-effort. 7.0 - 8.5 sec Cut to macro close-up of a pen on the desk. Silence beat. Then the pen nudges forward only a few centimeters with tiny scrape sound. a small movement. 8.5 - 11.0 sec Cut back to medium shot of boy. He notices tiny movement but interprets it as failure. Expression falls. Long disappointed sigh. Arm drops limp. He flops backward onto bed dramatically, mattress compresses and rebounds. He stares at ceiling. 11.0 - 15.0 sec One beat of stillness. Then sudden fast crash zoom backward from boy, through bedroom, smashing out the window in one continuous camera move. Pull rapidly into huge exterior reveal: the entire house is floating 20 feet above the ground, gently hovering in the air above the street of houses which are on ground level. Trees and yard below. Subtle debris drifting downward. House tilts slightly. Golden sunlight. End on wide heroic comedic reveal. STYLE NOTES Fast readable cuts, strong anticipation and payoff, expressive animation arcs, clean eye lines, comedic timing, whimsical realism, smooth camera transitions, polished render quality. NEGATIVE PROMPT text, watermark, logo, extra limbs, creepy face, broken anatomy, jittery motion, flicker, noisy render, dark scene, horror tone, low detail, stiff animation, warped room geometry, duplicate objects, camera stutter. #seedance #AIart️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️ #madewithrunway

@so_ainsight Most of the reactions focus on materials, but the bigger issue is structure. If the mesh can’t survive retopo or rigging, isn’t this locked to still renders and product shots?

これ、ガチで時代変わる。 Microsoftが「写真1枚から3Dモデルを作るAI」を無料公開しました。名前は TRELLIS.2。 写真を1枚渡すだけで、 ・金属やガラスの質感までそっくり再現 ・コップの中の空洞、葉っぱの薄さ、穴あきの服もそのまま再現 ・速いモードなら3秒、高精細でも1分 しかも個人でも会社でも、自由に使ってOK。 ゲーム素材も商品モックも、写真さえあれば作れる時代です。 深掘り続きます👇

@icreatelife @T1000_V2 When you combine multiple models like that, are you thinking in terms of roles (one for visuals, one for motion), or is it still trial and error?

@T1000_V2 I asked it to use Nano Banana 2 and Veo 3.1 but if you don’t specify it will choose it for you.

Adobe just dropped Firefly AI Assistant in public beta I immediately turned my day into a video game One conversation. No manual edits. The future of creative tools is here 🎮

@Kashiko_AIart What’s interesting is how quickly people read this as a game, not just a video. It shows how AI clips are starting to function like prototypes in people’s heads. The expectation jumps straight from “this looks fun” to “this should be playable,” even when no mechanics exist yet.

存在しないレースゲーム by Grok

@andreeliasdev What’s interesting here isn’t just speed, it’s how the pipeline shifts where time goes. If assets come almost instantly, most of the work moves into making systems readable and consistent for players. Curious how much time you’re now spending on that vs asset creation.

After 25 days working on my game for the #vibejam I'm glad to share the building system! Now you can: ✅ Build houses ✅ Set respawn position placing a bed ✅ Keep your items safe putting them in a chest ✅ Craft items in the workbench I'll share the link to play it tomorrow if everything goes well 🤞🏻

Update: AltNext 2026 will take place on June 12–13, 2026 (not June 13–14). See you in Shanghai!

Interactive narrative is growing fast, but interaction still doesn’t consistently shape the story. Better tools aren’t the issue. Design is. AltNext 2026 — speaker applications open: b8xui27hs2.feishu.cn/share/base/for… #InteractiveNarrative #GameDev #XR

@runzhuotao That makes sense. If live animation becomes viable, it shifts from tooling to interaction design. The challenge won’t just be speed, but how you translate player intent into something that still feels precise, expressive, and consistent.

@AltnextSummit if a low latency model like gemini flash becomes competent enough, live animation is possible

Refining the text-to-animation system I've been working on. Here are a few showcases of combat animations, but the tech can be applied to pretty much anything. #gamedev #robloxdev

@hasan_ab_hasan RL could make behaviors less predictable, but it also risks turning this into a spectator system. How do you make sure players can still shape outcomes instead of just watching agents evolve?

@AltnextSummit yeah, you're right. maybe I can try something similar to RL, so we can see what do they do by themselves

I built a low-poly town builder from scratch with ThreeJS You can place houses, trees, animals, people, play with weather, day/night cycles, rain, snow. But here's the fun part: hit one button and your villagers become AI agents. They talk, trade, make decisions, and live their own little lives. Honestly thinking about turning this into a full game where you can build and manage your own towns and villages. Would you play something like this? #threejs #vibjeam #gamedev

@heyshmoo Maybe the question isn’t “more cinematics,” but whether Roblox can make them scalable, not just possible.

Roblox games need more cinematics, I think there's a big opportunity for interactive film-style games on the platfofm.

The risk isn’t AI. It’s staying attached to a model that no longer reflects how value is created. Join the conversation at AltNext 2026. #AltNext #AIGC

Not the tools nor the hours. You're paying for the ability to create something that holds attention.

If it no longer takes months and millions to produce content… what exactly are you paying for?

Every media shift starts with fear. AI is no different.

That’s where the pressure really is. Not on creativity itself, but on how creative work is priced.

And when execution becomes cheap, it stops being valuable.

When you can produce 500 videos a day, execution stops being the bottleneck.

It was this: AI is accelerating everything. More content. Faster production. Lower costs.

At AltNext last year, we discussed this in a panel on AI-generated brands with Scarlett Li, Pablo Mauron, and Patrice Nordey. But the real issue wasn’t fear.