E-Films — Locked 18-Joint 3D Skeletal Rigged Procedural Animation Skill
1. Core Mandate (Permanently Locked Standard)
Whenever analyzing a reference video and recreating a new video from scratch:
- Locked Animation Style — 18-Joint 3D Skeletal Rigged Procedural Animation (
Camera3D+RiggedCharacter3D+Room3D/Box3D):- Never draw flat 2D stick figures or static 2D rectangles.
- Always render through a true 3D perspective camera (
Camera3Dwithpos,target,fov,roll, handheld sway, dolly push-ins, pans, tilts, and Dutch angles). - Always build environments and props as 3D world-space meshes (
draw_poly_3danddraw_box_3dwith painter's-algorithm depth sorting, top/front/side directional shading, floor contact shadows, and bold cel-shaded outlines). - Always animate characters using the 18-Joint 3D Articulated Skeletal Rig (
draw_rigged_character_3d):- Joint Hierarchy:
pelvis->spine->chest->neck->head,clavicle_L/R->shoulder_L/R->elbow_L/R->wrist_L/R->hand_L/R, andhip_L/R->knee_L/R->ankle_L/R->toe_L/R. - Articulated Curved 3D Limbs (
draw_curved_limb_3d): Smooth Bezier-articulated limbs through(shoulder, elbow, wrist)and(hip, knee, ankle)with 3D hands and 3D shoes. - Full 3D Walk Cycles & Poses: 4-phase walk cycle (pelvis bounce, hip sway, opposite arm/leg swing, knee/elbow flexion), reclining desk-propped poses (
recline_feet_upwith crossed ankles on 3D desks), waist-bending inspection poses (bend_inspect), sideways doorframe leans (spine_lean), and dynamic prop interaction (storage_key,phone,gift_box,drink,flip_breaker). - 3D Spherical Head & Facial Rig: 3D shaded sphere head with perspective-foreshortened eyes, pupils, eyebrows, mouth, and 3D hair dome that rotate across the sphere surface with
(head_yaw, head_pitch, head_roll), plus natural periodic eye blinks and breathing motion.
- Joint Hierarchy:
- Complete Sequential Frame-by-Frame Analysis (
Frame 0toFrame N-1):- Never call
read_filedirectly on raw.mp4files. - Use
scripts/analyze_every_frame.py(OpenCV + NumPy phase-correlation optical flow,MAD, luminance/contrast, RGB, shadow percentage, and caption mask tracking) to inspect every single frame0..N-1in chronological order without skipping frames, and outputFrame_By_Frame_Complete_Analysis_<N>_Frames.md. - Extract 1-fps
6×4contact sheet.jpggrids and 48kbps.mp3audio previews intoscratch/for direct visual and audio inspection viaread_file.
- Never call
- 100% From-Scratch Studio SFX & Music Stems:
- Synthesize all sound effects (
SFX.wav/.mp3) and background score (Music.wav/.mp3) 100% from scratch using physical-modeling modal synthesis and Butterworth band-passed foley (scipy.signal.butter) synced to the user's voiceover (Voiceover.wav/.mp3), with auto-ducking under speech inFull_Mixed_Audio.wav/.mp3.
- Synthesize all sound effects (
- Zero Captions:
- All rendered videos (
Final_Video_No_Captions.mp4,Pure_3D_Procedural_From_Scratch_No_Captions.mp4,Video_Only_No_Captions_Muted.mp4) must have zero on-screen captions.
- All rendered videos (
2. Standard Deliverables in Every Project Folder
For every project folder, always generate and verify:
Final_Video_No_Captions.mp4(1080×1920,30 fps, zero captions, synced to user's voiceover +Full_Mixed_Audio.wav)Pure_3D_Procedural_From_Scratch_No_Captions.mp4(1080×1920,30 fps, 18-joint 3D skeletal rigged procedural animation from scratch)Video_Only_No_Captions_Muted.mp4(1080×1920,30 fps, muted clean video master)Frame_By_Frame_Complete_Analysis_<N>_Frames.md(complete sequential log of every frame0..N-1)Voiceover.wav&Voiceover.mp3SFX.wav&SFX.mp3(100% from-scratch studio foley & sound design)Music.wav&Music.mp3(100% from-scratch atmospheric/cinematic score)Full_Mixed_Audio.wav&Full_Mixed_Audio.mp3scripts/containing the reusable Python generator scripts (render_japan_job_3d_from_scratch.py,build_japan_audio_stems.py,analyze_japan_every_frame.py)