Video Decode

Generative AI for Video: Runway ML + After Effects VFX Compositing Mastery

Lines To Designs Grade F· AI Ad Workflows For Freelancers

Here is exactly what makes this video win, decoded into reusable templates you can apply to your own niche: the title formula, the thumbnail recipe, the hook, the script structure, and the description pattern.

Title

Tool Stack + Outcome Claim

Reusable template

[Outcome] with [Tool 1] + [Tool 2]: [Specific Skill/Result]

77 chars · no number · trigger: none

Title verbatim

"Generative AI for Video: Runway ML + After Effects VFX Compositing Mastery"

What to fix

Title exceeds 70 characters (77 total) — YouTube will truncate on mobile and browse surfaces. Front-load the most compelling hook before the colon.
No emotional trigger or curiosity gap — the title is purely descriptive. Consider: 'How I Create Hollywood VFX with Runway ML + After Effects' or 'Runway ML + After Effects: The VFX Shortcut' to add intrigue or benefit.
Missing a number or specific outcome — 'Mastery' is abstract. Stronger: 'Runway ML + After Effects: 5 VFX Techniques That Look Hollywood-Grade' or 'How to Composite AI Video in After Effects (No Experience Needed).'

Thumbnail

Reusable template

[Shocked face, center-right, dominant size] + [Surreal AI composite background: [LANDSCAPE/EFFECT TYPE]] + [Tool logos, top-left] + [2-line text: '[TOOL 1]' + '[TOOL 2]', white bold, centered-left] + [Complementary color: magenta/orange + cool blue]

subject center-right · emotion: shock · face large · complementary · palette: Dominant magenta/purple background with hot orange/yellow accents (fire, desert warmth). Cool blue tones in mountains and sky. High saturation throughout. Complementary contrast (warm subject environment vs. cool sky backdrop). Creates electric, energetic feel.

On-thumbnail text

"RUNWAY ML +AFTER EFFECTS" (4 words)

What to fix

Text could be bolder and more condensed—'RUNWAY ML + AFTER EFFECTS' reads as 4 words but feels spread across two lines; consider stacking as 'RUNWAY ML + AE' or using abbreviations to tighten visual weight
The shocked expression is strong, but the body language (shoulders neutral, not leaning in) could amplify the 'amazed' read with a slight forward tilt or hand gesture

Hook

Pattern interrupt

Reusable template

0-2s: [SPECIFIC OUTCOME: number + result, e.g., 'I created [visual result] in [time] for [cost]'] -> 2-4s: [CONTRADICTION or OBSTACLE: 'Most people think [false belief]'] -> 4-7s: [TOOL REVEAL: 'Here's the [tool name] trick nobody knows about'] -> 7-30s: [DEMO START: show the first step that proves the claim]

device: none

First 30 seconds

so I'm going to show you how I created this using generative AI and after effects the future is here I had a chance to play around with Runway ML's new video to video gen 3 tool and let me show you how I created this using Runway ML and after effects okay so I'm in Runway ML and you can see it has tons of different tools that you can work with but today I'm just going to be working with the video to video gen 3 Alpha so when I get here I'm going to click on select app assets and I have my clip

What to fix

Lead with the outcome or the surprise, not the process: 'I just created a photorealistic video from a single image using AI—here's how' lands infinitely harder than 'I'm going to show you how I created this.'
Add a concrete number or specific result in the first 3 seconds: 'This took me 12 minutes and cost $2' or 'This video got 50K views in 24 hours' gives the viewer a reason to care before you explain the tool.
Open a loop immediately: hint at a problem or contradiction ('Most people think this requires a $5K camera setup, but I just did it with free AI') so the viewer NEEDS to know how.

Short script

Tutorial collapsed

Reusable template

[0-2s] [VISUAL: final composite result] [HOOK: bold claim about AI tool + speed] [SFX: dramatic reveal]
[2-35s] [VISUAL: rapid montage of tool interface, prompts, generation, masking] [NARRATION: 2-3 key feature callouts only, no step-by-step] [SFX: 5-7 transition hits]
[35-50s] [VISUAL: before/after split screen or loop back to opening result] [NARRATION: punchy payoff line]
[50-60s] [CTA: tool name + link/call-to-action] [MUSIC: upbeat, tech-forward]

Hook

so I'm going to show you how I created this using generative AI and after effects the future is here

What to fix

Hook is buried and weak—opens with 'so I'm going to show you' instead of a polarizing claim or visual shock. First 2 seconds must show the FINAL RESULT or a jaw-dropping visual, not a promise of explanation.
No loop-back or rewatch incentive—ending is flat ('Here's the final result') with no emotional payoff or twist that makes the viewer want to rewatch from the start.
Runtime is 3+ minutes, not 30-60 seconds—this is a tutorial video, not a Short. For Shorts, compress to 45-60 seconds max, front-load the payoff, and cut all step-by-step explanation. Show the magic, not the mechanics.

Long script

Tutorial

Reusable template

[0:00–0:15 | Cold open: final composite result (the payoff) with bold claim, e.g. 'I created this entire video using AI in under 30 minutes']
[0:15–0:45 | Branding + tool intro: 'Here's how I did it using [Tool Name]']
[0:45–1:30 | Step 1: Setup + first feature demo + output reveal + micro-hook]
[1:30–2:15 | Step 2: Second feature + output + 'here's the thing' re-hook]
[2:15–3:00 | Step 3: Third feature + output + 'this saves so much time' pattern interrupt]
[3:00–3:45 | Step 4 (optional): Advanced feature or alternative method + output]
[3:45–4:00 | Final composite reveal + emotional reaction]
[4:00–4:15 | CTA: 'Try it yourself, link in description, subscribe for more']

3 chapters

Cold open, first 30s

so I'm going to show you how I created this using generative AI and after effects the future is here I had a chance to play around with Runway ml's new video to video gen 3 tool and let me show you how I created this using Runway ML and after effects okay so I'm in Runway ML and you can see it has tons of different tools that you can work with but today I'm just going to be working with the video to video gen 3 Alpha so when I get here I'm going to click on select app assets and I have my clip here of a woman walking down a path Barefoot

What to fix

Cold open has no hook—it's a statement of intent ('I'm going to show you') rather than a bold claim or surprising result. Open with the final composite result (the 'wow' moment) first, then reveal how it was made.
Zero mid-roll re-engagement hooks between 1:00 and 2:00—the masking section is technically dense and visually repetitive (clicking, masking, feathering). Plant an open loop here ('but the real trick that makes this look photorealistic comes in step 3') to pull viewers through the slower teaching section.
No strategic CTA or value statement—the video ends abruptly with 'here's the final result' and no call to action, no invitation to try it themselves, no link to Runway ML. Add a 10-second CTA at 3:00 that invites viewers to experiment and subscribe for more AI tool breakdowns.

See the full decode

Unlock the script structure, editing cues, and description template behind this video, then get the same decode and reusable templates for your own niche.

Decode My Niche Free

Free. Build your own creator card + niche breakdown in minutes.