An avatar is a synthetic on-screen presenter. Pure B-roll uses demonstrations, screen recordings, graphics, or rights-cleared footage under voiceover. The right choice depends on what the viewer must trust and see.

By the end, you will have a six-scene format decision board and a reusable policy for choosing avatar, B-roll, or a hybrid for each content type on your channel.

// concept

Decide With Five Production Questions

Score the video, then decide each scene.

Decision factor	Prefer an avatar when…	Prefer pure B-roll when…
Trust	An authorized presenter must explain or take responsibility.	The viewer must see the product, screen, place, document, or process.
Visual proof	Direct explanation is the main value.	The claim should be inspectable.
Privacy	The presenter consented and required disclosure is ready.	Use rights-cleared objects, hands, diagrams, or captures without a face.
Pacing	Short statements benefit from eye contact and stable framing.	Steps, comparisons, or story beats carry the rhythm.
Budget	Repeated script or language updates justify testing an avatar workflow. Check current HeyGen pricing.	You own usable footage or can record it on a phone, avoiding generation fees.

An avatar cannot manufacture authority or replace an accountable expert. “Faceless” is not rights-free: people, brands, private screens, and licensed footage still need permission.

// concept

Turn the Choice Into a Channel Policy

Set a default per content type; record the reason for exceptions.

Content type	Default format	Why
Product explainer	Avatar-led hybrid	Presenter frames the problem; cutaways prove product behavior.
Software tutorial	B-roll with voiceover	Screen recording is the evidence; an avatar would cover useful pixels.
Story or mini-documentary	B-roll with voiceover	Objects, places, and sequences carry the narrative.
Privacy-sensitive case study	B-roll, diagrams, or text	Avoids exposing a client or creating an unauthorized likeness.

The reusable hybrid is avatar intro → B-roll body → optional avatar close. Identify the speaker, move to evidence when narration names an object, interface, place, or action, and return only for a direct summary.

In HeyGen AI Studio, select an authorized avatar, add the script, and generate only presenter clips; confirm the current menu path. A custom Digital Twin requires the depicted person’s consent. Free alternatives are a phone-recorded intro and your own shots, captures, or diagrams.

// concept

Build a Six-Scene Format Decision Board

Use one row per scene. “Proof required” stops a talking head from hiding the evidence.

// prompt — copy me10 lines

Using only my six-scene script, return a Markdown table with: scene, viewer
question, proof required, AVATAR/B-ROLL/HYBRID, visual, rights check, and reason.

Rules:
- AVATAR requires useful direct address; B-ROLL requires visible evidence.
- HYBRID requires both. Never invent footage, consent, behavior, or results.
- Flag missing consent or source assets as BLOCKED.

Channel policy: [paste your policy]
Six-scene script: [paste six numbered scene summaries]

For generated B-roll, make the evidence visible. Sample Veo prompt for a fictional product:

// prompt — copy me3 lines

Close-up demonstration, unbranded steel bottle on a plain desk; a hand opens
the lid and tilts it over a dry white cloth; continuous side-angle shot, slow
movement, soft window light, documentary style, no text or identifiable person.

Check Veo’s current official access path before generating; availability can change.

// worked_example

Worked Example

This hypothetical sample is not a real campaign or result. A Lahore online seller is planning a 45-second explainer for a sample order tracker.

Scene	Viewer question	Proof required	Format	Planned visual	Reason
1	Who guides me?	Authorized speaker	Avatar	Consented presenter introduces problem	Direct trust
2	What does it do?	Product overview	Avatar + B-roll	Presenter beside a sample interface card	Bridge to proof
3	How is an order added?	Screen action	B-roll	Redacted sample recording	Interface is evidence
4	What changes?	Before/after state	B-roll	Sample status changes	Inspectable proof
5	What stays private?	Data warning	B-roll	Blurred phone/order fields	Protect customer data
6	What next?	Clear action	Avatar	Try one sample order	Direct close

Draft one wrongly kept the avatar full-screen in scenes 3–5, hiding steps and captions. The fix: export scenes 1 and 6 from HeyGen, use screen evidence for scenes 3–5, and transition in scene 2. In CapCut, place voice first, add presenter clips, put B-roll above them, then split and trim cutaways to each spoken action.

A second hypothetical follows a handmade item from sketch to packing. Pure B-roll of materials, consented hands, quality check, and packing supplies proof. An avatar would interrupt the observational pace.

// failure_cases

Failure Cases to Diagnose

6 cases to diagnose

Presenter used as proof
the avatar says a product works, but no demonstration appears. Add a capture or test shot.
B-roll contradicts narration
the spoken and visible steps differ. Re-cut or rewrite the line.
Avatar blocks the interface
use full-screen capture or move the avatar to the intro.
No consent trail
permitted likeness uses are undocumented. Block the scene until consent is recorded.
Stock footage implies a false claim
label an illustrative office or use a graphic.
Restless format switching
change only when the viewer’s question or proof changes.

// pakistan_angle

Pakistan Angle

During load-shedding or unstable mobile data, generate only the short avatar clips, edit with low-resolution proxies, keep originals offline, and queue final uploads for a stable connection. Check the export on an ordinary Android phone. Test Roman-Urdu names and technical terms aloud; awkward pronunciation calls for your own voiceover.

USD subscriptions stack in PKR, and Pakistani cards may not work on every service. Check official price and payment pages before promising a tool. Pay for the bottleneck: HeyGen for repeat presenter updates, or phone B-roll plus CapCut when proof matters. Never upload a client’s CNIC, phone number, WhatsApp chat, address, or face without written permission and redaction.

// hands_on

Hands-On Exercise

6 steps

Choose one six-scene video you are authorized to plan.
Write the viewer question and required proof for every scene.
Apply the five-factor matrix and choose avatar, B-roll, or hybrid per row.
Add the exact planned visual and a privacy/rights check to each row.
Run the prompt; reject choices lacking proof or permission.
Save the final board plus a one-sentence channel policy. Done means all six rows have a justified format, producible visual, and rights status.

// completion_rubric

Completion Rubric

6 checks — tick as you verify

0/6

// sources

Sources

5 official sources — check every claim yourself

// check_yourself

Check yourself

4 questions · answers and options are taken word-for-word from this course

0/4

1 / 4 · diagnose
Your work shows this failure mode: “Presenter used as proof.” The lesson describes it like this: “The avatar says a product works, but no demonstration appears.” What does the lesson tell you to do about it?