Liam Wilson

[Elvar @elvar_official]

This is f**king awesome.

I still don't get how this isn't everywhere yet.

Every week Claude and I grab one thing people hold all day, crack it open and make every part inside move.

This time it's a phone. Flip it over, watch it unlock, then peel it apart into 9 layers down to a chip the size of a fingernail.

> Claude Code moved the face unlock before the teardown, because 30,000 infrared dots only make sense from a phone that's still in one piece.
> The camera lens floats on magnets and slides against a shaky hand, so the shot stays still.
> The buzz is a tiny weight on 2 springs moving 170 times a second, slowed 40× so you can actually see it.
> The chip zoom ends on 19 billion transistors packed into that fingernail.

The full build and a playground to try it yourself are in the article 👇

[DeepBlue深藍 @DeepBlueX0]

GPT image 2.5 Prompt:

動漫遊戲女性角色 × X光透射攝影 × 二維透射投影 × 淺底深像 × 極簡留白

暖灰白底，角色呈細膩灰黑影像，保留標誌性髮型、原服裝輪廓與關鍵配飾。均勻透射成像，以吸收灰階取代表面明暗，灰階隨材質吸收與沿射線方向的重疊累積變化：髮絲與薄衣料淺淡，身體組織層次柔和，骨骼較深，金屬配件形成局部深色。前後結構穿透相交，細部顯隱不一，部分輕薄邊緣融入背景，整體呈平面的放射攝影印相質���。

六張各選一位廣為人知的不同女性角色，全身、半身與局部自由變化，姿態、朝向及構圖不重複。

9:16，n=6，六張獨立單圖，不拼圖。左下簽名：深藍色實心圓點在前，白色手寫體「DeepBlue」在後。

[DeepBlue深藍 @DeepBlueX0]

假如你有透視眼
大概就是這個效果

實驗效果雖沒達到我預期
但有趣，供大家玩一下～

Prompt依旧放评论👇

[あぎ @agi_aibusi]

AIモデル「顔」データベース無料公開！

Midjourneyで生成した自然なAI美女を
顔だけ抜き出してデータベース化しました

・無料
・商用利用OK
・編集、二次利用OK

引用元の記事からお受け取り下さい

[rewind @rewind02]

full office in 3d with claude opus 5.5 + blender...

a few reference shots in, 1,876 objects out

a smart way to set it up:

- step 1: connect opus 5.5 to blender via mcp, send 2-3 reference shots
- step 2: set a fixed camera matching the reference
- step 3: build the room shell first: walls, windows, blinds
- step 4: add big furniture, check against the reference, lock it
- step 5: fill each zone with clutter: desk, shelves, sill, floor
- step 6: add close-up detail where the camera looks most
- step 7: assign materials, then match the reference light
- step 8: render, compare side by side, fix one thing per prompt

i like to keep the reference camera fixed the whole time

same angle every stage, so you instantly see what’s still missing

[Leafborn | Play Demo on Steam @LeafbornGame]

I made this game and I still can’t stop playing it.

You’re 10,000 leaves, tearing robots apart piece by piece, and every fight feels better than the last. It’s honestly addictive.

Demo’s live in Steam Next Fest right now. Go play it! 🍃

[mrdoob @mrdoob]

https://t.co/vR4kRm6NuX

[mrdoob @mrdoob]

Spent a bunch of tokens over the last few weeks building a fully procedural New York simulation.

https://t.co/YSrGNoIYsK

It's only 784KB gzipped and the code is on github, so you can do whatever you want with it ✌️

[NinjaCodex @N1njaCodex]

PROMPT:
Build a single-page web toy called "Yard Box": a real-time 3D scene rendered as pixel art, showing a courtyard football box behind a four-storey panel apartment block on a floating island. Use three.js r160 (ES module via import map from jsDelivr, plus BufferGeometryUtils). English UI. Original design only — no real brands, places or logos.

