Huge thanks to the @hytopia team for giving me early access to the new NEO version.
This is a big step forward — it allows me to push things to the next level with HYTOPIA STUDIO V2 NEO.
Exciting things coming.
English
Sethos AI creation
4.8K posts
Hello everyone,Small change of plans — I’m putting development on hold for now while I wait to integrate Hytopia NEO. The goal is to avoid having to redo everything after such a major update, so I’d rather take the time to analyze the situation before going further.In parallel, I’ve reached out to the team to request access to the SDK so I can run some integration tests. We’ll see how it goes — stay tuned.
Sethos AI creation@Sethos_AI
Leak video is closer than you think. 👀
English
Alright, I was waiting for a new model, but I won’t be picky.
Lentils@Lentils80
🚨 Veo 3.1 with R2V (reference-to-video) and voice replication was spotted. Courtesy of discord.gg/z-ai
English
@sharecashBTC @AdonisHY @hytopia @Temptranquil @iamarkdev Once I’ve finished the marketplace phase of Hytopia Studio, I’d be interested in letting you list your castle there at whatever price you want.
English
public deal announcement for @AdonisHY - you have more connections than I do so I’ll make an offer - help me get this castle in @hytopia and it's yours. A true 1/1 build. Use it for your game. I'll send you the file. @Temptranquil @iamarkdev
FULLY FURNISHED
250x250 castle 🏰
English
That’s exactly it — creating an experience generates a kind of memory file for the entire game, allowing it to be re-read before each action.This is especially relevant with Claude, which has a much higher token limit, but I imagine that over time this will become easier and more seamless overall.
English
Not sure what you mean 'memory'....but I mean, I assume you must be storing your data (nodes, placements, logic) of a specific game in some format?
If you ensure there is a claude.md or equivalent of how that format works, and all of the 'options', hopefully the AI can understand and modify it?
English
There is indeed memory in LLMs, along with an ability to read the map in its current state. However, I am absolutely not satisfied with the results. The constructions and modifications still feel too basic. The AI creates and remembers, but it doesn’t take the time to properly develop its ideas, which results in underdeveloped houses.
English
@Sethos_AI Looks pretty cool! Wondering how u save the data, and can the LLM see/access it 'easily' so it can both help debug, and build along side you?
English
So what we’re basically saying here is that you can turn a really crappy game into a beautiful one…
NVIDIA GeForce@NVIDIAGeForce
Announcing NVIDIA DLSS 5, an AI-powered breakthrough in visual fidelity for games, coming this fall. DLSS 5 infuses pixels with photorealistic lighting and materials, bridging the gap between rendering and reality. Learn More → nvidia.com/en-us/geforce/…
English
We finally know why Seedance 2.0 is taking longer to release.
Honestly… not really a surprise.
They’re mainly working on securing the copyright system.
Once that’s solved, we’ll finally be able to enjoy some of the best AI video models out there.
Patience. 🚀
The Information@theinformation
ByteDance has suspended the global launch of its latest video-generation AI model after copyright disputes with major Hollywood studios. Read more from @JuroOsawa and @QianerLiu 👇 thein.fo/4rsECmD
English
Post exceptionnellement en français aujourd’hui.
C’est un projet français, mais je pense sincèrement que son message peut toucher bien au-delà des frontières.
Très heureux de vous présenter “Qu’est-ce qu’il me reste” de William Elger.
Cette chanson aborde un sujet profondément humain : la mémoire qui s’efface, et le combat face à la maladie d’Alzheimer.
J’ai eu la chance de travailler sur ce projet pour la création de l’album et de réaliser le clip.
Oui, certaines instrumentales ont été créées avec l’aide de l’IA — mais derrière, il y a eu un énorme travail humain : adaptation, direction artistique, réglage des timings, écriture des paroles…
Et pour le mix & mastering, nous avons fait appel aux meilleurs.
🎧 youtu.be/vJSSDIl1Ooc?is…
L’album arrive très bientôt. Restez à l’écoute.
YouTube
Français
👀Just a few more details and I’ll be able to make a demo video. We’re still building. 💪
Sethos AI creation@Sethos_AI
What if I dropped a small HYTOPIA STUDIO leak? 👀 Still deep in development, but things are starting to look really interesting. Should I show a sneak peek? 😏
English
1m of tokens, I just want to say "perfect."
GIF
Claude@claudeai
1 million context window: Now generally available for Claude Opus 4.6 and Claude Sonnet 4.6.
English
@sc_ttrblx @RBXBloxy Your comment is interesting. Do you mean for the thumbnail or do you have something else in mind?
English
@Sethos_AI @RBXBloxy If you’re gonna use AI, don’t use AI images at the very least.
English
Hey everyone 👋
I know many of you are excited to see the studio in action. I'm doing my best to show it in the next few days with some features running.
The internet AI is still a bit lazy and needs some adjustments, but it pushed me to go deeper into the sandbox and creative side.
And that might actually lead to something interesting sooner than expected: asset sharing.
For example:
I build a castle made of multiple blocks → publish it in the shop (free or paid) → other developers can instantly reuse it.
Same for scripts, sounds, props…
Basically, anything created can become an asset.
A little more patience — it's coming soon 🚀
Sethos AI creation@Sethos_AI
Hey everyone 👋 It’s been a few days since I shared news about Hytopia Studio — and there’s a good reason. We’re currently in the most complex phase of development: calibrating the AI so it can correctly understand assets and place them in the world automatically. At the same time, I’ve successfully integrated the official Hytopia shop and drastically improved the user interface. Still a bit of patience… Once this is finished, anyone will be able to create a game with (almost) a single sentence. The goal: turn ideas into playable worlds instantly. 🚀
English
It doesn’t always follow the instructions, and you don’t really see the last frame properly yet, but it feels like the beginning of something.
English
I’m not sure if this was already available, but I just noticed you can now do first & last frame generation with Grok.
Here’s a quick test I made 👇
Pretty interesting for controlling motion and transitions.
English