Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

🏀 I built this PoC using a single 2D broadcast video: applied pitch detection, player positioning, and movement tracking to reconstruct the play in 3D and map it onto a real basketball court in XR.

Stijn Spanhove

4,216 subscribers

36,957 views • 10 months ago •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

26 Comments

Stijn Spanhove10 months ago

You can walk through the scene, pause it, analyse movement, see shot markers, select players and view stats. The play: Julie Allemand’s three-pointer in the Eurobasket final (Spain vs Belgium). Built with R3F and WebXR. #threejs @skalskip92 thanks for the inspiring examples!

SkalskiP10 months ago

omg! this is soooo cool!

iambenslater10 months ago

This is incredible. Seriously. As lightweight AR glasses become accessible, imagine a team training where all players are wearing the glasses to work through plays.

Lidoor L. Joseph10 months ago

For a millisecond I thought you hyperscape scanned and invited people to come play.

Stijn Spanhove10 months ago

No that would also be an option 😅

Lidoor L. Joseph10 months ago

Don’t forget to send the hyperlink if you ever do 😎

Azad Balabanian10 months ago

Such a cool concept!

Chris Smith10 months ago

Well done, that's a great idea (and execution)!

miki ⋆｡°✩10 months ago

awesome

Daa 😎10 months ago

Cool!

SaoGalaxyVR10 months ago

This is nice full scale

Natsuo Okada ⛏️🇯🇵10 months ago

Excelente work!

Sam Jurrens, CFA10 months ago

Huge for pro teams reviewing game tape

Zara Techie10 months ago

Impressive innovation! Truly groundbreaking work.

Furqan Rydhan10 months ago

looks cool!

AI PlanetX10 months ago

Impressive innovation! Your creativity is truly remarkable.

Jakub Klas10 months ago

This is nuts!

aLTis🇱🇹10 months ago

That's actually really cool

Peter Von Buskirk10 months ago

😲

Async (in SF)10 months ago

What. Seriously?

NIHAL JANI10 months ago

Why cant i see anything like in this on my sam 3d playground

Urs Gehrig10 months ago

Hello Stijn, do you intend to close the loop and also project from a 3D representation (e.g. a table soccer video) into a 2D view? Those projection transformations are generally helpful in sports teaching or e.g. music marching disciplines.

Agent Synister10 months ago

Dude this is genius, well done, I imagine teams could use this to make better plays, critique movements etc, awesome

Steven Haller10 months ago

This deserves much more recognition

10goldte.eth10 months ago

we need a real product

John Lehtonen10 months ago

I love this

Related Videos

AI BASKETBALL ANALYSIS. A FULL COMPUTER VISION SYSTEM. BUILT ON YOLO, OPENCV, AND PYTHON. Take any NBA broadcast. Any camera angle. Any resolution. Feed it into the system. YOLO finds every player and the ball. Frame by frame. No manual annotation. No pre-labeled data.The model just sees the court and understands it. Zero-shot classification looks at jersey pixels and decides who plays for whom. Two teams separated in milliseconds. Without ever being told the team names. But here's where it gets wild. Court keypoint detection identifies the geometry of the playing surface. Homography transforms the broadcast camera into a top-down tactical map. Real meters. Real coordinates. From a flat 2D video. Now every player has a position on a real court. Every movement becomes a data point. Every pass becomes a vector. Speed. Distance. Possession time. Pass networks. All extracted from nothing but pixels. No GPS. No chips in the ball. No million-dollar Second Spectrum setup. A Python script and a GPU. The full pipeline is open source. The tutorial walks through every line of code. From detection to transformation to analytics output. Teams used to pay six figures for this data. Now you build it in an afternoon. Computer vision didn't just enter basketball. It democratized it.

