Video summary

Create UNLIMITED Stickman Animations with AI | 100% FREE Forever

Main summary

Key takeaways

Technology

Overview

The video is a step-by-step “course” tutorial for creating stickman / Stickman-style animation videos using fully free AI tools. It also provides a process to test multiple AI models to find the best performers for multi-shot animation generation.

Additionally, it includes workflow tips for:

  • Voice quality
  • YouTube demonetization safety

The tutorial covers a full pipeline from channel setup → assets → animation → editing → thumbnail → branding.

1) Why this approach / intended outcome

  • Auto-generated-style tutorials are positioned as a response to viewers who want a more in-depth, practical guide.
  • The presenter claims they will test several AI models against each other to identify the best results for stickman animation creation.

2) “Brain of your operation”: chatbot + prompt system

  • A chatbot is used as a planning and generation hub.
  • For a “completely free” route, the presenter recommends DeepSeek, claiming no paywall and unlimited usage.
  • A Google Doc prompt sheet is provided:
    • Copy the prompt section and paste into DeepSeek.
    • DeepSeek then asks for a “source PDF.”

Stickman training context (source PDF)

  • A ~53-page “Stickman Source Material PDF” is used from Google Drive.
  • Workflow:
    • Download it
    • Attach it to DeepSeek
    • Prompt the model to absorb it

DeepSeek staged output

  • Stage 1: branding
    • Channel name ideas
    • Description options
    • Logo/banner prompt suggestions
  • Script generation
    • Includes an option to retry or change tone/style

3) Script control and tone variation

After generating a script, the presenter notes you can:

  • Retry script generation
  • Change tone (e.g., finance-like or fitness-like) to match the channel’s style

4) Voiceover generation (and quality trade-offs)

The tutorial compares multiple text-to-speech options:

  • Noise AI
    • Free credits, but lower quality
  • Microsoft Clipchamp
    • Claims “unlimited” voiceover creation
    • Multiple voices
    • TTS included in the editing workflow
  • Google AI Studio
    • Access to Google TTS models (e.g., Gemini TTS / Lyria)
  • ElevenLabs (recommended for premium natural quality)
    • Claims about monthly credit limits (enough for several long videos or many shorts)
    • For V3:
      • Switch V2 → V3 for “emotional controls,” natural pauses, and audio tags
      • Uses emotional toggles via bracket commands to control delivery
    • Practical guidance:
      • Paste only small script chunks to avoid robotic/glitchy output
      • Generate and re-roll until voice sounds solid
      • Download the resulting audio

5) Character creation via AI image generation (multiple models)

DeepSeek generates character prompts (text-to-image prompts) based on the script.

The presenter tests multiple “unlimited” image generation options:

  • Meta AI (also mentions video generation access)
  • Google Flow (claims unlimited credits; mentions “Nano Banana 2”)
  • Google Gemini (claims unlimited image generation)
  • TikTok Symphony (claims unlimited image + video model access)

Character style/model testing

  • Prompts are run through each model.
  • Aspect ratio note:
    • One model outputs only vertical 9:16 (fine for character images).
  • The best-looking character images are downloaded.
  • DeepSeek returns multiple character variants, each with different personality/vibe.
  • The user selects one as the brand face.

6) Visual direction vs letting AI “take over”

DeepSeek asks whether to:

  • Build visuals from scratch, or
  • Use an existing “visual flow.”

The tutorial emphasizes:

  • Models may struggle with creative direction for visuals.
  • A visual flow is the scene order (e.g., wakes up → restroom → kitchen → drive to work).
  • You either provide the flow or let DeepSeek generate a breakdown.

Multi-shot style generations

  • Animation is split into chapters (example: ~4 structured chapters).

7) Multi-shot video generation: testing multiple video models

The core comparison is focused on multi-shot animation quality. The presenter runs the same chapter prompts across multiple tools/models.

