Back to Home11/12/2025, 5:13:30 PM

Marble by World Labs: Multimodal world model to create and edit 3D worlds

dmarcos
47 points
18 comments

Mood

excited

Sentiment

positive

Category

tech

Key topics

AI

3D modeling

world models

Debate intensity20/100

World Labs has introduced Marble, a multimodal world model for creating and editing 3D worlds, sparking excitement and discussion about its potential applications and limitations.

Snapshot generated from the HN discussion

Discussion Activity

Active discussion

First comment

3h

Peak period

17

Day 1

Avg / period

9

Comment distribution18 data points

Based on 18 loaded comments

Key moments

  1. 01Story posted

    11/12/2025, 5:13:30 PM

    6d ago

    Step 01
  2. 02First comment

    11/12/2025, 7:51:01 PM

    3h after posting

    Step 02
  3. 03Peak activity

    17 comments in Day 1

    Hottest window of the conversation

    Step 03
  4. 04Latest activity

    11/13/2025, 9:23:21 PM

    5d ago

    Step 04

Generating AI Summary...

Analyzing up to 500 comments to identify key contributors and discussion patterns

Discussion (18 comments)
Showing 18 comments
thetoon
6d ago
3 replies
Not to belittle this or anything (it does look good and show promise), it feels like they somehow generate several consistent (but discrete) views of a given world, then feed all that to the good old pose estimation + gaussian splatting workflow. Whenever you leave the generated area (which isn't exactly huge on the few I tested) you get tell-tale signs of GS.
kkukshtel
6d ago
This was my take as well — this is just pose estimation from generated stereo panoramic images.
embedding-shape
6d ago
Yeah, it's more of a somewhat 3D-drawing of a frame that you can navigate inside, rather than a world up that happen to fit with whatever image you use as an input, but makes sense as a standalone world when you walk around. For being a "world" model, it doesn't seem to grasp physical space very well.

The interior scenes look and walks great, but any scenes with/in exteriors seems kind of bad.

xg15
6d ago
Yeah, if the entire point is that you can move around inside those worlds, I'd have expected a bit more "walkability" - maybe a few different viewpoints that each have their own Gaussian splatting? Right now, it dissolves pretty quickly once you change the location.
pedalpete
6d ago
1 reply
It's amazing to see how this space is developing. About 7 years ago I was building "spatial media" with https://ayvri.com

Nobody believed us when we said AI would create 3D virtual worlds that were indistinguishable from the real thing, and we'd be able to transport people to different places.

I particularly like the artistic effect of the drawing that brings the person into this world. Like a point-cloud that then gets "filled in".

I have little doubt this was a design decision and I think it is very well executed.

jaccola
6d ago
1 reply
Even more amazing to me is that the tech to create these really existed 7 years ago (would have been slower to train but most methods don't need the latest GPUs). This means there are no doubt more improvements just waiting to be discovered!
pedalpete
5d ago
The tech was not available, but it was the direction we were heading.

Digital Twins were a thing, and we had developed a high-resolution 3d world, outside of cities.

At the time, we thought that NERFs were going to allow us to increase resolution and fill in the gaps of what we didn't know about the world. Then Gaussian Splats came in and just took over.

There are definitely still improvements and techniques.

However, people occasionally still reach out to me to ask how to build a replica of Ayvri, and I tell them you wouldn't build it today like we did back then.

Today, you wouldn't go through the processes of setting up tile-servers, I think you can get current AI to build a scene frame by frame and transition between frames, rather than tile by tile.

But others in the gaming world may have different opinions as to where the industry is heading.

proof_by_vibes
6d ago
1 reply
Are there any experts that could help me bootstrap myself on the current literature on "world models?"
jaccola
6d ago
1 reply
In this current generation, "world models" is basically a marketing term. You can research gaussian splatting, novel view synthesis, neural radiance fields (nerf), etc... I find Mr Nerf is good to follow: https://x.com/janusch_patas

There is another thing called world models that involves predicting the state of something after some action. But this is a very very limited area of research. My understanding of this is that there just isn't much data of action->reaction.

Same issue with gaussian splatting/nerf really, very little data (relative to text/images/videos) of text -> 3d splats. I'd guess what world labs are doing is text -> image -> splats, but of course it is just speculation.

cl42
6d ago
> There is another thing called world models that involves predicting the state of something after some action. But this is a very very limited area of research. My understanding of this is that there just isn't much data of action->reaction.

Folks interested in this can look up Yann LeCun's work on world models and JEPA, which his team at Meta created. This lecture is a nice summary of his thinking on this space and also why he isn't a fan of autoregressive LLMs: https://www.youtube.com/watch?v=yUmDRxV0krg

ogogmad
6d ago
1 reply
Slightly off-topic: I've just watched this takedown of an AI-generated chart-topping song: https://www.youtube.com/watch?v=rGremoYVMPc&lc=UgxfDvqX1G6kp...

OK, so I've talked about this phenomenon with ChatGPT, and I think that the issue here is that to a lot of people, a song needs to be more than just a "song". There's some sort of requirement for it to be the un-faked result of having certain experiences. It has to relate to something happening in reality, and to be derived from it, and cannot exist in a vacuum separated from the rest of reality. Otherwise to them, the music isn't "real".

embedding-shape
6d ago
Endless droned ambient music disagrees with you that there is any sort of "requirement of certain experiences". Some of it is basically someone hitting play on a modular synth patch and letting it play until it sounds done, (some) people are still fine with listening to it.
the_real_cher
6d ago
Thats totally insane and amazing.
lvl155
6d ago
It would be nice to have these world models integrated with Blender.
ganelonhb
6d ago
wow, it’s slop!
alyxya
6d ago
Something about the camera perspective creates a skew that makes things feel artificial to me. It's a minor thing that bothers me, but I'd like the geometry to feel more like what I normally see. Video generation models tend to feel more natural in perspective.
MarsIronPI
6d ago
I'm looking forward to the future of games and movies if these world models keep improving. Imagine if anyone with an interesting idea could sketch it, plug it into a world model and share the result with everyone. It'd open up a huge amount of possibilities.

Not to mention being able to explore worlds from already existing works. Care to go for a ride on a broomstick? How about simply walking into Mordor? It's exciting.

View full discussion on Hacker News
ID: 45902732Type: storyLast synced: 11/17/2025, 6:02:30 AM

Want the full context?

Jump to the original sources

Read the primary article or dive into the live Hacker News thread when you're ready.

Read ArticleView on HN