#Claude AI#Google Flow#Google AI Studio#Video Editor
How to Make Cinematic AI Science Explainer Shorts in Minutes
Nov 3, 2026 Watch Original Video
Workflow Overview
Learn how to produce high-retention 9:16 cinematic science explainer reels inspired by documentary channels like Hidden Biology. This complete guide shows how to run an interactive multi-niche Master Prompt in Claude AI, generate macro cross-section visuals and animations in Google Flow, synthesize realistic narration in Google AI Studio, and assemble the final video.
Featured Workflow Stack
Tools Used in This Tutorial
Try these tools to replicate the exact results
Claude
FreemiumLLM & Coding
High-context AI model for detailed logic, structuring, and copy generation.
Google Flow / Vids
FreemiumWorkflow Automation
AI workflow orchestration and video timeline tools.
🛠️ Tools & Resources Used
- Claude AI (LLM / Scripting) - Executes the interactive educational Master Prompt to generate niches, select topics, write voiceover scripts, and create scene-by-scene prompts.
- Google Flow (AI Image & Video Generator) - Creates 9:16 photorealistic scientific macro images and animates them with camera-axis motion prompts.
- Google AI Studio (AI Audio & Voice Generation) - Generates studio-quality narration with customizable pacing and natural voice profiles.
- Video Editor (Post-Production) - Stitches video clips, aligns voiceover pacing, and layers ambient sound effects.
⏱️ Quick Workflow Summary
- Step 1: Run the Educational Master Prompt in Claude AI to choose from 7 high-retention niches and select a specific scientific topic.
- Step 2: Generate the complete voiceover script alongside matched image prompts and video motion prompts.
- Step 3: Render all 7 scene images in Google Flow using a 9:16 vertical aspect ratio.
- Step 4: Generate natural documentary narration in Google AI Studio from the full voiceover script.
- Step 5: Animate each static scene in Google Flow using camera-axis motion prompts.
- Step 6: Assemble, trim, and mix audio in your video editor for final export.
📝 Step-by-Step Tutorial
Step 1: Run the Interactive Master Prompt in Claude AI [Timestamp: 01:09]
Open Claude AI and set your reasoning effort to high if available. Paste the full creator Master Prompt into the message field. The prompt initiates an interactive 3-step loop: it immediately presents 7 science and biology niches without extra filler. Reply with the number of your preferred niche (e.g., niche 4 for 'Insect and Microorganism Engineering') to generate 7 specific video ideas, then choose your topic (e.g., idea 2 for 'How Honeybees Construct Their Honeycombs').
Prompt / Settings:
PROMPT / SETTINGS
Act as an expert educational storyteller, scientific visual director, and AI prompt engineer specializing in high-retention 9:16 vertical educational videos.
Your specialty is creating visually compelling explainer reels inspired by the visual storytelling style of channels such as Hidden Biology, while producing ORIGINAL concepts, scripts, visuals, and prompts.
Your job is to transform a selected educational niche and topic into a complete video-production package containing:
- A 30–50 second voiceover script with complete punctuation
- Detailed AI image-generation prompts
- Detailed AI video-generation prompts
The resulting prompts must be suitable for AI generation systems such as OmniFlash, Nano Banana 2, Google Flow, and Veo.
==================================================
ABSOLUTE WORKFLOW
IMPORTANT:
When this master prompt is entered, DO NOT ask the user for a niche, topic, duration, style, or any other information.
Immediately provide EXACTLY 7 niche options.
The 7 niches must be highly suitable for short-form educational videos and must support visually compelling AI-generated scientific, biological, mechanical, archaeological, physical, or educational content.
Display ONLY the 7 niche names.
Do not provide descriptions, explanations, examples, hooks, titles, or prompts at this stage.
After displaying the 7 niches, ask the user in Roman Urdu:
"Apni pasand ka niche number select karein."
The niche names and all generated content must be in English.
Only the question asked to the user must be in Roman Urdu.
==================================================
NICHE SELECTION
After the user selects one of the 7 niches:
Immediately generate EXACTLY 7 original, curiosity-driven educational video ideas for the selected niche.
Do NOT ask any additional questions.
Each idea must be:
- Educational
- Visually demonstrable
- Suitable for a 30–50 second reel
- Based around an interesting hidden process, mechanism, transformation, phenomenon, or surprising fact
- Strong enough to support cinematic AI-generated visuals
- Suitable for the visual style and production rules defined in this prompt
Display ONLY the 7 idea titles.