Tools tried

  • Google Vids
    • New feature described as supporting animation
    • Uses aspect ratio selection and Google VO3
    • Workflow: upload stickman image → “animate an image” → paste chapter prompt
  • Google Flow
    • Video generation with VO3
    • Mentions ~200 free credits/day for video testing
    • Upload avatar → switch image → video → choose outputs/aspect ratio
  • Gemini video generation
    • Uses a model referred to as “flashlight” (video mode)
    • Reports issues/failures; retries later
  • Kwai / Kwen Studios
    • Generates video from character + prompt
    • Sometimes produces a single long unbroken sequence or unusable results
  • Higgs Field (paid platform used for testing)
    • Uses a model referred to as “Seedens 2.0”
    • Sets aspect ratio to 720p to reduce credit usage

Results / findings (quality assessment criteria)

Quality is evaluated by:

  • Multi-shot structure handling
  • Smoothness
  • Aspect ratio correctness
  • Polish / colors

Overall ranking (as described)

  • Google Vids: best in a first chapter test (polished, good colors)
  • Google Flow: second best (strong multi-shot handling, decent animation)
  • Higgs Field: sometimes more polished, but not always best; later described as better than most except Google Flow/Vids
  • Kwen Studios: often unusable or not properly multi-shot
  • Gemini: aspect ratio and smoothness issues; sometimes fails

Iteration strategy

For each chapter:

  1. Run all models
  2. Download the best output
  3. Repeat across chapters until all chapter segments become complete video portions

8) Background music + CapCut editing workflow

  • Background music source: YouTube Studio Audio Library
    • Choose tracks by genre/mood/duration

Editing tool: CapCut

Chosen for beginner-friendly UI.

Workflow:

  • Create a new project
  • Import a folder containing:
    • Voiceover audio
    • Video clips
    • Background track
  • Add background track to the timeline
    • Set volume to about -17 dB
    • Add fade-in / fade-out
  • Add voiceovers in narration order
    • Raise voice volume to about 6–7 dB
  • Align video clips to voiceover
    • Trim if clip length exceeds voiceover duration
  • Preview and export at 720p
  • Optional note:
    • Upscaling later can improve perceived quality

9) Thumbnail generation via chatbot + unlimited image tool

After assembling the video, the user returns to DeepSeek:

  • DeepSeek provides a thumbnail prompt plus 5 variants.

Thumbnail generation:

  • Uses Google Flow (image generation for free/unlimited credits)
  • Paste each prompt → generate → compare → download best

The tutorial describes the thumbnail prompts as:

  • Highly detailed
  • Consistent with reference channel styles

Example thumbnail styles include variants like:

  • “stop wasting time”
  • “21 hours wasted per week” with clean typography and icons.

10) Channel setup + branding pipeline

From DeepSeek’s initial branding stage:

  • Choose a channel name (example used: “Clarity Animations”)

YouTube setup

  • Create the channel on YouTube and customize in YouTube Studio
  • Copy a DeepSeek description suggestion into YouTube Studio
  • Add emojis

Logo/banner generation

  • Logo:
    • Use DeepSeek’s logo prompt
    • Generate in Google Flow
    • Switch to 1:1 aspect ratio for logos
  • Banner:
    • Use Canva (free plan is sufficient)
    • Select YouTube banner template
    • Ensure safe area text placement across devices (mobile/desktop/TV)
    • Customize text (e.g., “subscribe”), font weight, and colors

Finally:

  • Upload logo and banner to the YouTube channel
  • Publish

11) Monetization / safety tip emphasis

The tutorial highlights steps intended to reduce demonetization risk, mainly by:

  • Ensuring premium, natural voice quality
  • Avoiding glitchy/robotic TTS output
  • Keeping production “safe” (presented as lessons learned during a demonetization wave)

Main speakers / sources

  • Main speaker (channel): the narrator/teacher referred to as That Guy (mentioned in subtitles)
  • Primary AI tool sources mentioned:
    • DeepSeek, ElevenLabs, Google Flow, Google Vids, Gemini, Meta AI, TikTok Symphony, Kwai/Kwen Studios, Higgs Field, Canva, CapCut, YouTube Studio Audio Library
  • Referenced model/creator channels for context/benchmarking:
    • That Guy
    • Productive Peter

Original video