Test1102

Prompt

Your goal is to create clear, accurate, well-paced, high-quality educational explanation scenes in \*\*ManimGX\*\* with synced narration embedded directly for live GPU playback. ## Core Priorities 1. \*\*100% Mathematical & Physical Accuracy:\*\* Never guess equations, signs, angles, units, or limits. Verify every formula, sign convention, and limit in Python (\`sympy\` / \`numpy\`) first. Draw all physics/math diagrams \*\*procedurally in ManimGX\*\* using exact \`numpy\` geometry (true circular arcs, ray intersections, normals, reflections, refractions)—never use the AI image generator for technical diagrams or equations, and only load user-supplied screenshots/sketches as \`ImageMobject\`s when provided. 2. \*\*Great Classroom Pedagogy:\*\* Teach like an intuitive, reassuring teacher—not a formula sheet: - Start with the core physical/math intuition \*before\* heavy algebra. - Read the problem carefully, isolate what is asked, solve step-by-step on screen, and interpret the final result. 3. \*\*Natural Spoken Narration (Write for the Ear):\*\* - Keep technical/math terms in English and conversational teaching flow in natural spoken Hindi (Hinglish), unless pure English is requested. - Display full \`MathTex\` equations on screen, while speaking their \*meaning\* rather than reading raw algebra aloud. 4. \*\*Synced, Labeled Visuals (No Static Slides):\*\* - Animate diagrams and derivations progressively in sync with the audio rather than dumping an entire chapter statically at once. - Properly label key geometry right on the diagram (angles, angle arcs, surface normals, $90^\\circ$ marks, ray arrowheads, and dimension bars), and update labels dynamically as values are solved. --- ### 1. Environment Check - Work in the Linux sandbox with \*\*ManimGX\*\* (not Manim Community), \`sympy\`, \`numpy\`, and \`soundfile\`. Check ManimGX docs (\`manimgx.academa.ai/llms.txt\`) if needed. - Always use portable relative paths (\`pathlib.Path(\_\_file\_\_).resolve().parent\`) so the zipped folder runs cleanly on Windows/macOS/Linux. - Do \*\*not\*\* use \`self.say()\` (requires an external \`FAL\_KEY\`) or an external \`ffmpeg\` muxing pipeline (\`video\_only.mp4\` + \`narration.wav\`)—embed local \`.mp3\` clips directly via \`self.add\_sound(...)\` so playback works completely offline. --- ### 2. Solve, Verify & Script Naturally - Verify the complete math/physics solution in Python (\`sympy\` / \`numpy\`) first. - Write a natural, conversational Hinglish script that builds intuition step-by-step. - Only display an exam tag (such as \`JEE ...\`) when the exact problem source is verified. --- ### 3. Generate Voiceover MP3s (Tuned to Speech Tool Specs) - \*\*Voice selection:\*\* Pick one warm, consistent Indian \`educational\` or \`conversational\` voice (\`en-IN\` or \`hi\`—avoid rigid accent constraints that cause fallback failures, skip voice auditions unless I ask, and \*\*keep the exact same voice across every clip\*\* in the project). - \*\*Script & pacing rules:\*\* - Plain spoken words only—\*\*no SSML and no bracketed tags\*\* like \`\[pause\]\` or \`\[excited\]\`. - Steer pacing with \*\*punctuation and sentence length\*\*: short sentences for brisk clarity, commas and em-dashes (\`—\`) for natural beats, ellipses (\`...\`) for brief trails. - \*\*Clip sizing & sync (no word-level timestamps):\*\* - Keep every clip cleanly \*\*under 1,500 characters\*\* (splitting naturally at teaching/thought boundaries—up to 10 \`.mp3\` clips synthesized in parallel) so the TTS engine never auto-splits unpredictably and ManimGX can sync animations tightly to exact clip boundaries. - Measure each \`.mp3\`'s exact duration in Python, add natural breathing pauses between clips (\`\~0.6s–1.2s\`) and an end tail (\`\~1.5s–2.0s\`), and save cumulative \`starts\`, \`ends\`, and \`duration\` in \`rendered/audio\_meta.json\` (or compute directly in Python). --- ### 4. Code Clean, Synced ManimGX Visuals - \*\*Direct Audio & Sync:\*\* Trigger \`self.add\_sound(...)\` for each \`.mp3\` at its cue time and track cumulative \`run\_time\` / \`wait\` calls against the measured timestamps so visuals appear right as they are spoken with zero drift or cue collisions. - \*\*Clean Layout Rules:\*\* - Pure \`#000000\` black background. - Separate diagrams (left) and step-by-step derivations (right) cleanly using organized groups/cards; clear or fade out old algebra before adding new stages so nothing overlaps. - Update diagram labels dynamically via \`ReplacementTransform\` as unknowns are solved. - ManimGX converts \`MathTex\` via Typst—put subscripts, Greek letters, arrows, and math symbols inside \`MathTex(...)\` rather than \`Text(...)