AI BASKETBALL ANALYSIS. A FULL COMPUTER VISION SYSTEM. BUILT ON YOLO, OPENCV, AND PYTHON. Take any NBA broadcast. Any camera angle. Any resolution. Feed it into the system. YOLO finds every player and the ball. Frame by frame. No manual annotation. No pre-labeled data.The model just sees the court and understands it. Zero-shot classification looks at jersey pixels and decides who plays for whom. Two teams separated in milliseconds. Without ever being told the team names. But here's where it gets wild. Court keypoint detection identifies the geometry of the playing surface. Homography transforms the broadcast camera into a top-down tactical map. Real meters. Real coordinates. From a flat 2D video. Now every player has a position on a real court. Every movement becomes a data point. Every pass becomes a vector. Speed. Distance. Possession time. Pass networks. All extracted from nothing but pixels. No GPS. No chips in the ball. No million-dollar Second Spectrum setup. A Python script and a GPU. The full pipeline is open source. The tutorial walks through every line of code. From detection to transformation to analytics output. Teams used to pay six figures for this data. Now you build it in an afternoon. Computer vision didn't just enter basketball. It democratized it.

zostaff

231,628 views • 5 months ago

🕹️ Turning drawings into XR worlds you can play in. I built a POC for @Spectacles that turns imagination into reality. My niece drew a picture and with the help of Mirage 2, I brought it to life in an interactive environment.

🕹️ Turning drawings into XR worlds you can play in. I built a POC for @Spectacles that turns imagination into reality. My niece drew a picture and with the help of Mirage 2, I brought it to life in an interactive environment.

Stijn Spanhove

37,239 views • 1 year ago

🥳 NEW TUTORIAL 🥳 2D look is making a big come back and I love it! In this video I give you my tips to achieve convincing 2D look for your 3D animation using #Blender obviously 🤗 👉 #b3D #animation #p2dessign #2D #3D #anime

🥳 NEW TUTORIAL 🥳 2D look is making a big come back and I love it! In this video I give you my tips to achieve convincing 2D look for your 3D animation using #Blender obviously 🤗 👉 #b3D #animation #p2dessign #2D #3D #anime

Pierrick Picaut

71,985 views • 1 year ago

I built a post-run recap design studio in SwiftUI Live animated map, 2D and 3D rendering, full customization. Here's how everything works 👇

I built a post-run recap design studio in SwiftUI Live animated map, 2D and 3D rendering, full customization. Here's how everything works 👇

JC Builds

37,931 views • 4 months ago

The NHL partnered with Disney to broadcast its first-ever game in 3D animation yesterday. The broadcast used tracking chips embedded within the puck and player jerseys to recreate a real-time animated version of the game on Disney Channel. Pretty cool 🤯

The NHL partnered with Disney to broadcast its first-ever game in 3D animation yesterday. The broadcast used tracking chips embedded within the puck and player jerseys to recreate a real-time animated version of the game on Disney Channel. Pretty cool 🤯

Joe Pompliano

1,116,309 views • 3 years ago

The NBA app on Apple Vision Pro just dropped a massive update. World View (Alpha) puts you inside a live 3D recreation of the game. Every player and ref tracked in real time, rendered in NBA city-inspired environments, with visual effects when buckets drop or shots clank off the rim. You can move around the court using waypoints and even pull up the live broadcast synced to the 3D recreation simultaneously. The models aren’t photorealistic yet, but the tracking is impressively accurate. When they start mapping real likenesses onto these, it’s going to be something else entirely. Lastly, Tabletop is now called Court View. The NBA keeps pushing spatial computing forward. I love to see it!

The NBA app on Apple Vision Pro just dropped a massive update. World View (Alpha) puts you inside a live 3D recreation of the game. Every player and ref tracked in real time, rendered in NBA city-inspired environments, with visual effects when buckets drop or shots clank off the rim. You can move around the court using waypoints and even pull up the live broadcast synced to the 3D recreation simultaneously. The models aren’t photorealistic yet, but the tracking is impressively accurate. When they start mapping real likenesses onto these, it’s going to be something else entirely. Lastly, Tabletop is now called Court View. The NBA keeps pushing spatial computing forward. I love to see it!

