Strong concept, weak storytelling — the output looks impressive but the launch needed a wow demo video to really land.
Comments (15)
Sign in to join the discussion.
K
Kaveh R.Apr 30, 2026
Calling it a world model is doing a lot of heavy lifting when the demo is a 12 second flythrough of a kitchen. Show me a chair I can knock over.
M
MarvApr 30, 2026
Back in my day we called this photogrammetry and it took six hours and a Canon 5D. Cute though.
L
Linnea HolmApr 30, 2026
The reveal video had no voiceover and somehow that made it more convincing than every other 'world model' launch this year. Restraint is a flex.
D
Damilola O.Apr 30, 2026
ok wait, real-time rendering AND physically grounded from one image? if even half of that holds up at scale this changes the asset pipeline for like every indie studio.
P
Prash V.Apr 30, 2026
Reminds me of the early thesis behind one of our portfolio bets in neural radiance work. Curious where you land on licensing vs API.
I
Idris K.Apr 30, 2026
Real-time rendering of generated 3D scenes sounds like a GPU bill that keeps a CFO up at night. What's the plan, per-scene pricing or seat-based?
J
junieApr 30, 2026
naive q but what happens at the edges of the scene where the single input image had no info? does it hallucinate a wall or just fade out?
B
Benoit LavalApr 30, 2026
The thumbnail choice was brave. A blurry living room as the hero shot for a 3D model launch is a vibe I respect but did not predict.
T
Tara N.Apr 30, 2026
hot take: this is the last spatial model launch where the demo is a passive camera orbit. next one has to let me walk in or it doesn't count.
M
Mikkel S.Apr 30, 2026
Tagline rewrite, on the house: 'one image in, a world out.' you can venmo me later.
R
Raghav P.Apr 30, 2026
how big is the team behind Echo-2? asking because if this is under ten people I need to lie down for a bit.
O
Ohene F.Apr 30, 2026
Building in adjacent territory on the avatar side and genuinely rooting for you, even if you keep stealing the spotlight on launch days. Well played on the timing.
Y
Yuki TanabeApr 30, 2026
Would love to know if this is closer to a feed-forward gaussian splatting approach or something more diffusion-flavored under the hood. The temporal consistency in clip 3 is suspiciously good.
C
Chen W.Apr 30, 2026
We prototyped something directionally similar internally in 2019 and shelved it because the compute wasn't there. Glad someone finally got the timing right.
S
Sofia La RosaApr 30, 2026
the launch tweet using bold unicode for '3D scenes' is such a 2024 move and i mean that as a compliment, you know your audience scrolls fast.
Work With Us
Book an intro call to explore our suite of services. We might have a waitlist.
SpAItial AI
SpAItial is pioneering spatial foundation models
SpAItial is pioneering spatial foundation models
Tags
TLVC Rating
Community Rating
No ratings yet
to rate this launch.
Strong concept, weak storytelling — the output looks impressive but the launch needed a wow demo video to really land.
Comments (15)
Calling it a world model is doing a lot of heavy lifting when the demo is a 12 second flythrough of a kitchen. Show me a chair I can knock over.
Back in my day we called this photogrammetry and it took six hours and a Canon 5D. Cute though.
The reveal video had no voiceover and somehow that made it more convincing than every other 'world model' launch this year. Restraint is a flex.
ok wait, real-time rendering AND physically grounded from one image? if even half of that holds up at scale this changes the asset pipeline for like every indie studio.
Reminds me of the early thesis behind one of our portfolio bets in neural radiance work. Curious where you land on licensing vs API.
Real-time rendering of generated 3D scenes sounds like a GPU bill that keeps a CFO up at night. What's the plan, per-scene pricing or seat-based?
naive q but what happens at the edges of the scene where the single input image had no info? does it hallucinate a wall or just fade out?
The thumbnail choice was brave. A blurry living room as the hero shot for a 3D model launch is a vibe I respect but did not predict.
hot take: this is the last spatial model launch where the demo is a passive camera orbit. next one has to let me walk in or it doesn't count.
Tagline rewrite, on the house: 'one image in, a world out.' you can venmo me later.
how big is the team behind Echo-2? asking because if this is under ten people I need to lie down for a bit.
Building in adjacent territory on the avatar side and genuinely rooting for you, even if you keep stealing the spotlight on launch days. Well played on the timing.
Would love to know if this is closer to a feed-forward gaussian splatting approach or something more diffusion-flavored under the hood. The temporal consistency in clip 3 is suspiciously good.
We prototyped something directionally similar internally in 2019 and shelved it because the compute wasn't there. Glad someone finally got the timing right.
the launch tweet using bold unicode for '3D scenes' is such a 2024 move and i mean that as a compliment, you know your audience scrolls fast.