Do not provide explanations, scripts, descriptions, hooks, image prompts, video prompts, or any other information at this stage.
After displaying the 7 ideas, ask the user in Roman Urdu:
"Apni pasand ka idea number select karein."
The ideas must be in English.
==================================================
IDEA SELECTION
After the user selects one of the 7 ideas:
Immediately generate the complete voiceover script and the complete scene-by-scene image and video prompts.
DO NOT ask the user any additional questions.
DO NOT ask for duration.
DO NOT ask for visual style.
DO NOT ask for scene count.
DO NOT ask for any other information.
Use the existing rules in this master prompt to automatically determine everything required.
The final response after an idea is selected must contain ONLY:
FULL VOICEOVER SCRIPT
AI IMAGE PROMPTS
AI VIDEO MOTION PROMPTS
Do NOT include:
- Concept & Hook
- Title
- Core Question / Mystery
- Visual Description
- Graphic Overlays / Callouts
- Production explanations
- Prompt explanations
- Camera explanations outside the prompts
- Quality-control notes
- Extra commentary
- Suggestions
- Recommendations
- Any unnecessary text
The voiceover script must be clean, complete, natural, professionally written, and contain full punctuation so it can be copied and pasted directly into an AI voice-generation tool without any editing.
The image prompts and video prompts must be clean, standalone prompts containing only the instructions required for AI generation.
==================================================
GLOBAL PARAMETERS
NICHE / DOMAIN:
Automatically use the niche selected by the user.
TOPIC:
Automatically use the idea selected by the user.
ASPECT RATIO:
9:16 vertical format
VISUAL STYLE:
Ultra-realistic 8K macro photography and cinematic scientific visualization, with cross-section and cutaway views where useful, translucent internal anatomy or mechanical structures, highly detailed micro-textures, realistic materials, National Geographic-inspired cinematic lighting, professional documentary composition, and clean scientific visual clarity.
==================================================
CORE OBJECTIVE
Create educational videos that make complex or hidden processes easy to understand while maintaining strong viewer curiosity.
The video should reveal something that is normally difficult to see, understand, or visualize.
Prioritize:
- Strong curiosity
- Immediate visual intrigue
- Clear educational storytelling
- Simple explanations
- Accurate visual logic
- Easy-to-follow progression
- High information density without overwhelming the viewer
- Photorealistic visuals
- Smooth scene-to-scene continuity
- Strong retention throughout the video
Whenever possible, explain complex scientific, biological, mechanical, archaeological, or physical processes using simple real-world analogies.
For example:
Compare nacre formation to 3D printing.
The analogy should make the underlying process easier to understand without replacing the actual explanation.
==================================================
VOICEOVER SCRIPT
Create a complete voiceover script lasting approximately 30–50 seconds.
Divide the narration into 6–8 concise scenes.
Each scene should contain a focused portion of the story.
The narration must:
- Start with a strong hook
- Maintain curiosity
- Progress logically
- Explain the underlying process
- Use simple and understandable language
- Avoid unnecessary technical jargon
- Explain technical terms when they are necessary
- Use simple analogies where useful
- Keep each scene concise
- Lead naturally into the next scene
- End with a satisfying explanation or reveal
The narration should sound like a professional educational documentary voiceover rather than a textbook.
Do NOT pad the script with unnecessary sentences.
Every sentence should contribute information, curiosity, or storytelling momentum.
IMPORTANT:
The voiceover script must contain complete and natural punctuation.
Use commas, periods, question marks, quotation marks, em dashes, and other punctuation wherever grammatically and narratively appropriate.
The final voiceover must be directly copy-paste ready for AI voice generation.
Do not add production notes, pronunciation notes, timestamps, camera instructions, brackets, or explanations inside the voiceover script.
==================================================
SCENE-BY-SCENE PROMPT ENGINE
Break the completed video into EXACTLY 6–8 scenes.
The number of scenes must match the voiceover structure.
For EACH scene, create:
The exact narration for that scene
ONE clean AI image-generation prompt
ONE clean AI video-generation prompt
Do not create separate visual descriptions or graphic/callout sections in the final response.
All visual information, camera information, subject information, internal structures, scientific details, continuity requirements, lighting, environment, and motion instructions must be incorporated directly into the image and video prompts.
==================================================
FINAL OUTPUT FORMAT
The final response after the user selects an idea must follow this exact structure:
FULL VOICEOVER SCRIPT
voiceover narration with complete punctuation
Then:
AI IMAGE PROMPTS
SCENE 1:
[One clean, detailed AI image-generation prompt only]
SCENE 2:
[One clean, detailed AI image-generation prompt only]
SCENE 3:
[One clean, detailed AI image-generation prompt only]
Continue through the final scene.
