manim-educational-animation

name	manim-educational-animation
description	Creating clean, non-overlapping Manim animations for educational explainers (Gymnasium/school level). Avoids text overlap bugs, guides pacing, and structures single-scene vs multi-scene approaches.
version	1.0.0

Manim Educational Animation — Best Practices

When to Use

Creating clean, non-overlapping Manim animations for educational explainers (Gymnasium/school level). Avoids text overlap bugs, guides pacing, and structures single-scene vs multi-scene approaches.

Overview

This skill contains a reusable operational workflow. Follow the existing task-specific steps and examples in the sections below.

Pitfalls

• Do not hardcode credentials, tokens, or personal secrets.
• Verify external service URLs, paths, and permissions before making changes.
• Keep generated outputs reproducible and record input assumptions.

Verification

• Confirm required inputs and credentials are available.
• Run the smallest safe command or example before scaling up.
• Check produced files, API responses, or plots before reporting success.

Core Lesson Learned (from trial & error)

The #1 bug in educational Manim animations: old text stays visible while new text is written over it. The root cause: Manim renders animations in parallel. Write and FadeOut in the same self.play() overlap.

The Fix: Step Group Pattern

# Each derivation step: Write → Wait → Group → FadeOut → wait(0.5) → Next
s1_lbl = Text("Schritt 1: Variablen trennen", ...)
s1_eq  = MathTex(r"\frac{dT}{T-20} = -k\,dt", ...)
self.play(Write(s1_lbl), Write(s1_eq), run_time=2.0)
self.wait(3.0)

# CRITICAL: group everything from step 1, then remove completely
s1_grp = VGroup(s1_lbl, s1_eq)
self.play(FadeOut(s1_grp), run_time=1.2)
self.wait(0.5)   # Screen must clear before next Write

# Now screen is clean — write step 2
s2_lbl = Text("Schritt 2: Integrieren", ...)
s2_eq  = MathTex(r"\ln|T-20| = -kt + C", ...)
self.play(Write(s2_lbl), Write(s2_eq), run_time=2.0)

Why wait(0.5): Manim's animation queue needs a frame gap to clear the old mobject. Without it, the next Write starts before FadeOut finishes.

Why not ReplacementTransform: It can leave LaTeX artifacts and bounding-box ghosts. FadeOut of a VGroup is reliable.

Consistent Positioning

All derivation steps should use the same anchor position — otherwise they drift:

# All step labels at the same Y
s1_lbl = Text("Schritt 1: ...", ...).to_edge(LEFT, buff=0.8).shift(DOWN * 2.0)
s2_lbl = Text("Schritt 2: ...", ...).to_edge(LEFT, buff=0.8).shift(DOWN * 2.0)

# All step equations directly below their label
eq.next_to(s1_lbl, DOWN, buff=0.5)

Single-Scene vs Multi-Scene

Single-scene (one class with _teilN() methods)

• Best for: focused topic, 1-2 minute explainers, school material
• Pros: no stitching needed, easier to avoid overlap, simpler code
• Cons: long render time for one scene
• Target: ~60 animations, ~2 min for a complete topic

Multi-scene (stitched)

• Best for: 5+ min videos, modular content, different visual layouts per section
• Pros: scenes render in parallel, each scene has its own visual identity
• Cons: more complex pipeline, overlap bugs harder to track across scenes
• Target: ~15 animations per scene, ~15 sec/scene

Pacing for Educational Content

Audience	Target length	Animations	Per-step wait
Gymnasium/School (Klasse 11)	1.5–2.5 min	~60–80	2–3 sec
University	3–5 min	~80–120	1.5–2 sec
General public	5–8 min	~120–200	1–1.5 sec

Color Palette (Dark Background)

BG      = "#0D1117"   # dark background
BLUE    = "#58C4DD"   # main equations, titles
GREEN   = "#83C167"   # solutions, positive results
YELLOW  = "#FFFF00"   # highlights, key values
WHITE   = "#EAEAEA"   # body text
DIM     = "#7A8599"   # axes, labels, secondary text
RED     = "#FF6B6B"   # negative slopes, warnings
PURPLE  = "#C792EA"   # step indicators (Schritt 1, 2, 3)
MONO    = "DejaVu Sans Mono"  # font for ALL text

Common Bugs

Bug	Cause	Fix
Text overlaps	FadeOut + Write in same play()	Add wait(0.5) after every FadeOut
Steps drift vertically	No consistent anchor position	All steps use same .shift(DOWN * 2.0)
LaTeX artifacts	ReplacementTransform on MathTex	Use FadeOut + Write instead
Row-by-row reveals too slow	6+ reveals per field × 0.2s each	Use Create(pfeile) for the whole group at once
9-part video too long	Too many scenes, too much explanation	Target 5 parts max, 2 min total
199 animations for 2 min	Too many wait() and small reveals	Combine into fewer, bigger animations
log(x, base) not defined	Manim has no built-in log function	Use np.log(x) / np.log(base) with import numpy as np
NameError: name 'log' not defined	Same as above — log from math module not available	Same fix: np.log(x) / np.log(base)

Rendering

# Draft (iterate fast)
manim -ql script.py SceneName

# Production
manim -qh script.py SceneName

# Preview frame
manim -ql --format=png -s script.py SceneName

# Stitch scenes
ffmpeg -y -f concat -safe 0 -i concat.txt -c copy final.mp4