HeyGen Bootcamp

Introduction to Hagen Bootcamp

Welcome and Overview

• The host greets participants from around the world, expressing excitement for the sessions ahead.

• Participants are encouraged to take a quick break before the first session starts in 10-15 minutes.

• The agenda is introduced, focusing on setting goals for 2026 that attendees hope to achieve through the bootcamp.

Agenda Breakdown

• Six sessions are scheduled for today, with a break in between; tomorrow will feature breakout sessions.

• The bootcamp aims to empower attendees in mastering avatars, voice clones, and AI video production workflows using Hagen's tools.

Hagen's Tools and Goals

Empowering Communication Strategies

• Hagen has developed a powerful suite of AI video tools aimed at scaling authentic messaging across various formats like training videos and corporate communications.

• A team of product managers, engineers, and researchers will guide participants on utilizing these tools effectively for their businesses.

Focus on 2026 Goals

• Attendees are encouraged to think about their communication strategies as they prepare for 2026 during the bootcamp sessions.

Engagement and Community Input

Participant Interaction

• Participants are prompted to share their specific goals for 2026 in the chat, fostering community engagement.

Inspiration from Past Successes

• Examples of impactful goals include increasing course completion rates, expanding market reach through localization, or achieving sustainable growth via effective video content.

Session Previews

Upcoming Sessions Overview

• The first session features Adam Helper discussing how to create professional avatars.

• Subsequent topics include perfecting AI voice authenticity and streamlining creative workflows using existing content within Hagen’s framework.

Technical Insights

• A focus on editing techniques within the AI studio will be covered later in the day, emphasizing tips for enhancing engagement through B-roll and motion graphics.

Session Overview and Structure

Introduction to the Day's Sessions

• The session promises valuable insights, focusing on scaling messages globally using Hagen's translation features.

• Attendees are encouraged to save their calendars for breakout sessions scheduled for the following day.

Q&A and Community Engagement

• Instructions provided on how Q&A and chat functionalities will operate during the sessions; community sharing is emphasized as a learning tool.

• Participants can return to the same link throughout the day, with all sessions being recorded for later viewing.

Q&A Management and Feedback Collection

Engaging with Questions

• Attendees are directed to submit questions via the Q&A tab at community.hgn.com, ensuring that queries can be addressed effectively despite high attendance.

Importance of Feedback

• A survey will be distributed after each session, encouraging immediate feedback while content is fresh in participants' minds.

Technical Setup and Session Start

Addressing Technical Issues

• Viewers experiencing audio issues are advised to refresh their stream or adjust audio settings for optimal experience.

Introduction of Speakers

• Adam and Nick, experts in avatar technology, are introduced as speakers who will guide attendees in creating effective digital avatars.

Creating Effective Avatars with Hagen

Overview of Avatar Creation Process

• Adam explains that Hagen allows users to create videos featuring avatars that mimic their appearance without needing constant filming.

Key Principles of Avatar Development

• High-quality training footage is essential for creating effective avatars; quality directly impacts avatar performance compared to public examples.

Best Practices for Filming Avatars

Filming Guidelines

• The session will cover best practices for setup and filming avatars tailored to specific use cases.

Safety Measures in Avatar Creation

• Emphasis on user control over identity; consent is required before creating digital twins, ensuring ethical standards are upheld.

Creating Effective Avatars for Client Communications

Filming Environment Considerations

• When creating an avatar for client communications, consider filming in a relevant environment. If not possible, alternatives like solid color backgrounds can be used.

• A solid wall or green screen is recommended for filming; ensure strong contrast between the subject and background to facilitate effective background removal later.

Technical Quality Factors

• High-quality background removal allows for swapping backgrounds or overlaying avatars on presentations. Good lighting is crucial to avoid shadows and overexposure.

• HN offers a live webcam experience for quick avatar creation, but using high-quality cameras or smartphones yields better results. Aim for footage of at least 30 seconds to two minutes depending on the use case.

Framing and Stability

• Proper framing is essential; your face should occupy over 50% of the frame to maintain pixel data quality during video creation.

• Keep camera movement consistent with your own movements to avoid artifacts in the final video output.

