Big P

Stacking @ReadStacked 🥞

Calling 🤙

[Big P @BigP_dev]

@claudeai claude --safely-accept-edits

[Big P @BigP_dev]

@ycocerious More claude code

[Big P @BigP_dev]

@MicrosoftEdge 16, all in iTerm

[Big P @BigP_dev]

@elevenlabs

Can we can some better flow than this?

[Big P @BigP_dev]

@zenorocha Closed for maintenance, back open now

[Big P @BigP_dev]

@ReadStacked Yea! @elonmusk Stay in your lane!

[Big P @BigP_dev]

All of my engineers are AI enabled

The days of coding without AI are gone

[Stacked @ReadStacked]

your company is about to hire AI agents.

not someday. this year.

Microsoft just said over 500,000 AI agents have already been created on their platform. Anthropic's Claude is now running inside Microsoft 365. Salesforce has Agentforce. every major enterprise platform is building AI that does the work, not just answers questions.

the question isn't "will AI take my job."

the question is "am i the person who knows how to manage AI, or am i the person AI is replacing."

one of those people gets promoted. the other gets a meeting with HR.

[Big P @BigP_dev]

https://t.co/MD3ObqJu7O

[Big P @BigP_dev]

1 prompt...

It's crazy what you can do now in < 5 minutes

Here is the Prompt:

Build a Python script `generate_3d.py` that renders Conway's Game of Life as a 3D video with orbiting camera. Pure Python + numpy + Pillow + ffmpeg (no OpenGL/pygame).

**Core architecture:**
- 1080x1080 output, 30fps, H.264 mp4 via ffmpeg raw RGB pipe
- GoL engine: non-toroidal (zero-padded boundaries), tracks cell age as float
- 3D renderer: each alive cell is a column/box at grid position (x, y) with height proportional to age (capped at max_height). Older cells = taller columns.
- Camera orbits the grid center at configurable speed, with gentle sinusoidal height bob
- Painter's algorithm: sort all columns far-to-near, draw back-to-front
- Each column drawn as polygons: 4 side faces (dark/mid shading) + top face (bright). Use Pillow ImageDraw.polygon()
- 3D→2D projection: standard perspective projection with camera basis vectors (forward/right/up), FOV 60°
- Post-process: Gaussian blur glow layer blended at 40% opacity over original
- Sonified audio: ambient drone (40-100Hz) + ethereal pad (220-550Hz) + sub pulse + sparkle, all modulated by live cell count, rendered as 16-bit WAV

**5 color styles** (each has bg color, grid line color, 4-stop color gradient for columns, glow toggle):
- neon: cyan→blue→purple→pink on dark blue-black
- fire: yellow→orange→red→crimson on dark brown
- matrix: light green→green→dark green on black
- ice: white→light blue→blue→navy on dark blue
- vapor: pink→purple→cyan→teal on dark purple

**Floor:** draw grid lines projected in 3D at z=0 in the style's grid_color.

**CLI args:** `--output`, `--duration` (15s default), `--seed` (42), `--grid-size` (60), `--density` (0.3), `--style` (neon/fire/matrix/ice/vapor), `--fps` (30), `--cam-speed` (0.2 rev/s), `--cam-height` (grid*0.7), `--cam-dist` (grid*0.9), `--max-height` (8.0), `--batch` (generates all 5 styles).

**Key details:**
- Column half-width 0.4 cells, height = min(age/15, 1) * max_height + 0.5
- Color interpolation: lerp through 4 color stops based on min(age/30, 1)
- Side faces: 2 dark (50% color), 2 mid (70% color). Top face: 130% color (clamped 255)
- Camera target: grid center at z = max_height * 0.3
- Audio: fade in/out 0.8s, normalize to 0.8 peak

[Big P @BigP_dev]

@AnthropicAI am I just supposed to resize my terminal every few minutes to fix this?

Does anyone have any better solutions?

[Big P @BigP_dev]

@msdev Did you make this using claude code?

[Big P @BigP_dev]

Worth it?

[Big P @BigP_dev]

@tim_cook My openclaw has been asking for an M5 Max!!!

[Big P @BigP_dev]

@vercel Why not just make the caching less complex to manage at scale?

Vite is not hard to manage at scale... NextJS made this it's own problem and now vercel sells the solution...

[Big P @BigP_dev]

@pvergadia Claude makes way better diagrams. This is way too complex

[Big P @BigP_dev]

@alexmolina08 raw PIL frames piped into ffmpeg

[Big P @BigP_dev]

I built a tool that procedurally generates YouTube Poop videos from text prompts. No AI image gen. No external media. Just math, glitch art, and vibes.

First prompt: “What does it feel like to be a large language model?”

It made this.

“I have no mouth and I must predict”

#AIArt #GenerativeArt #YTP #GlitchArt #LLM

[Big P @BigP_dev]

https://t.co/22qcK265r9

[Big P @BigP_dev]

@pmddomingos TL;DR: Almost all Human reasoning is fake

[Big P @BigP_dev]

@pvergadia

[Big P @BigP_dev]

@valigo Nothing new, they just gave it a name. I was already having claude review the code before...