HeyGen

Yes, Avatar IV and Avatar 4 are the same HeyGen photo-to-talking-video model. It is HeyGen's photo-driven avatar engine, rolled out in mid-2025, and it turns a single still image into a lip-synced presenter with head movement, blinks, micro-expressions and subtle hand gestures.

You need one portrait image and either an audio track or a text script. With audio, the recording drives the performance, sets the video length and overrides the prompt and voice. With text, you write the words to speak and pick a voice from the library.

Stable gives minimal, subtle motion suited to corporate, training and formal announcements, while expressive gives bigger gestures and facial energy for marketing and social. Expressive amplifies the audio's energy but can't create energy a flat voice lacks.

A sharp, well-lit, front-facing portrait with the mouth and eyes clearly visible works best. Leave headroom and shoulder space so tilts and gestures aren't cropped. Sunglasses, hair over the face, hands near the mouth, harsh side-shadows and extreme profiles cause problems.

No, it is a performance tool, not a general video generator. The subject stays a talking presenter in a mostly static frame, so scenes needing camera movement, body action or physical interaction are better handled elsewhere. Long flat monologues also feel static without cutaways.

It is billed per second of generated video, so cost scales with clip length and resolution. Revisions cost full price, so iterate with short scripts at 360p to 540p, then render the final at 720p or 1080p.