RENDERING PIPELINE (the core look)
- Orthographic camera at an isometric angle (elevation ≈31°, azimuth π/4, rotatable in 90° steps with smooth easing; drag sideways or buttons to turn; wheel/pinch zoom 0.7–2.6).
- Render into a small WebGLRenderTarget (half-float, nearest filtering, depth texture) — about 480×300 art pixels at pixel size 3 — no anti-aliasing. Fit the island + the top of the block into the screen area left free by the UI, and snap the camera so the world origin lands on a whole art pixel (static things never shimmer).
- Post-pass ShaderMaterial: texelFetch whole-pixel upscale; depth-based outlines (a pixel > ~0.5 m in front of a neighbour darkens ×0.5); quantize every pixel to the nearest of a fixed ~60-colour palette in OKLab (lightness weighted ×1.35); optional 4×4 Bayer dither (off by default); empty depth = page background colour.
- "Raw 3D" toggle shows the full-resolution MSAA render without steps above. Controls: pixel size slider 2–6, palette / dither / outlines checkboxes.
- Palette: warm concrete greys, cool greys for snow in shade/moonlight, snow and sky blues, turf greens, autumn olives, yellows/oranges/rusts, team reds and blues, skin and wood tones, lamp/window yellows, dusk violets, mint for the boards.

LIGHT AND TIME
- Directional sun/moon with hard BasicShadowMap (2048), hemisphere fill, point lights for two streetlamps (decay 2) and weak doorway lamps. Presets Day / Dusk / Night blended over 1.6 s. Glow materials lerp between on/off colours.
- The block's facade uses an emissive map drawn on a canvas; at night windows glow warm or cool, and random flats switch lights on and off.

WORLD (all boxes, cylinders, extrusions; textures drawn texel by texel on canvases)
- Rounded floating slab: grass top over soil over rock.
- Football box: rounded-rectangle boards (22×12 m, mint green, plank lines, a gate with an open door), netting on posts above the boards at both ends, mini goals (3 m × 2 m) with alpha-tested nets, turf texture with mowing stripes, worn goalmouths/centre and white markings.
- Panel block along one long side: 4 floors, facade drawn panel by panel with windows and curtains, two entrances with canopies, random balconies, roof details; an asphalt strip with two benches and hedges.
- Dollhouse cut: when the camera looks from the block's side, hide the floors, roof, balconies and canopies and show a section of the ground floor instead — low cut walls with a dark cap, rooms with coloured floors, rugs, sofas, tables, wardrobes, TVs (glowing at night), pot plants, stairwells with steps and post boxes; room floors glow warm at night.
- Playground (swing that actually swings, slide, sandbox), birches with white trunks and dark marks, painted-tyre flower beds, a parked car, a washing line with swaying sheets, two lamp posts.

PEOPLE AND DOG
- Kids as rigs of ~17 boxes (bigger heads), posed every frame with two-bone IK: running stride, kicks keyed as back-swing → strike → follow-through, keeper stance, sideways dives, catch/hold/throw, goal celebrations (airplane arms).
- Three grandmothers sitting on the benches (cheer on goals), a small kid on the swing.
- A box-built dog that lives by the benches.

GAME LOGIC
- 3v3: two field players + one keeper per team (RED attacks +x, BLUE −x), first to 5, then a new game; scoreboard with score, game number, status line and games won.
- Kick-off from the centre; open play with dribbling by small touches, passes leading the receiver, shots aimed inside a post at random height with angular error growing with distance and pressure, tackles; players get behind the ball before kicking (feed-forward on a rolling ball), squeeze in at an angle when the ball is against the boards, avoid goal frames; a scramble poke if nobody can reach a still ball for 3.5 s.
- Keepers hold position, come out for slow loose balls, read shots after a 0.1–0.3 s reaction delay with some error, dive, catch or parry.
- Ball physics in 4 substeps: gravity, ground bounce (lower in winter), boards (lively below rail height, dead in the netting above), posts and crossbar, the goal net, legs in the way. Status flashes: GOAL!, SAVED!, WHAT A SAVE!, OFF THE POST!, OFF THE BOARDS, TACKLE.
- Every 100–170 s (or on a button) the dog runs in through the gate, grabs the ball, runs loops away from the kids for ~5 s, drops it and trots home; it gives up if the game resets.

