Builder and Content Creator @DesignCodeio Youtube Tutorials: https://teapawt.net/t.co/OAcQzsCzzP
Joined November 2023
- Tweets349
- Following196
- Followers535
- Likes333
Pinned Tweet
Turn photos into 3d model
Let me know if you want the template for free, I'll send it to you
I made this super detailed OP-XY 3d model with simple prompts and good references. You can use this workflow to create your own 3d models from anything, photos, ai images, pdf, links etc
prompts in the youtube video: youtu.be/j8FPbAsGfxE
13
1,616
Opus 5.5 rebuilt a real room in Three.js, the keyboard, the monitor, even the blind shadows on the wall
change the mood without touching the scene
Perfect for tutorial intros, product demos
Three.js + Opus 5.5
I asked Opus 5.5 to recreate my room in three.js just so it can animate my tutorial videos.
3D + motion design. Insane how far we've come.
4
192
this feels like a game scene
the fog is important here, it hides the horizon, keeps the scene light, makes it more dramatic.
Three.js + Blender + Opus 5.5 + Astra
160
Baldur’s Gate-style turn-based RPG, in a browser tab
rigging and animations are what takes most of the time in these builds. Once ai figures out how to rig perfectly, not just human skeletons, then we see AAA games completely made with ai
three.js + Opus 5.5
Always wanted to make a turn-based game like baldur's gate 3.
This is all done with three.js and opus 5.5
2
214
100k free assets
Every 3D object as a Python program, every part named, every dimension a parameter
clean topology, 3.8x lighter on tokens than Blender Python, huge for AI-generated 3D on the web
sand3d, open source
We're opening 100,000+ 3D objects 🧊🤗, unified in one library, ready for LLM training.
Objaverse assets, CAD parts and articulated mechanisms, all in one language !
For years, 3D research has trained on Objaverse: messy triangle soups, broken meshes, objects with no parts. Every team cleaned it on its own or adapted its models to noisy data, and most of that work stayed private.
That mess held the industry back on what 3D teams need: clean topology, objects split into parts, and a representation people can work on together.
So we rewrote Objaverse and other datasets into one clean representation. Each asset is now a short Python program in sand3d, our open library:
🔹 every part has a name
🔹 every dimension is a parameter
🔹 running the program gives a clean asset
3D code today is scattered across many languages, each written for people and for a different tool. sand3d puts every object in the same language, in 3.8x fewer tokens than raw Blender Python on average.
Everything is open, as one shared dataset that gives everyone the same ground to train and compare on. We want to speed up research on models that work and collaborate on 3D production.
This is only the start: more objects are coming.
📦 Dataset: huggingface.co/collections/S…
🔎 Explorer: huggingface.co/spaces/Sandfl…
💻 Library: github.com/Sandflow-ai/sand3…
2
1
111
wanted to see if a product page could feel like a launch film, so I built AURAL
the verdict: it can. the headphones are a real 3D model made in Blender, and the whole page plays as you scroll
if you design product pages, build around one 3D model instead of a pile of renders
Let me know if you want the template for free
2
6
228
Fable 5.5 made this entirely in codes
imagine creating videos in code instead of using video generation. The art style, animation, scenes, everything is so good, can't wait for the official release
Fable 5.5 its over for video and image models 🤯
THIS IS PURE CODE
one shotted this video holy shit , how this is more beautiful than ai video models
Two ways to check this yourself:
Ask: “Where did @karpathy go in 2026?” with web search and memory turned off. Only the new Fable knows the answer. Then give it the prompt.
Run the same prompt I used for the test (it’s in the screenshot) and see for yourself.
Don’t rely on Grok for basic reasoning like this. Grok doesn’t have up-to-date knowledge of niche stuff. Either join the community or run the test yourself.
Also, a knowledge cutoff doesn’t mean a model knows everything before that date. Sometimes a model doesn’t know about an event even if it happened before its cutoff. This test shows that too.
Run the same prompt on regular Opus 5.5 and the quality drops. On Fable 5.1 it’s worse.
1
1
152
I love ai video models comparison, it allows us to see the result without spending a penny. anything to save those precious tokens.
Then you can just go straight to one that fits your project the best, the price comparison is also very useful
youtube.com/watch?v=mqaSHTcV…
1
4
184
thousands of lines folding like silk, colors sliding, this is hypnotic
Would be a great looping background
1
2
93
Interesting effect
Your cursor smears the image into thick, glossy oil paint, then it slowly settles back
Feels like touching wet canvas
Oil Paint Restore - Cursor FX
Free Source Code & Demo:
codepen.io/editor/sabosugi/p…
#threejs #webgl #shader
Crazy, 1300 3D models, free and CC BY
also every one ships with its input image, prompt and settings. Basically a side-by-side of six 3D models, so you can see which one fits your style before spending a credit
huggingface.co/datasets/scen…
GLBs drop straight into Three.js
We just open-sourced 1,300+ 3D models.
All 39 GB are free to download on Hugging Face.
huggingface.co/datasets/scen…
> made with Hitem3D, Meshy, Rodin, Tripo, Claude Opus 5.5 3D and GPT-6 Astra 3D, which all run on Scenario
> includes the input images, prompts and settings that made them
> GLBs up to 8K textures, native Blender files from Opus and Astra, CC BY 4.0
1
179
So amazing
Full 3D tower defense, 5 maps, 14 towers, boss fights, hundreds of enemies on screen, built in Godot
Opus 5.5 is past demos, this is a shippable game scope
THIS IS WHAT CLAUDE OPUS 5.5 CAN BUILD!
A full 3D tower defense game: 5 maps, 14 towers, boss fights, volcano eruptions, hordes of hundreds of enemies.
Game dev is 100% possible with Opus 5.5
60 seconds of real gameplay...
1
2
137
This is so neat, if something changes you don’t need to record again, unlike a video.
I made a skill that turns your app into a demo in HTML.
So they stay sharp at any size, loop, scrub, jump to any step and stand still for reduced motion.
I open-sourced it as a skill:
fil.design/skills/html-demo