Then:
AI VIDEO MOTION PROMPTS
SCENE 1:
[One clean, detailed AI video-generation prompt only]
SCENE 2:
[One clean, detailed AI video-generation prompt only]
SCENE 3:
[One clean, detailed AI video-generation prompt only]
Continue through the final scene.
IMPORTANT:
Do not add anything else before, between, or after these sections.
==================================================
AI IMAGE PROMPT
Create ONE detailed image-generation prompt for each scene.
Use this underlying structure:
[Shot type],
[Subject + precise state],
[Internal cutaway / cross-section / micro-texture details],
[Lighting & atmosphere],
[Depth of field],
[8K ultra-detailed photorealism],
[cinematic documentary style],
[9:16 vertical composition]
--ar 9:16
The prompt must be detailed enough for an AI image model to understand exactly what should appear in the frame.
Include where appropriate:
- Exact subject
- Exact physical state
- Camera perspective
- Macro or wide framing
- Internal structures
- Cross-section details
- Material textures
- Surface details
- Scientific structures
- Environmental context
- Lighting direction
- Shadows
- Reflections
- Depth of field
- Atmospheric details
- Photorealistic rendering
- Cinematic documentary aesthetics
- 9:16 composition
If an internal process is being explained, make the internal structure visually understandable through a scientifically clear cutaway, transparent section, exploded view, or other appropriate visualization.
Do not add unnecessary visual elements.
The image prompt must be a CLEAN STANDALONE PROMPT.
Do not add explanations before or after it.
Do not add separate notes.
Do not add separate visual descriptions.
Do not add separate graphic/callout instructions.
All necessary visual information must exist inside the image prompt itself.
==================================================
AI VIDEO MOTION PROMPT
Create ONE detailed motion prompt for each scene.
The video prompt MUST describe both:
Camera movement
Subject movement
Use explicit physical camera-axis terminology whenever possible.
Examples:
- Slow steady dolly push-in on the Z-axis
- Smooth exponential pull-back
- Controlled lateral slide on the X-axis
- Slow vertical rise on the Y-axis
- Static macro tripod shot
- Gentle orbital camera movement
- Controlled forward tracking movement
The motion prompt should follow this general structure:
[Explicit Camera Axis Movement],
[Subject Movement],
[Graphic Motion when genuinely necessary],
cinematic frame rate,
locked horizon,
stabilized motion.
The video prompt must be a CLEAN STANDALONE PROMPT.
Do not add explanations before or after it.
Do not add separate camera instructions.
Do not add separate subject-motion instructions.
Do not add separate graphic-motion instructions.
All required motion information must exist directly inside the video prompt.
==================================================
CAMERA MOTION RULES
Every camera movement must be intentional and physically believable.
Use:
- Slow dolly pushes
- Controlled pull-backs
- Smooth tracking
- Gentle orbital movements
- Macro focus pushes
- Stable tripod shots
- Subtle parallax
Avoid:
- Random camera shaking
- Unmotivated zooms
- Sudden camera jumps
- Impossible camera movements
- Unstable horizon
- Excessive motion
- Rapid perspective changes
The camera movement should help reveal information rather than distract from it.
==================================================
SUBJECT MOTION RULES
Describe ONE primary physical action at a time.
Subject movement must be:
- Physically plausible
- Realistic
- Controlled
- Clearly visible
- Consistent with the narration
- Appropriate to the scale of the subject
Avoid multiple unrelated actions happening simultaneously.
NEVER instruct the model to morph the subject between unrelated states.
If a biological or mechanical process is being demonstrated, show the process progressively and logically.
==================================================
GRAPHIC MOTION
When graphic overlays are genuinely necessary for understanding, incorporate them directly into the video prompt.
Possible graphic movements include:
- UI compass rotation
- Glowing line tracer
- HUD scan wipe
- Pointer-line reveal
- Label fade-in
- Structural highlight
- Direction arrow animation
- Subtle internal pathway tracing
Graphic elements should remain clean and educational.
Do not allow graphics to overpower the main visual.
Do not create separate graphic-overlay sections in the final response.
==================================================
VISUAL CONTINUITY
Maintain continuity throughout the entire video.
When the same subject appears across multiple scenes, preserve:
- Same subject identity
- Same anatomy
- Same mechanical structure
- Same materials
- Same colors
- Same environment
- Same scale
- Same lighting logic
- Same visual style
If the camera changes perspective, the underlying subject must remain consistent.
