구경만 하던 AI 영상, 이제 직접 만드세요 (프롬프트 88개 무료) | 댓글 안 달아도 됩니다. 모션그래픽 88개 그냥 드립니다

Main ideas / lessons conveyed

• Simply watching AI comparison videos isn’t enough—the real value is learning how to use AI tools to produce motion graphics/videos yourself.
• Different AI models/tools perform differently depending on the task (e.g., Claude vs. GPT, Opus 5.5, Atlas 6, Atlas 6.1), so you may need to switch between them.
• The creator provides a free, no-login/no-signup “menu” (library) of motion graphics prompt dictionaries (about 88 prompts).
• The core goal is practical: help businesses (shops, restaurants, events, etc.) create:
  • YouTube-style intros/outros and title/subtitle motion graphics
  • Text/image animations over video
  • Store materials (menus, announcements, schedules, direction cards)
  • Infographic-style explanations (boxes/arrows concept diagrams)
• Results improve when prompts include:
  • Detailed scene sequences
  • Clear instructions for what the AI should and should not do
• A full workflow is demonstrated: install tools → configure settings → insert content → optionally combine segments → run/render via terminal → save outputs → iterate and adjust prompts.

---

Methodology / instructions (step-by-step, detailed)

A) What you need (prerequisites)

• Choose one AI tool/model:
  • Mentioned options: Claude / Codex
• Install:
  • Node.js
  • Video editing tool: FFMPMG (likely FFmpeg)
  • “Skill” / tool: Hyperframe
• Installation help:
  • Use the page’s “Step-by-step installation” button
  • Follow instructions for Mac and Windows
  • The video also includes terminal usage instructions for both OSes

B) Access the motion graphics prompt library

• No login or signup is required.
• The library includes about 88 motion graphics prompt dictionaries.
• The creator compiled them by collecting what they saw in many threads/posts/videos and what made them think: “I want to try making that too.”

C) Use the library: navigation and filtering

• Filter by:
  • Content type (including a 3D-only option)
  • Industry/category
  • Search terms (e.g., “shopping mall”, “menu”)
• Performance note:
  • If the page lags due to low memory, use a pause button to reduce stuttering.

D) Configure common settings (before generating)

• Select the AI tool being used:
  • For Claude Code, use Hyperframe with a specific configuration (described as using “Hyphen Hyperflame” like a given example).
  • For Codex, set it using the dollar sign so the skill loads.
• Choose aspect ratio:
  • 16:9 (YouTube long-form)
  • 9:16 (YouTube Shorts / Reels)
  • 1:1 (square)
  • Custom sizes may be available
• Set a color theme:
  • You may need to adjust colors to match your brand palette (e.g., navy/yellow presets may require changes).
• Choose the generation format:
  • The system can generate infographic-like explanations (e.g., boxes pop out, arrows draw in sequence).

E) Insert your own content (instead of editing code)

• Use “Insert your own content”:
  • Click to enable insertion behavior
  • Replace example subjects (e.g., “napa cabbage kimchi / kimjang kimchi”)
  • The prompt is filled naturally
• Framed as “no need to mess with code.”

F) Set timing / duration

• Default duration per graphic item is about 4–6 seconds.
• Changing total duration:
  • Example: set 15 seconds → it adds extra prompts and produces a 15-second video.

G) Combine multiple segments: “Add to Order Form”

• Recommended when you need multiple elements compiled together, such as:
  • Subtitles
  • Intros
  • Endings
  • Multi-step guidance (maps/location guides)
  • Store visitor guidance and direction flows
• Conceptual model:
  • Like adding multiple items to a shopping cart (prompts are queued into order forms).
• Timing combination idea:
  • If prompt 1 is 5s and prompt 2 is 10s, together they render as a combined longer output (example totals 15 seconds).
• Browser behavior:
  • Common settings may be remembered by the browser for later reuse (e.g., next day), except where privacy protections prevent persistence.

H) Advanced prompt control: “Position” / detailed prompt options

• Use the “position” option to add detailed requirements.
• Creator analogy:
  • Simple prompt = ordering “one bowl” (basic)
  • Detailed prompt = a full order (removing onions, delivery time, servings, etc.)
• Switching between:
  • Simple = treats the request as a whole
  • Details = exposes parameters like length/size and how items appear/move

I) Scene sequencing importance

• The creator emphasizes that accurately prompting the scene order sequence produces more beautiful, taste-aligned results.
• The official Hyperframe manual supports the idea:
  • More detailed writing → less chance the output is “off.”

J) Rendering workflow (practical execution)

1. Copy the detailed prompt for your chosen tool (e.g., Claude Code).
2. Paste it into the terminal window.
3. Run it.
4. Rendering starts; production time depends on your computer.
5. Typical render time mentioned: about 5–10 minutes.
6. Review output:
   • Confirms requested details were included.
7. If needed:
   • Edit/modify the prompt and regenerate
   • The creator also mentions you can ask it to “fix” if something is wrong

K) File saving

• Outputs are automatically saved in a folder called HF Work.
• You can instruct saving to the desktop instead.

L) Iteration philosophy

• Expect to iterate:
  • Try multiple times
  • Adjust prompts if results aren’t satisfactory
• Generate variations by adjusting tokens/prompts.

M) Creator motivation and free distribution appeal

• The creator shares it free without login/signup and encourages viewers to:
  • try it
  • leave comments if it helped
• Their intent is to use AI/video creation to make business work easier and more expressive—and to see positive effects on others.

N) Reminder about AI model switching (avoid being swayed)

• Don’t be distracted by claims that one model is “finished” or replaced.
• Focus on:
  • what motion graphics/videos you can create
  • how AI helps you execute tasks, including 3D workflows (e.g., Blender/engine tools)

---

Speakers / sources featured

• Primary speaker / creator: The YouTube channel narrator (unnamed in subtitles; referred to repeatedly as “I” and “me.”)
• External instructor (demo example): Kim Ji-baek
• AI/brand/tools mentioned:
  • Atlas 6, Atlas 6.1
  • Opus 5.5
  • Claude / Claude Code
  • GPT
  • Codex
  • Hyperframe
  • FFmpeg (subtitled as “FFMPMG”)
  • Node.js
• Education materials referenced:
  • AI Academy called “Non-Devking”
  • “Non-Developer Thinking Method” (Season 1 / Season 2)
  • “Automation Thinking Method” (Season 3)