Audio Quality Tips

• Record audio in quiet environments where your voice is clear. Subtle ambient sounds can enhance realism if they do not overpower your voice.

• Ensure even soft lighting across the face to prevent overexposure or shadowing that could affect video quality.

Examples and Best Practices in Avatar Creation

Importance of Real Examples

• Nick emphasizes showcasing real examples from his own likeness to illustrate common patterns observed at Hager, enhancing understanding beyond theory.

Creativity and Feedback Encouragement

• The session encourages creativity; personal style and input are valued as they contribute to product improvement at Hager.

Building Better Workflows

• Understanding user feedback helps improve workflows and models, ensuring users can create their ideal digital twins effectively.

Creating a Digital Twin: Key Insights

Understanding the Importance of Looks

• The speaker emphasizes two critical aspects of creating a digital twin: understanding the use case and recognizing that there are no limitations or charges for creating additional looks.

• Different types of looks can be created, showcasing various outfits and backgrounds while maintaining the same person as the subject.

Expression and Use Cases

• The ability to upload different expressions is highlighted, allowing users to convey various emotions such as neutrality or excitement in their digital twins.

• Various body shots (upper, half, full body) can be utilized to create diverse looks, with features like zooming in and out enhancing visual representation.

Workflow Efficiency

• Users can upload multiple recordings from the same setup without needing to re-upload content, streamlining workflow significantly.

• Each script can be matched with different digital twins based on hand gestures and expressions, facilitating personalized content creation.

Input vs. Output Video Quality

• The importance of quality input videos is stressed; specific gestures must align with desired output videos for effective results.

• Challenges arise when input recordings do not match intended expressions or gestures; users may need to re-record for better alignment.

Addressing Monotony in Avatars

• Customer feedback indicates that many avatars appear too monotone; collaboration is encouraged to enhance expressiveness in digital twins.

• Acknowledgment of challenges faced by models when processing rapid hand gestures or overly expressive performances highlights areas for improvement.

Understanding Avatar Recording Techniques

Importance of Initial Setup

• The speaker emphasizes the need to consider the type of avatar used for a specific duration, suggesting that initial recordings may not represent the best experience.

• Off-camera interactions can disrupt recording quality; looking away from the camera during these moments can be distracting in digital twin presentations.

Preparing for Quality Recordings

• Natural occurrences during recording (e.g., distractions or physical discomfort) should be treated as opportunities for practice rather than failures.

• A critical question before uploading content is whether the input video meets quality standards, including lighting and clarity.

Mastering Lighting Techniques

• Prior to uploading videos, it's essential to check various aspects like lighting conditions and visual artifacts to ensure a polished final product.

• The speaker suggests offering separate sessions on lighting due to its significant impact on video quality, as previously mentioned by Adam.

Equipment and Setup Considerations

• Different setups are possible; even basic equipment can yield satisfactory results if understood correctly regarding lighting.

• Cost-effective equipment options exist that can enhance avatar creation and improve overall digital twin performance.

Effects of Lighting on Performance

• Variations in lighting significantly affect input video quality; mismatched lighting can lead to poor avatar performance.

• While natural light is preferred by many users, it poses challenges due to changing weather conditions affecting consistency in recordings.

Framing and Distance Guidelines

• Proper framing is crucial; common issues include cutting off heads or improper distances leading to unclear visuals.

• Three typical framing examples illustrate common mistakes: too close-up shots reduce lip sync accuracy, while distant shots blur details.

Camera Positioning Tips

• Maintaining an appropriate distance (ideally 1–2 meters from the camera) enhances clarity and accuracy in recordings.

• Adjustments such as elevating laptops with books can improve camera angles for better sight lines during recordings.

Video Presentation Tips

Importance of First Frame in Video Recording

• The model struggles when the first frame lacks a person; efforts are underway to assist users in generating videos even without a subject in the initial frame.

• Avoid uploading footage where the first frame is unclear, as it can confuse viewers about who is present.

Camera Setup and Stability

• A higher laptop position can enhance the first impression; using a tripod for smartphones is recommended for stability during recording.

• While webcams can be useful for certain contexts like webinars, it's essential to consider how you want your avatar to appear.