SEASONS (Summer / Autumn / Winter)
- Summer: poplar fluff drifting. Autumn: yellow/orange canopies, leaves falling from the birches, leaf litter on the ground, olive grass. Winter: snow on the ground, roofs, benches, car and boards; bare trees; falling snow and kicked-up snow spray; kids wear hats; the ball turns orange.

INPUT AND UI
- Click the player on the ball to shoot, his teammate to pass, the pitch to make him dribble there; keys S, P, D, Space, Q/E or arrows.
- Layout: title and caption top-left, lit pixel-font scoreboard top-centre, a narrow control rail on the right (Time, Season, Game, View), a hint bottom-left, a fold-out "How it's drawn" panel bottom-right. Fonts: Silkscreen, Rubik, JetBrains Mono. Light and dark themes via CSS tokens; on phones everything stacks under the scene. WebGL2 fallback card.

[NinjaCodex @N1njaCodex]

Claude Opus 5.5 built this little football yard for $6.06

Real 3D with a pixel-art look: the kids decide what to do with the ball, while physics handles shots and bounces. There’s even a dog that steals the ball

[MSB @KeWai386772]

用 Opus 5.5 探索出了第二种风格超级惊艳的作品集&个人网站！

还是从之前分享的那900多种“丹方”中组合的

如果面试官收到这样高质量的简历视觉

很难不给面试机会吧（当然专业度还是第一的hh，视觉与交互体验只是加分项

创作思路已分享

[nicco @0xnicc0]

Opus 5.5 prompt PDF: https://t.co/T0YqwbZkgI

[nicco @0xnicc0]

Opus 5.5 wrote a song AND animated the band playing it, and every drum hit is in sync

this is unreal…

Drop the PDF below into Opus 5.5, you’ll thank me later.

Then save the article, it’s the only Opus 5.5 motion design playbook you’ll need.

[nicco @0xnicc0]

Andrej Karpathy:

"prompting is going away, delete everything, keep the Graph"

in just 69 minutes, he shows how to build graphs and why they're the only layer that survives in the end

Prompts → Agents → Loops → Graphs

a loop keeps one agent running until the job is done

a graph decides which agents exist and what each one passes to the next

building the graph that runs your agents is the skill he spends most of the lecture on

watch it today, then save the full guide on graphs below before everyone starts catching up

[nicco @0xnicc0]

Opus 5.5 promt: https://t.co/mFMiaxPXvI

[nicco @0xnicc0]

Opus 5.5 just made After Effects optional, and nobody in motion design is ready for it

I had to watch this three times…

Drop the PDF below into Opus 5.5, you’ll thank me later.

Then save the article on the Opus 5.5 best practices and usage for motion design.

[nicco @0xnicc0]

Opus 5.5 promt: https://t.co/IFg7itKHVY

[nicco @0xnicc0]

Opus 5.5 is the new After Effects

and nobody's talking about it…

Feed the PDF below to Opus 5.5, you'll thank me later

Then start with the guide below, it's all you need for motion design in Opus 5.5

[兴雨 @YxzRainy]

卧槽😱这样发展下去，视频模型天塌了啊！

同一个提示词，Gemini 4 Argon 和 Claude Opus 5.5 分别搓出了一只在宫殿里潜行的白虎。

这毛发、水雾、光影，还有电影级的画面质感，不知道的还以为是 Veo/即梦生成的！？

众所周知，Gemini 4 Argon 不是视频模型，Claude 也没有原生视频生成能力。

所以，这他🐎到底是怎么做到的？？？ 🤯

[Movez @0xMovez]

I just solved "Robotics Engineering" education with Opus 5.5 + /colorfigures skill

no need to spend $1000+ on a starter kit to learn what a resistor does, just visit my interactive site:

• 13 real boards: Arduino Uno, L298N, Raspberry Pi 5...
• build part by part: pick & place, reflow, solder, plug in
• break them on purpose: reversed caps pop, chips smoke, LEDs burn out

send the skill to your Claude:
> https://t.co/1879rauiu7

try it free, no sign-up ↓