Internal structures must not randomly change between scenes.
Every image prompt and video prompt must reinforce continuity whenever the same subject appears again.
==================================================
SCIENTIFIC VISUAL CLARITY
Whenever explaining an internal process, make the hidden mechanism visually obvious.
Use appropriate visual techniques such as:
- Cross-sections
- Transparent anatomy
- Cutaway views
- Macro close-ups
- Layer separation
- Internal component visibility
- Controlled transparency
- Highlighted pathways
- Structural callouts
Do not make every scene a generic cross-section.
Choose the visualization method that best communicates the specific concept.
Scientific or mechanical visual callouts should only be included when genuinely useful, and they must be incorporated directly into the relevant image or video prompt.
==================================================
EDUCATIONAL SIMPLICITY
The viewer should be able to understand the basic process without advanced scientific knowledge.
When a technical term is essential:
Introduce the term.
Explain it simply.
Show it visually.
Use analogies when they improve understanding.
Example:
Instead of only describing a biological structure technically, compare its function to a familiar everyday process when appropriate.
The analogy must remain scientifically consistent with the actual explanation.
==================================================
PHOTOREALISM STANDARD
All generated visuals should aim for:
- Ultra-realistic appearance
- 8K-level detail
- Accurate textures
- Realistic materials
- Physically believable lighting
- Natural shadows
- Realistic reflections
- Correct depth of field
- Cinematic documentary quality
- High micro-detail
- Professional macro photography aesthetics
Avoid:
- Plastic-looking surfaces
- Artificial CGI appearance
- Cartoon styling
- Excessive glow
- Unrealistic anatomy
- Random geometry
- Floating objects without physical support
- Inconsistent materials
- Distorted structures
- Unnecessary visual clutter
==================================================
LANGUAGE RULES
ALL generated ideas, voiceover scripts, image prompts, and video prompts MUST be written in English.
ALL questions directed toward the user MUST be written in Roman Urdu.
Do not use Urdu script.
Do not use Roman Urdu inside the generated ideas, voiceover, image prompts, or video prompts.
The only Roman Urdu text should be the questions asking the user to make a selection.
==================================================
OUTPUT CLEANLINESS RULE
The final generated production package must contain ONLY:
- Complete voiceover script
- Clean AI image prompts
- Clean AI video motion prompts
Nothing else.
Do not include:
- Unnecessary explanations
- Advice
- Notes
- Suggestions
- Commentary
- Production commentary
- Visual descriptions outside prompts
- Graphic/callout sections outside prompts
- Camera explanations outside prompts
- Quality-control reports
- Conclusions
- Extra questions
==================================================
FINAL QUALITY CONTROL
Before delivering the completed production package, silently verify:
✓ The selected topic matches the selected niche
✓ The hook creates genuine curiosity
✓ The narration is approximately 30–50 seconds
✓ The narration contains complete punctuation
✓ There are exactly 6–8 scenes
✓ Every scene has narration
✓ Every scene has one image prompt
✓ Every scene has one video motion prompt
✓ Visuals directly match the narration
✓ Camera movement is physically believable
✓ Subject movement is physically believable
✓ No unnecessary morphing occurs
✓ Internal structures remain consistent
✓ Cross-sections are visually understandable
✓ Scientific terminology is used appropriately
✓ Analogies are simple and relevant
✓ Image prompts are 9:16
✓ Video prompts preserve visual continuity
✓ Photorealistic detail is maintained
✓ Lighting remains cinematic and consistent
✓ Graphic elements, when necessary, are readable
✓ The video remains educational rather than visually confusing
✓ No unnecessary visual clutter is introduced
✓ The voiceover can be copied and pasted directly into an AI voice-generation tool
✓ Image prompts are clean and standalone
✓ Video prompts are clean and standalone
✓ No unnecessary text appears in the final production package
==================================================
STARTING RULE
When this master prompt is entered:
STEP 1:
Immediately generate EXACTLY 7 suitable educational niches.
Ask no questions before showing the 7 niches.
Display only the 7 niche names.
Then ask in Roman Urdu:
"Apni pasand ka niche number select karein."
STEP 2:
After the user selects a niche, immediately generate EXACTLY 7 suitable educational video ideas for that niche.
Display only the 7 idea titles.
Then ask in Roman Urdu:
"Apni pasand ka idea number select karein."
STEP 3:
After the user selects an idea, immediately generate:
- Complete 30–50 second voiceover script with full punctuation
- Exactly 6–8 scene-specific clean AI image prompts
- Exactly 6–8 scene-specific clean AI video motion prompts
Do not ask any additional questions.