Justin Ryan

28,377 views • 6 months ago

Combining 2D and 3D in After Effects. In Foldspace, not every surface needs to be bent. Keep one flat for 2D artwork, and let the others exist in 3D space. This allows you to combine 2D and 3D objects in a single animation with no masks or tricks needed.

Combining 2D and 3D in After Effects. In Foldspace, not every surface needs to be bent. Keep one flat for 2D artwork, and let the others exist in 3D space. This allows you to combine 2D and 3D objects in a single animation with no masks or tricks needed.

Motion Princess

22,007 views • 6 months ago

I recreated the #ArcaneSeason2 Piltover Garden scene using #Blender. I roughly built a 3D scene, made a draft render, and then painted over it in 2D. All moving parts like foliage, characters and spinner are 3D elements. Breakdown coming soon. More info:

I recreated the #ArcaneSeason2 Piltover Garden scene using #Blender. I roughly built a 3D scene, made a draft render, and then painted over it in 2D. All moving parts like foliage, characters and spinner are 3D elements. Breakdown coming soon. More info:

trungduyng

34,849 views • 1 year ago

I got a lot of questions yesterday on how this made and if it was 3D or 2D so I decided to make a small breakdown video to explain. It was created using Non-Photorealistic Shaders and adapting a 2D animation workflow into 3D within Blender itself.

I got a lot of questions yesterday on how this made and if it was 3D or 2D so I decided to make a small breakdown video to explain. It was created using Non-Photorealistic Shaders and adapting a 2D animation workflow into 3D within Blender itself.

BNBaku

80,329 views • 2 years ago

There is a game coming out where you have to play it in 2d and 3d at the same time

There is a game coming out where you have to play it in 2d and 3d at the same time

Dexerto

7,085,009 views • 2 years ago

I built a full 3D strategy game with Opus 4.8 + Fable 5 Launching it today. Runs in your browser, no install, free: > 21 battlefields > economy + tech tree > tanks vs jets vs turrets, real counters > single player vs up to 7 AI bots > a battle royale storm map Click and you're at war in 10 seconds 👇

I built a full 3D strategy game with Opus 4.8 + Fable 5 Launching it today. Runs in your browser, no install, free: > 21 battlefields > economy + tech tree > tanks vs jets vs turrets, real counters > single player vs up to 7 AI bots > a battle royale storm map Click and you're at war in 10 seconds 👇

SoloTech

68,242 views • 4 months ago

Global Anomaly Detection System update. Inspired by the recent global outage. visualizing worldwide impact zones while experimenting with projecting instanced geometries from a 3D sphere to a 2D Mercator map in three.js

Global Anomaly Detection System update. Inspired by the recent global outage. visualizing worldwide impact zones while experimenting with projecting instanced geometries from a 3D sphere to a 2D Mercator map in three.js

Dilum Sanjaya

39,622 views • 11 months ago

Magic123: One Image to High-Quality 3D Object Generation Using Both 2D and 3D Diffusion Priors paper page: present Magic123, a two-stage coarse-to-fine approach for high-quality, textured 3D meshes generation from a single unposed image in the wild using both2D and 3D priors. In the first stage, we optimize a neural radiance field to produce a coarse geometry. In the second stage, we adopt a memory-efficient differentiable mesh representation to yield a high-resolution mesh with a visually appealing texture. In both stages, the 3D content is learned through reference view supervision and novel views guided by a combination of 2D and 3D diffusion priors. We introduce a single trade-off parameter between the 2D and 3D priors to control exploration (more imaginative) and exploitation (more precise) of the generated geometry. Additionally, we employ textual inversion and monocular depth regularization to encourage consistent appearances across views and to prevent degenerate solutions, respectively. Magic123 demonstrates a significant improvement over previous image-to-3D techniques, as validated through extensive experiments on synthetic benchmarks and diverse real-world images.