Eye Contact and Engagement

• Maintaining eye contact with the camera is crucial; aim for recordings of at least 30 seconds, ideally two minutes, to ensure engagement.

• Examples illustrate varying levels of eye contact and positioning that affect viewer perception.

Body Language and Performance

• Understanding your use case helps in presenting yourself effectively as a digital twin; practice talking naturally to convey body language accurately.

• Controlled facial expressions and gestures contribute significantly to professional video performance.

Technical Specifications for Quality Recording

• Basic technical requirements include 4K resolution recording options (landscape or portrait), with 30 FPS recommended for input footage.

Understanding Avatar Creation and User Engagement

The Impact of Frame Rate on User Experience

• There are rumors within the HNN community suggesting that a 60 FPS input does not yield better results for users, indicating a need for further exploration into frame rates and their effects.

Utilizing Avatars for Educational Purposes

• Introduction of "Annie," a public avatar designed for educational use, emphasizing the importance of facial expressions over hand movements in avatar interactions.

Understanding Use Cases in User-Generated Content (UGC)

• Differentiation in UGC styles; understanding specific use cases is crucial when creating avatars for social media or professional settings.

Enhancing Video Quality with Green Screen Techniques

• Two methods to improve video quality using green screens: uploading raw footage directly to HGEN or pre-editing backgrounds before upload. This allows for higher quality outputs.

Recording Strategies for Engaging Content

• Importance of recording multiple angles and short clips to enhance editing possibilities, even if only A-roll footage is used. This approach can lead to more engaging final videos.

Creative Uses of Avatars in Various Contexts

• Encouragement to explore creative applications of avatars through various activities like walking, cycling, or drone shots, enhancing audience engagement through diverse content formats.

Key Takeaways from Avatar Creation

Technical Considerations When Filming

• Introduction of a video score feature that evaluates uploaded footage based on visual and audio quality, providing timestamped feedback on issues detected during filming.

Essential Tips for Effective Avatar Performance

• Three main takeaways:

• Keep eyes at camera level to maintain viewer connection.

• Be extra expressive with voice as AI may mute natural expressiveness.

• Limit excessive hand gestures; focus on subtle movements to avoid distracting the avatar's performance.

Customization Options Post-Creation

• After creating an avatar, users can customize its appearance extensively by changing outfits, poses, and settings using prompts. Additional branding elements can also be integrated seamlessly.

Avatar Customization and Realism

Generating New Looks for Avatars

• Users can create new looks for their avatars through the avatar tab or within the studio, allowing for specific environments like Tokyo or in front of a house's "for sale" sign.

• The customization process is user-friendly, enabling complete appearance changes with prompts. Look packs and style controls are available to enhance this experience.

Understanding Photo vs. Video Looks

• A question arises about the necessity of filming in real environments if avatars can be customized to any outfit and setting.

• Customized photo looks may not achieve the same realism as video looks, particularly for those familiar with the individual; video captures micro-expressions and personal nuances.

• While both looks appear realistic, video representations are more authentic due to their alignment with real motion and expressions.

Tools for Enhancing Video Footage

• For optimal realism, it’s recommended to film using a video look directly; however, photo looks serve as excellent alternatives when filming isn't possible.

• Background removal tools allow users to change their video's background instantly or overlay avatars on presentations.

• AI eye contact correction is available to adjust gaze direction towards the camera but is best achieved by filming correctly from the start.

Importance of Audio Quality

• Avatar models rely on audio input; dull voice quality results in less engaging avatar visuals. A strong expressive voice enhances avatar performance significantly.

Getting Started with HayGen

Initial Steps for Beginners

• Beginners often inquire about where to start within the HayGen app; guidance is provided on navigating initial steps effectively.

User Interface Walkthrough

• A live screen share demonstrates how users can create a new avatar by selecting options within HayGen's interface.

Creating an Avatar

• Users should begin by tapping "create new avatar," which leads them through a guided flow including uploading footage or recording live via webcam.

Finalizing Avatar Creation

• After following instructions, users can generate their avatars easily by tapping "Create" in Studio, marking an accessible entry into video creation.

Conclusion and Next Steps

Transitioning Between Sessions

• The session wraps up with thanks to participants while introducing John for an upcoming discussion focused on perfecting AI voice technology.

