Communitygithub.com

Render Rigged 3D

Complete 18-Joint 3D Skeletal Rigged Procedural Animation, Sequential Frame-by-Frame Video Inspection, and From-Scratch Audio Production Skill following the user's exact folder-based instruction workflow.

What is Render Rigged 3D?

Render Rigged 3D is a Claude Code agent skill that complete 18-Joint 3D Skeletal Rigged Procedural Animation, Sequential Frame-by-Frame Video Inspection, and From-Scratch Audio Production Skill following the user's exact folder-based instruction workflow.

Works with~Claude Code~Codex CLI~Cursor
npx skills add https://github.com/abdulahadofficial432-arch/Gemini_Argon_Video_Skills/tree/HEAD/Render%20Rigged%203D%20Videos%20Skill/Skills/Render%20Rigged%203D

Ask in your favorite AI

Open a new chat with this agent skill pre-loaded.

Preview

From the skill README

Train Video 3D Sheet 01 (0s–25s): Crouched Hacksaw on 3D Rail, Steam Locomotive Pass-By, Parallel Canyon Bridges, Union Station Gold Bullion & Purple-Faced Boss
Train Video 3D Sheet 02 (25s–50s): Side-Profile Macro 3D Rail Joint with Vernier Caliper, Welded Rival Track, Vandergold Parlor Car, Jolting Coach & Unobstructed Newspaper
Train Video 3D Sheet 03 (50s–76.67s): Silver Coin in May Gap, White Paper in July Gap, Crushing Rival Rails, Beach Umbrella Tycoon & Steel Arch Bridge Plunge
Japan Job Culture 3D Sheet 01: 3D Office Environments, Reclining Desk-Propped Poses & Articulated 18-Joint Walk Cycles
Japan Job Culture 3D Sheet 02: Dynamic 3D Camera Angles, Expressive Spherical Heads & Prop Interactions
Japan Job Culture 3D Sheet 03: Climax 3D Sequences & Zero-Caption Studio Finish

Documentation

Render Rigged 3D — Complete Skill & User Instruction Workflow

1. How the User Gives Instructions (The User's Workflow Method)

Whenever the user starts a new video recreation task, they follow this exact workflow:

  1. New Project Folder in Main Workspace (C:\Users\WayWe Work-Station\Downloads\Mate\<Project Folder>\):
    • The user creates a new folder inside the main workspace folder named after the video topic (for example: "Train Video", "Job Culture of japan", "Railroad Video", "The Window That Breaks Japan's Gambling Law", "80 Victories in a Plane Nobody Could See").
  2. Reference Video + Voiceover Inside the Project Folder:
    • Inside that project folder, the user places:
      • A Reference Video (e.g., "Example video 1.mp4", "Jap vid.mp4", "My video 0.mp4", "My video 2.mp4").
      • A Voiceover Audio File placed either directly in the project folder or inside a voiceover subfolder ("VoiceOver", "VoiceOver 01", "VoiceOver 02"), named e.g. "Voiceover.mp3" or "Audio.mp3".
  3. The User's Prompting Pattern:
    • The user asks you to:
      1. Read and analyze the reference video frame-by-frame (Frame 0 to Frame N-1) in strict chronological order without skipping or sampling frames, checking every single frame for:
        • Exact frame number & timestamp (frame / fps)
        • Camera position, angle, movement (dx, dy, zoom, tilt, roll), and shot framing
        • Every visible person, position, movement, facial expression, body pose, hands/fingers, and clothing/accessories
        • Objects, vehicles, background/environment, lighting, shadows, text/logos/signage, and visual effects
        • Frame-to-frame comparison (what changed, what stayed the same, direction/amount of movement, new/removed objects, morphing, flickering, or generation artifacts)
      2. Recreate the exact same video 100% from scratch using the user's Voiceover file, matching the exact visuals, hyper-active audience-grabbing animation, character rigging/movements, camera choreography, and clear visual explanations — rendered in the locked 18-Joint 3D Skeletal Rigged Procedural Animation style with zero captions and 100% from-scratch SFX and Music stems.