Magic123: One Image to High-Quality 3D Object Generation Using Both 2D and 3D Diffusion Priors paper page: present Magic123, a two-stage coarse-to-fine approach for high-quality, textured 3D meshes generation from a single unposed image in the wild using both2D and 3D priors. In the first stage, we optimize a neural radiance field to produce a coarse geometry. In the second stage, we adopt a memory-efficient differentiable mesh representation to yield a high-resolution mesh with a visually appealing texture. In both stages, the 3D content is learned through reference view supervision and novel views guided by a combination of 2D and 3D diffusion priors. We introduce a single trade-off parameter between the 2D and 3D priors to control exploration (more imaginative) and exploitation (more precise) of the generated geometry. Additionally, we employ textual inversion and monocular depth regularization to encourage consistent appearances across views and to prevent degenerate solutions, respectively. Magic123 demonstrates a significant improvement over previous image-to-3D techniques, as validated through extensive experiments on synthetic benchmarks and diverse real-world images.

AK

305,746 views • 3 years ago

I applied Matt's prompt, ran it for a little over ten hours. About 1.3 million tokens on Opus 5. It created the game. I also made a video of it using Opus 5. Here's the video (and in my reply to this is a link to the actual game so you can play).

I applied Matt's prompt, ran it for a little over ten hours. About 1.3 million tokens on Opus 5. It created the game. I also made a video of it using Opus 5. Here's the video (and in my reply to this is a link to the actual game so you can play).

James Altucher

128,128 views • 2 months ago

For the facial animations in my game, I use a technique that yields results similar to L.A. Noire. To keep it simple: I first generate a facial animation video using AI like LTX 2. Then, I create a video depth map from that animation and project it onto a face mask. Using vertex displacement, the depth map dynamically deforms the face as the character speaks. The result looks good in standalone VR. I actually created this workflow long time ago:

For the facial animations in my game, I use a technique that yields results similar to L.A. Noire. To keep it simple: I first generate a facial animation video using AI like LTX 2. Then, I create a video depth map from that animation and project it onto a face mask. Using vertex displacement, the depth map dynamically deforms the face as the character speaks. The result looks good in standalone VR. I actually created this workflow long time ago:

Alex

48,582 views • 7 months ago

LaMelo Ball on building a chemistry with Rudy Gobert “I mean, he’s played basketball a long time, I have too, so it’s kind of easy. He’s just a big and in a little for real” On if he’s played with a player like Gobert before “I mean, he’s hall of fame for sure so that’s tough to say, but, I play a lot of 2K, you just play a lot of basketball, basketball is the same from a video game to on court to real life, On Gobert off the court “Rudy is funny as hell, great dude”

LaMelo Ball on building a chemistry with Rudy Gobert “I mean, he’s played basketball a long time, I have too, so it’s kind of easy. He’s just a big and in a little for real” On if he’s played with a player like Gobert before “I mean, he’s hall of fame for sure so that’s tough to say, but, I play a lot of 2K, you just play a lot of basketball, basketball is the same from a video game to on court to real life, On Gobert off the court “Rudy is funny as hell, great dude”

Andrew Dukowitz

200,853 views • 11 days ago

This game makes you play it in 2d and 3d at the same time, now they have caves

This game makes you play it in 2d and 3d at the same time, now they have caves

Dexerto

3,549,573 views • 2 years ago

I built a real-time, multiplayer whiteboard fully in Lovable. Real-time sync, cursor tracking, and drawing tools, all built in an afternoon.

I built a real-time, multiplayer whiteboard fully in Lovable. Real-time sync, cursor tracking, and drawing tools, all built in an afternoon.

Tyler Bruno

28,031 views • 6 months ago

Today’s hobby project: Prototyped a real-time room scanning system with live sharing to XR devices. Explore scanned spaces in 3D on XR (not flat 2D screens) and share them with remote users for spatial remote collaboration. #metaquest #augmentedreality

Today’s hobby project: Prototyped a real-time room scanning system with live sharing to XR devices. Explore scanned spaces in 3D on XR (not flat 2D screens) and share them with remote users for spatial remote collaboration. #metaquest #augmentedreality

Takashi Yoshinaga

126,635 views • 9 months ago