How to Perfect Your AI Voice

Introduction to AI Voice Creation

• The session will cover advanced topics including editing in the AI studio, translation, and personalization.

• Adam introduces himself as a product manager for Avatar and Voice, alongside John, the software engineering manager.

Importance of a Great AI Voice

• A high-quality voice is crucial as it drives the visuals of avatar models; monotone audio results in dull avatars.

• The session aims to teach how to create a high-quality voice clone and utilize tools for emotional delivery.

Methods for Cloning Your Voice

• Two methods exist for cloning: automatic voice cloning from uploaded video footage or using a standalone voice clone if quality issues arise.

• Automatic cloning captures ambient sounds that enhance realism but may not always meet quality expectations.

Recording Techniques for High Quality

• Ideal equipment includes a high-quality microphone or smartphone; iPhone users should enable lossless audio in the Voice Memos app.

• When recording, maintain steady speech with brief pauses while ensuring the script matches intended use cases (e.g., ads vs. lectures).

Script Selection and Delivery Tips

• Choose scripts that reflect your avatar's purpose; include accent-specific words to improve authenticity.

• Be expressive during delivery to counteract potential monotony introduced by the cloning process.

Tools Available in AI Studio

• The "voice director" tool allows setting overall mood through presets like excited or calm tones.

• Inline script control can be achieved with voice tags for specific actions like laughter or emphasis on certain words.

Advanced Control Features

• "Voice mirroring" enables users to record their own delivery style which avatars can replicate accurately.

• Most users find success simply entering scripts without additional adjustments, but these tools provide options for fine-tuning.

Voice Control Features in AI Studio

Introduction to Voice Director

• The session begins with a demonstration of the VoiceDirector feature, showcasing its capabilities for controlling voice delivery at the scene level.

• Users can select presets or create custom prompts to adjust the tone of voice, such as making it more excited during previews.

Enhancements with 11Labs V3 Engine

• The integration of 11Labs V3 Engine enhances voice quality and expressiveness, even without additional tags.

• Voice tags provide an extra layer of control, allowing users to add specific emotions like laughter or emphasis; however, results can vary significantly.

Previewing and Regenerating Outputs

• Users are encouraged to preview their outputs due to the non-deterministic nature of the engine; results may differ on subsequent attempts.

• A one-click enhance button is available that automatically adds context-based tags, serving as a helpful starting point for further customization.

Utilizing Public Voices and Tags

• For those using public avatars where personal voice matching isn't crucial, a curated list of recommended voices is provided for optimal results.

• Users can update or add their own tags after enhancing the voice based on script context, allowing for precise emotional delivery.

Practical Application and Contextual Relevance

• Tags last until replaced by new ones; this allows for nuanced control over specific moments within a script.

• Effective use of tags depends on contextual relevance—matching emotional tone with script content yields better outcomes (e.g., excitement in positive news).

Voice Mirroring Capabilities

Introduction to Voice Mirroring

• The concept of voice mirroring is introduced, enabling users to record their own voices and have avatars replicate them accurately.

Demonstration of Voice Mirroring Functionality

• A live demo showcases how well the system matches tone and pacing when transcribing recorded audio into avatar speech.

Customization Options in Voice Mirroring

• Users can modify characteristics such as pitch (e.g., switching from female to male tones), demonstrating flexibility in output options.

Conclusion on Tool Usage

• While advanced tools offer enhanced control over outputs, simply entering a script will still yield satisfactory results without additional adjustments.

Voice Doctor: A New Feature for Voice Issues

Introduction to Voice Doctor

• The speaker introduces a new feature called "Voice Doctor," designed to assist users in resolving voice-related issues. It will be available to all users within two weeks.

Common User Complaints About Voice

1. Voice Doesn't Sound Like Me

• Users often report that their voice does not match their natural accent or pitch, leading to dissatisfaction with the output.

2. Inconsistent Voice

• This issue arises when the voice changes accents or speeds up/slows down unexpectedly during a script, disrupting the flow of speech.

3. Incorrect Pronunciation

• Users need precise pronunciation for brand names and acronyms, which can sometimes be mispronounced by the system.