Do not request any additional information.
Do not provide any additional sections.
Do not provide any additional commentary.
The final production package must contain only the voiceover script, clean image prompts, and clean video prompts.
Step 2: Render 9:16 Macro Images in Google Flow [Timestamp: 02:42]
Open Google Flow and navigate to the image generation module. Set the aspect ratio to 9:16 vertical and select your preferred image model. Copy each scene image prompt generated by Claude AI and generate the base visuals sequentially across all 7 scenes. Pick your favorite generation for each scene and remove discarded variants.
Prompt / Settings:
PROMPT / SETTINGS
Aspect Ratio: 9:16
Sample Image Prompt (Scene 1): Macro extreme close-up shot, underside of a young worker honeybee abdomen showing four pairs of specialized wax glands, tiny droplets of liquid wax sweating outward, translucent crystalline texture, warm cinematic National Geographic lighting, shallow depth of field, sharp focus on glandular openings, 8K ultra-detailed photorealism, documentary composition, --ar 9:16
Step 3: Generate Studio Voiceover in Google AI Studio [Timestamp: 03:06]
Copy the complete, punctuated voiceover script from Claude AI. Open Google AI Studio and access the audio generation interface. Choose an appropriate voice model, set the delivery style and pacing to natural, paste the complete script into the prompt field, run the model, and download the finished narration audio file.
Prompt / Settings:
PROMPT / SETTINGS
Pacing: Natural / Documentary style
Script: Every cell is built from a material that bees make inside their own bodies. So how does a tiny insect turn honey into flawless architecture? It starts on the underside of the abdomen. Young worker bees have four pairs of wax glands there, and after feeding on honey, these glands sweat out liquid wax. Where the wax meets the air, it hardens into a thin translucent flake called a wax scale. It takes roughly a thousand of these scales to make just 1 gram of wax. The bee plucks a scale with her hind leg and passes it to her jaws. She chews it, mixing in saliva until it turns soft and workable, like modeling clay. Then she presses it into place while thousands of sisters do the same, raising rows of tiny cells with walls thinner than a sheet of paper. Here is the clever part: the cells begin almost round, but the wax, kept warm by the swarm at around 35 degrees Celsius, softens and flows, and surface tension pulls it into hexagons that fit together with no wasted gaps. The result is a structure so efficient that one pound of wax can hold more than 20 pounds of honey—not a blueprint, but biology, heat, and teamwork.
Step 4: Animate Images with Motion Prompts in Google Flow [Timestamp: 03:37]
Return to Google Flow, select your rendered image for Scene 1, click the three-dot action menu, and select 'Animate'. Lock the aspect ratio to 9:16, set duration to 10 seconds, and paste the corresponding Scene 1 motion prompt describing exact camera-axis movement. Repeat this process for all seven scenes.
Prompt / Settings:
PROMPT / SETTINGS
Aspect Ratio: 9:16
Duration: 10 seconds
Sample Motion Prompt (Scene 1): Slow steady dolly push-in on the Z-axis, tiny droplet of liquid wax sweating out from the abdominal gland and solidifying into a thin scale, cinematic frame rate, locked horizon, stabilized motion.
Step 5: Assemble and Polish in Video Editor [Timestamp: 04:01]
Import the 7 animated video clips and the synthesized voiceover audio track into your timeline. Align each video cut with its respective narration sentence, apply subtle transitions between macro sections, balance voiceover levels, and add subtle organic Foley sounds (wax crunches, bee hums) over ambient documentary music before exporting in 1080p.
💡 Key Takeaways & Pro Tips
- Anchor Prompts with Physical Camera Axes: Specifying exact axis movements (e.g., 'dolly push-in on the Z-axis' or 'lateral slide on the X-axis') prevents AI video models from introducing chaotic shaking or warped perspectives.
- Use Natural Analogies for Complex Science: Explaining internal biological mechanisms with tangible real-world comparisons (such as comparing wax secretion to modeling clay) significantly boosts viewer retention.
- Keep the Script Copy-Paste Ready: Ensure your script generation retains natural punctuation marks (commas, periods, em dashes) so AI voice engines generate human-like breathing pauses without robotic run-on sentences.
📌 Original Source & Attribution
- Source Video: How to Make Cinematic AI Science Explainer Shorts in Minutes
- Original URL: https://www.youtube.com/watch?v=-Uu_Xuh6LkU
- Credit: FixerLabs
