Video summary

BONGKAR TRIK! Cara Bikin AI Influencer 100% GRATIS Pakai Google Flow 🤑

Main summary

Key takeaways

Technology

Summary of the Video (AI influencer creation with Google Flow + free tools)

The video explains that creating AI influencers is no longer difficult/taboo and can be done without subscribing to a specific AI model, using free image-generation credits and Google Flow’s newest features.

Core workflow:

  1. Create a character reference
  2. Generate a consistent character sheet
  3. Build a custom AI character in Google Flow
  4. Generate image/video content (including motion control and outfit/pose control)
  5. Publish for consistent income opportunities (affiliate-style content)

Key Tutorials / Guides Shown

1) Create an AI influencer character (from reference)

  • Find a reference character image first
    • Example method: search Pinterest for “Indonesian influencer” and pick a model you like.
  • Download the reference image.
  • Use Clone:
    • Upload the image
    • Paste a prepared prompt (claimed to be available in the video description)
    • If the result is “cloudy” or refuses:
      • Clarify in the chat prompt that the character is fictional, not a real person
      • Then cloud.Aaii/Clone continues generating similar (but not identical) character outputs
  • Output: multiple generated results → choose the best one.

2) Generate a “character sheet” in Google Flow (free credits)

  • Open Google Flow and click New.
  • Use Image mode with:
    • 16:9 aspect ratio
    • Generate 4 images at once
    • Model: Nano Banana 2
  • Important claim: image generation costs zero credits (totally free).
  • Paste the garcloud.ai prompt into Google Flow.
  • Select the best result and name it (e.g., “model sheet / modal sheet”).

3) Build the full character sheet using Grid Architect

  • In Google Flow left menu: Tools → Grid Architect → Start
  • Settings:
    • Model: Nano Banana 2
    • Ratio: 16:9
    • Grid format: 3Ă—2
    • Global reference: use the sheet image created earlier
    • Use wizard prompt to auto-fill prompts for each shot
  • Shots/perspectives generated:
    • front view
    • additional angles
    • rear view
    • close-up (switch to “close-up” camera description)
    • full body (switch to “full shot”)
  • Download note:
    • Downloading directly may give a messy grid
    • Fix: go to All Media, select the top image for a cleaner output.

4) Create a custom AI character (new Google Flow feature)

  • Go to Characters section.
  • Click Add from project and select the model sheet image, then Add to character.
  • Click Create body:
    • Provide a simple prompt for full body
    • Use the format icon to refine/perfect prompts automatically
  • Rename the character (example: “Kirana”).
  • Voice setup:
    • Choose Select voice with multiple sound choices
    • Try a voice basis (example mentions Leda’s voice)
    • Add customization like “Indonesian Accent”
    • Test sample dialogue (e.g., “Hi, my name is Kirana…”)
    • Save the voice preset.

5) Generate image content while keeping character consistency

  • In content execution, start with image generation:
    • Example prompt: “Kirana relaxing while having coffee at a cafe in Bandung”
  • Tag the character:
    • Add tag → Character → select Kirana
  • Settings:
    • Ratio: 9:16
    • Mode: image
    • Model: Nano Banana 2
  • Consistency claim: the character remains consistent across outputs.

6) Use reference images for pose control (Copy tool approach)

To use the pose from a specific picture:

  • Save the reference pose image.
  • In Google Flow:
    • Add to prompt → upload media → select pose image
    • Include the pose image in the prompt workflow
  • Result claim: improved character consistency + aesthetic poses.

Video Generation Methods Compared (and why results differ)

The video describes two different approaches for video content in Google Flow:

Method A: Image-to-video (generate video from the generated image first)

  • Use:
    • Video mode
    • Frames settings
    • Model: Light version (example uses ~10 credits)
  • Output:
    • Visuals okay, but sound may not match voice settings because sound settings may not apply from the beginning.

Method B: Direct video generation with better sound control

  • Use the same character prompt idea, but switch model:
    • Model: Omniflash
    • Example credits: 12 credits
  • Change settings from frames → ingredients
  • Tag character properly (e.g., Kirana).
  • Result claim:
    • Sound aligns with the voice settings created earlier
    • Output feels better than Method A

Outfit + Affiliate-Style Video Creation (without generating an image first)

  • Save an outfit image and upload to Google Flow.
  • Prompt includes:
    • Tag character (Kirana)
    • Tag outfit in the prompt column
  • Settings:
    • Model: Omniflash
    • Duration: 10 seconds
    • Example credits: 15 credits
  • Claim:
    • Character remains consistent
    • Outfit fits perfectly
    • Suitable for affiliate videos

Motion Control Feature (movement reference video)

Demonstrated with:

  • A movement reference video (max 10 seconds)
  • Audio removed first
  • Outfit images prepared for the AI character

Steps:

  1. Upload outfit + dance/movement reference video
  2. Prompt tags:
    • tag character (Kirana)
    • tag outfit
  3. For movement reference:
    • no tag needed
    • add the video into the prompt using add to prompt

Credits:

  • Motion control feature uses 20 credits

Result claim:

  • Movements may not be 100% identical, but remain in rhythm and pleasing
  • Outfit automatically applies without extra drawing/music

Pricing / Credit Strategy (free daily quota)

Google Flow provides:

  • 50 free credits per day, automatically refreshed daily.

With 50 credits:

  • about two motion control videos/day (per the video’s claim)
  • or up to five videos/day using a Lite version

Additional claims:

  • Image generation is completely free (no credit deductions).
  • If needing more capacity:
    • create multiple Gmail accounts
    • or subscribe to a Pro account for higher limits.

Main Speakers / Sources (as identifiable)

  • Speaker(s): The narrator isn’t clearly named in subtitles, but it directly addresses viewers as “Hello Alarka friends” and “back again”—likely the channel host.
  • Tools / Sources mentioned:
    • Google Flow
    • Clone / cloud.Aaii
    • Pinterest
    • garcloud.ai
    • VO (voice) / Omniash / Omniflash / Nano Banana 2 (models/features referenced)

Original video