Solutions for Common Complaints

Addressing "Voice Doesn't Sound Like Me"

• The recommended solution is to use the fish engine in AI Studio, known for its high cloning similarity and ability to replicate challenging accents like New Zealand and South African.

• If issues persist, re-recording with a high-quality microphone in a quiet environment is suggested.

Tackling Inconsistent Voice

• Users are encouraged to preview and regenerate their audio frequently without changing scripts, as this helps achieve better results.

• A new button in the More menu allows users to regenerate voices easily without altering the script content.

Fixing Incorrect Pronunciations

• The "brand glossary" feature enables users to phonetically spell out words that are pronounced incorrectly, ensuring consistent pronunciation across all instances.

Advanced Integrations with Hagen Platform

Importing 11Labs Professional Voice Clone

• Users can integrate their 11Labs voice clones into Hagen using an advanced integration feature accessible from the voice panel.

Steps for Integration

• To set up integration, create a new API key with access permissions for various features including text-to-speech and speech-to-text functionalities.

• After creating an API key, navigate back to Hagen's platform where you can paste your API key into the designated area under New Voice settings.

Voice Cloning and Integration with 11Labs

Setting Up Voice Cloning

• Once the setup is complete, all voices on the 11Labs platform become accessible. Users can add voices like Alice and Adam, which will sound identical to their original versions in 11Labs.

Q&A Session Introduction

• Following the demonstration, a Q&A session begins, inviting questions from participants regarding voice cloning and related topics.

Getting Started with Voice Cloning

• A question arises about where beginners should start with voice cloning. The speaker suggests creating an avatar as a primary step for new users.

Methods of Voice Cloning

• There are two main methods to clone a voice:

• Create an avatar directly from video footage.

• Uploading video allows for automatic voice cloning that matches ambient sounds.

Using the Voice Doctor Feature

• The "voice doctor" feature is introduced as a tool for fixing voice issues. Users can check back later if they don't have access yet.

Accessing and Creating Voices

• If users are dissatisfied with their initial voice clone, they can create a standalone voice clone by accessing the voice panel and selecting options to integrate third-party voices or create new ones.

Options for Generating Voices

• Two options are available when creating new voices:

• Generate Voice: Design a character-specific AI-generated voice.

• Instant Voice Clone: Upload an audio file to create a digital twin's voice.

Considerations for Audio Quality

• When uploading audio files, users must decide whether to keep ambient sounds. While subtle background noise can enhance realism, excessive noise may degrade quality.

Choosing Between Different Voices

• Users can mix various types of voices (automatically cloned from footage, imported from 11 Labs, or created standalone). The integration process remains consistent across different sources.

Recommended Order of Operations for Best Results

• The recommended approach starts with using the automatically generated voice from video footage due to its environmental match. If unsatisfactory results occur, users should try recording standalone audio optimized for quality before seeking advanced integrations or using the "voice doctor."

Final Thoughts on User Experience

• Most users achieve excellent results through direct video look clones or standalone options without needing advanced features.

This structured summary provides insights into setting up and utilizing voice cloning technology effectively while addressing common user queries during the session.

Avatar Creation and Voice Consistency

Stock Avatars and Voice Options

• Hejan provides a library of stock avatars along with various voice options, allowing users to mix and match to find the ideal voice for their avatar.

• Users can opt for a voice clone or select from a public library. The voice design option allows for text-prompted creation of voices tailored to characters.

• In scenarios where an avatar needs to speak different languages, users may choose public voices that are specifically designed with the appropriate accent for better authenticity.

Language Support and Recording

• The system is multilingual but performs best when the recorded language matches the script's language, ensuring optimal output quality.

Maintaining Voice Consistency Across Scenes

• To maintain consistency in voice across scenes, users typically do not need to make adjustments; the system handles this automatically.

• If inconsistencies arise, clicking the regenerate button is recommended. This action will reprocess audio while considering context from previous and subsequent scenes.

• The platform manages much of the behind-the-scenes work to ensure consistent audio throughout videos without requiring user intervention.

Conclusion and Resources

• Acknowledgment of unanswered questions; resources such as a recording of the session and a shared voice guide were provided for further reference. Excitement was expressed regarding upcoming features like Voice Doctor.