Source photos for a realistic avatar

Read the video beyond pretty frames

Separate identity anchors, geometry support, expression, and body scale.

Primary source

Local video analysis and timestamp selection

Every recommendation in this course comes from the video, contact sheets, and eight extracted frames.

Read and test the interactive blocks.

The big idea

A frame may look appealing and still be weak for identity. Function comes before beauty: an image is an anchor, geometry support, expression, or body scale.

Think of a multi-page identity document. No single page solves everything; together they remove ambiguity.

72,77 s

video inspected

timeline samples

final frames

From video to new photos

The selected frame teaches the pose; the new photograph becomes the final reference.

role	use	avoid
Anchor	neutral geometry	extreme expression
Support	side and volume	use alone
Expression	mouth and cheeks	lock identity

Real evidence

The final timestamps are 00:16.0, 00:17.40, 00:18.5, 00:27.25, 00:57.50, 00:29.5, 00:50.80, and 01:00.0. The video has no reliable facial three-quarter B view; that pose must be reshot.

outputs/avatar-photo-selection/selection-sheet.jpg

ffmpeg -ss 00:16.0 -i IMG_2907.MOV -frames:v 1 frontal.png

Check your judgment

Which image should define base identity?

A neutral frontal view shows geometry without a strong expression deforming it. Smile and body are supporting references.