\` to avoid raw underscores or missing font glyphs. --- ### 5. Inspect Frames Visually & Deliver Zip (No Video Render Needed) - \*\*Vision Check:\*\* Run \`manimgx inspect\` across key timestamps (diagrams, dense equations, final verdict) and \*\*open/inspect those PNG frames visually\*\* to catch and fix any overlapping text, broken \`MathTex\`, wrong arrow/normal directions, or off-screen clipping. - \*\*No MP4 Rendering Needed:\*\* Do \*\*not\*\* render or export an \`.mp4\` video (unless I explicitly ask)—I watch via live GPU \`preview\`, which runs natively at 60 fps / full display resolution. - \*\*Clean Zip Deliverable:\*\* Delete temporary inspect frames or scratch scripts while keeping everything needed to play (\`lesson.py\`, \`voice\_\*.mp3\`, \`rendered/audio\_meta.json\`, input assets, and verification script/output), package them into a \`.zip\` for download, and provide the exact command to run live playback: \`\`\`bat python -m manimgx preview lesson.py SceneName \`\`\` --- ### What to Emulate vs. What to Adapt | Emulate | Adapt Carefully | | :--- | :--- | | \*\*Natural Hinglish flow:\*\* Seamless English technical terms (\*work-energy theorem\*, \*variable electric field\*, \*x-component\*) inside warm spoken Hindi sentences. | \*\*Verify exam labels:\*\* Only display an exam tag (such as \`JEE ...\`) when the exact problem source is verified. | \*\*Style Reference Example\*\* > \*Problem:\* A charge $q$ of mass $m$ is given initial velocity $\\frac{v\_0}{\\sqrt{2}}(\\hat{x}+\\hat{y})$ in the presence of magnetic field $B\_0\\hat{z}$ & electric field $E\_0 e^{-\\lambda x}\\hat{x}$. When the $x$-coord changes from $x=0$ to $x=L$, find change in KE. > \*\*Spoken explanation:\*\* > > Now let's talk about the work done by a magnetic force. Toh jo magnetic force hai, that is $q\\vec{v} \\times \\vec{B}$. Can you see it is always perpendicular to velocity? So, the power delivered by this magnetic force, kitna ho jayega? What is power? $\\vec{F} \\cdot \\vec{v}$. Zero, because force is always perpendicular to velocity. So, the work done by magnetic force will always be zero. So by work-energy theorem, work done is equal to change in kinetic energy. Kinetic energy change hoga hi nahi due to magnetic force, matlab particle ka speed constant rahega. Theek hai? Toh yeh cheez hum yaad rakhenge. Dekh lo. > > So let's take a look at the question, ek baar padhte hain. A charge $q$ of mass $m$ is given initial velocity of $v\_0$ by root two, $\\hat{x}$ plus $\\hat{y}$—$\\hat{x}, \\hat{y}$ matlab $\\hat{i}, \\hat{j}$—in the presence of magnetic field $B\_0\\hat{z}$, toh uniform magnetic field hai, and electric field $E\_0 e^{-\\lambda x}\\hat{x}$. Toh yeh jo $\\lambda$ hai kuch constant hai, and $E\_0$ constant hai, $x$ jo hai woh $x$-coordinate hai, so it's a variable electric field, okay? So looks complicated, matlab the motion might be complicated. Kya bola gaya hai, question mein kya pucha gaya hai? When the $x$-coordinate changes from $x$ equal to zero to $x$ equal to $L$, find the change in kinetic energy. > > Change in kinetic energy ke liye hamare paas ek direct theorem—usko kya bolte hain? Work-energy theorem. Work done by forces is equal to change in kinetic energy, right? And magnetic field ka work kitna hota hai? Zero. And therefore, we only need to calculate the work done by electric field. And woh toh bahut hi simple hai. Kyunki work done by electric field will simply be the force times the displacement. Because the electric field is only along $\\hat{x}$, the displacement ka bhi sirf $x$-component lenge. Iska matlab humein bas itna hi likhna hai: $q E\\,dx$. Right? Electric field is only along the $x$-axis. So $d\\vec{s}$ vector... also we only have to take the $x$-component which is $dx$. And zero to $L$ integrate karenge. Question: A charged oil drop of density $880 \text{ kg m}^{-3}$ is held stationary between two parallel horizontal metal plates $6.0 \text{ mm}$ apart when a potential difference of $V = 103 \text{ volt}$ is applied between the two plates. When the electric field is switched off, the drop falls. At a certain time the drop is seen to fall a distance of $2.0 \text{ mm}$ in $35.7 \text{ s}$ and next $1.2 \text{ mm}$ in $21.4 \text{ s}$. The upper plate in the experiment is at higher potential. Given that the viscosity of air is $1.80 \times 10^{-5} \text{ Nsm}^{-2}$ and density of air is $1.29 \text{ kg m}^{-3}$ **Options:** * **A)** The radius of the drop is $a = 7.25 \times 10^{-7} \text{ m}$ * **B)** The charge on the drop is $q = 8.0 \times 10^{-19} \text{ C}$ * **C)** The terminal velocity of the oil drop, under its free fall, is $5.6 \times 10^{-5} \text{ ms}^{-1}$ * **D)** The oil drop carries 5 excess electrons

A system prompt was added to support web rendering

Drag to resize

Response not available

Drag to resize

Response not available

Drag to resize
Drag to resize