2. Visual Reference & Rendered 3D Contact Sheets (screenshots/)

A. Train Video — 18-Joint 3D Rigged Recreation & Macro Thermal Physics (train_rigged_3d_sheet_01..03.jpg)

Train Video 3D Sheet 01 (0s–25s): Crouched Hacksaw on 3D Rail, Steam Locomotive Pass-By, Parallel Canyon Bridges, Union Station Gold Bullion & Purple-Faced Boss

Train Video 3D Sheet 02 (25s–50s): Side-Profile Macro 3D Rail Joint with Vernier Caliper, Welded Rival Track, Vandergold Parlor Car, Jolting Coach & Unobstructed Newspaper

Train Video 3D Sheet 03 (50s–76.67s): Silver Coin in May Gap, White Paper in July Gap, Crushing Rival Rails, Beach Umbrella Tycoon & Steel Arch Bridge Plunge

B. Job Culture of Japan — 18-Joint 3D Rigged Corporate Animation (japan_rigged_3d_sheet_01..03.jpg)

Japan Job Culture 3D Sheet 01: 3D Office Environments, Reclining Desk-Propped Poses & Articulated 18-Joint Walk Cycles

Japan Job Culture 3D Sheet 02: Dynamic 3D Camera Angles, Expressive Spherical Heads & Prop Interactions

Japan Job Culture 3D Sheet 03: Climax 3D Sequences & Zero-Caption Studio Finish


3. Locked 18-Joint 3D Skeletal Rigged Animation Architecture

All videos created with Render Rigged 3D must use a 100% from-scratch 3D procedural engine (PyCairo + NumPy + FFmpeg at 1080×1920, 30.00 fps) with zero 2D stick-figure shortcuts and zero polygon glitches:

A. Glitch-Free 3D Camera & Polygon Pipeline (Camera3D, clip_and_project_poly, draw_box_3d)

  1. True 3D Perspective Camera (Camera3D):
    • Supports world-space pos = (x, y, z), target = (tx, ty, tz), fov, Dutch-angle roll, organic handheld sway (math.sin(t * 2.5) * 4.0), fast dolly push-ins, whip pans, crane tilts, and high-frequency jolt shakes.
  2. Sutherland-Hodgman 3D Near-Plane Polygon Clipping (NEAR_Z = 45.0):
    • Mandatory Anti-Glitch Rule: Never project 3D vertices that cross behind the camera by simply clamping z_cam. Always clip every 3D polygon (clip_and_project_poly) against the camera near plane z_cam >= 45.0 before perspective division so long rails, floors, walls, and bridges never fold or warp across the screen.
  3. Deterministic 3D Outward-Normal Back-Face Culling:
    • Mandatory Anti-Flicker Rule: In draw_box_3d, never sort box faces by average vertex depth (which causes adjacent faces to swap depth order and flicker as the camera moves). Always use outward-normal back-face culling (np.dot(face_normal, cam.pos - face_center) > 0) and subdivide long geometry (such as railroad rails and ties) into back-to-front segments.

B. 18-Joint 3D Articulated Character Rig (draw_rigged_character_3d)

Every character is animated in 3D world space using an 18-joint skeletal hierarchy:

  • Torso & Head Chain (5 joints): pelvis -> spine -> chest -> neck -> head
  • Left & Right Arm Chains (8 joints): clavicle_L/R -> shoulder_L/R -> elbow_L/R -> wrist_L/R -> hand_L/R
  • Left & Right Leg Chains (6 joints): hip_L/R -> knee_L/R -> ankle_L/R -> toe_L/R
  • Curved 3D Articulated Limbs (draw_curved_limb_3d):
    • Smooth cubic-Bezier articulated limbs through (shoulder, elbow, wrist) and (hip, knee, ankle) with shaded 3D spherical hands and 3D shoes.
  • 3D Spherical Head, Hats & Facial Rig:
    • Radial-gradient 3D shaded sphere head with perspective-foreshortened eyes, pupils, eyebrows, and mouth expressions (neutral, grin, angry, rage, panic, worried, squint, closed_smile) that rotate across the 3D sphere surface according to (head_yaw, head_pitch, head_roll).
    • Includes periodic eye blinking, chest breathing (math.sin(t * 4.5)), sweat droplets, role-specific 3D hats/vests/suits (engineer, boss, tycoon, foreman, veteran, worker), and special visual states (e.g., purple_anger transitioning the head to violet-purple with 3D white steam puffs erupting from both ears).
  • Unobstructed Front-Prop Layering Rule:
    • Whenever a character holds a readable front prop (such as a newspaper, sign, or document), draw the 3D upper/lower arms first (so they tuck naturally behind the prop), draw the prop second in front of the chest, and draw the white 3D spherical hands third gripping the left and right outer edges so arm lines never cross over text or illustrations.

C. Crystal-Clear Visual Storytelling & Macro Physics Views

  • Every concept described in the voiceover must be visually explained on screen through clear 3D staging, props, and dedicated macro shots:
    • Macro 3D Cutaways & Gauges: When explaining measurements, physics, or mechanisms (e.g., thermal rail expansion), build a dedicated side-profile 3D macro view (draw_macro_rail_joint_side_3d) featuring 3D I-beam profiles, bolted fishplates, a sliding metal Vernier caliper clamping the exact gap, a rising mercury thermometer, a flipping calendar (MAY -> JUNE -> JULY), physical comparison objects dropping into the gap (silver coin vs. thin sheet of paper), and force/stress arrows (==><==).
    • Multi-Sub-Shot Visual Pacing: Break longer narrative sentences into distinct visual sub-shots (e.g., Sub-shot A showing the macro mechanism -> Sub-shot B showing the characters reacting; or Sub-shot A showing the newspaper headline -> Sub-shot B showing passengers walking toward 'ALL ABOARD!') so the visual pacing stays hyper-active and engaging.

3. Step-by-Step Execution Protocol

Whenever triggered on a new project folder:

Step 1: Complete Sequential Frame-by-Frame Inspection (Frame 0 to Frame N-1)

  1. Run analyze_every_frame.py on the reference .mp4 video to inspect every single frame from Frame 0 to Frame N-1 in chronological order without skipping frames.
  2. Save the complete sequential per-frame report to <Project Folder>\Frame_By_Frame_Complete_Analysis_<N>_Frames.md.
  3. Extract 6×4 contact sheet .jpg grids (1 frame per second) and a 48kbps .mp3 audio preview into scratch/, and inspect every contact sheet and audio file using read_file (never call read_file on a raw .mp4).

Step 2: Synthesize 100% From-Scratch Audio Stems Synced to the User's Voiceover

  1. Measure the exact duration (VO_DURATION) of the user's voiceover file (Voiceover.mp3 / Audio.mp3) and convert it to 44.1 kHz 16-bit stereo Voiceover.wav and Voiceover.mp3 in <Project Folder>\.
  2. Write and run build_<project>_audio_stems.py using NumPy + SciPy (scipy.signal.butter) to synthesize:
    • SFX.wav & SFX.mp3: 100% from-scratch physical-modeling foley synced to every visual action (e.g., hacksaw strokes, steam chugs, whistle Doppler, metallic clanks, coin/paper drops, thermal groans, crashes, explosions).
    • Music.wav & Music.mp3: 100% from-scratch multi-act cinematic/atmospheric score matched to the emotional arc.
    • Full_Mixed_Audio.wav & Full_Mixed_Audio.mp3: Master mix combining Voiceover + SFX + Music with automatic envelope ducking so the voiceover remains crystal clear.

Step 3: Build & Preview-Verify the 18-Joint 3D Rigged Render Script

  1. Write render_<project>_rigged_3d_from_scratch.py mapping every reference shot and visual explanation onto the exact VO_DURATION (TOTAL_FRAMES = int(math.ceil(VO_DURATION * 30))).
  2. Mandatory Pre-Render Visual Check (preview mode):
    • Run python render_<project>_rigged_3d_from_scratch.py preview to generate 3 contact sheets (sheet_01.jpg, sheet_02.jpg, sheet_03.jpg, 72 tiles evenly spanning 0.0s .. VO_DURATION).
    • Inspect all 3 contact sheets via read_file to verify:
      • Zero 3D polygon warping or box-face flickering
      • Zero prop/limb overlaps (hands on outer edges of newspapers/signs, hats/umbrellas properly layered)
      • Every story beat and physical mechanism is visually clear
      • Zero voiceover captions

Step 4: Render Full 30-FPS 1080×1920 Video & Verify Deliverables

  1. Run python render_<project>_rigged_3d_from_scratch.py to render all TOTAL_FRAMES frames at 1080×1920 (30.00 fps) and mux with Full_Mixed_Audio.wav.
  2. Verify that <Project Folder>\ contains all required deliverables:
    • Final_Video_No_Captions.mp4 (Full Mixed Audio, Zero Captions)
    • Pure_3D_Procedural_From_Scratch_No_Captions.mp4 (Pure 3D Rigged Render Copy)
    • Video_Only_No_Captions_Muted.mp4 (Clean Muted Video Master)
    • Frame_By_Frame_Complete_Analysis_<N>_Frames.md
    • Voiceover.wav & Voiceover.mp3
    • SFX.wav & SFX.mp3
    • Music.wav & Music.mp3
    • Full_Mixed_Audio.wav & Full_Mixed_Audio.mp3
    • scripts/ containing the complete Python generator scripts.

Individual skills in this repo

This repo contains 3 individual skills — each has its own dedicated page.

Related Skills