AI Talking Avatar

minimax/h3-max-turbo/image-to-video
  image_url = <your picture> · duration = 2 · resolution = 768P · prompt_expansion_mode = disabled

minimax/h3-max-turbo/extend-video   (once per piece)
  video_url = <the clip so far> · reference_audio_urls = [<piece>] · duration = piece + 0.2 s
  output = continuation · resolution = 768P · enable_prompt_expansion = false
  → cut the first 1.625 s (a repeat of the source’s end), append, re-upload as the next source

One continuous take from the given first image. The <man/woman/person> stays still and silent, looking straight into the lens, lips closed, one soft natural blink. The camera holds still. Everything stays exactly as in the first image. Quiet room tone only, no voice, no music.

The same woman now speaks straight to the camera. Audio 1 is her voice: she says exactly the words of Audio 1, her lips exactly in sync with it: "Rule number one: never go back for what you left behind. Rule number two... tomorrow." Small natural head movements and blinks while she talks, a faint smile on the last word. The camera holds still. Everything else stays exactly as in the video. No music, no other sounds.

Split speech at pauses (silence −35 dB, ≥ 0.3 s) into even 2–14 s pieces.
Attach each piece through your supported external tool and transcribe its words with Scribe.
Keep an illustrative assembly plan (not an installed runner or API):
{"name": "speaker-clip", "image_url": "<your picture>", "parts": [{"audio_url": "<piece 1>", "seconds": 10.6, "prompt": "<prompt 1>"}]}
Request a 2 s image-to-video stub, then extend once per piece with output continuation.
Trim the repeated tail, append the new footage, and remove the silent stub.

The same <man/woman/person> <starts/keeps> talking straight into the camera, casual and alive, like recording a story for <their> followers. Audio 1 is <their> voice: <they> say(s) exactly these words, <their> lips exactly in sync with Audio 1: "<the piece's words>" <acting beats tied to words>. The framing is slightly crooked and wobbles a little, like a handheld phone. <Their> face, hair and clothes, the place and the light stay exactly as in the video. One continuous shot, no cuts, no music, no text on screen.

The same woman starts talking straight into the camera, casual and alive, like recording a story for her followers. Audio 1 is her voice: she says exactly these words, her lips exactly in sync with Audio 1: "Okay, rule number three: never trust a man who irons his jeans. No, I'm serious. My ex did it every single Sunday, and you know what he left me with?" A small breath and a mischievous smile before "Okay". On "never" she leans in closer to the lens and narrows her eyes. She laughs after "jeans", tilting her head back a little. On "No, I'm serious" she lifts one hand, palm out. On "every single Sunday" she taps the beat with a finger. On "you know what he left me with?" she raises her eyebrows and pauses, waiting. The framing is slightly crooked and wobbles a little, like a handheld phone. Her face, hair and clothes, the place and the light stay exactly as in the video. One continuous shot, no cuts, no music, no text on screen.

The same woman keeps talking straight into the camera, casual and alive, like recording a story for her followers. Audio 1 is her voice: she says exactly these words, her lips exactly in sync with Audio 1: "A really nice iron. So rule number four: if something feels wrong at two in the morning, it's wrong at noon too. Trust me on that one. See you tomorrow." She sighs and rolls her eyes before "A really nice iron", deadpan. On "rule number four" she holds up four fingers. On "two in the morning" she drops her voice and leans in close to the lens. She whispers "Trust me on that one" with a small slow nod. A soft laugh and a little wave on "See you tomorrow", then she holds a warm smile. The framing is slightly crooked and wobbles a little, like a handheld phone. Her face, hair and clothes, the place and the light stay exactly as in the video. One continuous shot, no cuts, no music, no text on screen.

Picture 1 is a character sheet of one <man/woman/person>. A vertical 9:16 live-action selfie frame shot with a phone's front camera held at arm's length; the phone itself is never seen. The same <man/woman/person> stands <at the start of the route> of <{place}>, the place visible behind <them>: <every stop of the route, 5–7 concrete things>. A medium close-up from chest up, slightly high angle, <they> look(s) straight into the lens with a <mood> smile, about to speak. <Their> face, hair and clothes exactly as in picture 1. Natural phone camera look, <the light sources of the place>.

A vertical 9:16 selfie <room tour / walk>: the <man/woman/person> films <themselves> with the phone's front camera held at arm's length in one hand; the phone itself is never seen. Audio 1 is <their> voice: <they> say(s) exactly these words, <their> lips exactly in sync with Audio 1: "<the piece's words>" <the stop: where <they> walk(s) and what <they> do(es) on which word>. The camera moves with <them> as <they> walk(s), the framing bobs and sways a little with <their> steps like a real handheld phone, and the place passes behind <them>. <Their> face, hair and clothes stay exactly as in the video, and it is the same <place> as in the video: <the same 5–7 things as in the start frame>. <light>. One continuous shot, no cuts, no music, no text on screen.

Picture 1 is a character sheet of one woman. A vertical 9:16 live-action selfie frame shot with a phone's front camera held at arm's length; the phone itself is never seen. The same woman stands just inside the door of her small bedroom in the evening, the room visible behind her: an unmade bed with fairy lights on the wall above it, a small desk with a warm desk lamp, a big leafy plant in a pot, a clothes rack with jackets, a window with city lights at night. A medium close-up from chest up, slightly high angle, she looks straight into the lens with a small excited smile, about to speak. Her face, hair and clothes exactly as in picture 1. Natural phone camera look, warm practical light from the lamp and the fairy lights.

A vertical 9:16 selfie room tour: the woman films herself with the phone's front camera held at arm's length in one hand; the phone itself is never seen. Audio 1 is her voice: she says exactly these words, her lips exactly in sync with Audio 1: "Okay, you guys kept asking, so welcome to my room. It's tiny, like really tiny, but it's mine. So this is the bed." Bright and excited, she steps from the doorway into the room. On "welcome to my room" she sweeps her free hand out to show the room. She laughs on "tiny", scrunching her nose. On "So this is the bed" she turns so the unmade bed with the fairy lights is right behind her and points back at it with her thumb. The camera moves with her as she walks, the framing bobs and sways a little with her steps like a real handheld phone, and the room passes behind her. Her face, hair and clothes stay exactly as in the video, and it is the same room as in the video: the unmade bed with fairy lights and photos above it, the desk with the warm lamp, the big leafy plant, the window with city lights at night, the clothes rack with jackets. Warm light from the lamp and the fairy lights, soft bokeh. One continuous shot, no cuts, no music, no text on screen.

A vertical 9:16 selfie room tour: the woman films herself with the phone's front camera held at arm's length in one hand; the phone itself is never seen. Audio 1 is her voice: she says exactly these words, her lips exactly in sync with Audio 1: "Yes, it's unmade, and no, I'm not sorry. Fairy lights, non-negotiable. They've been up since like two Christmases ago. Over here is my desk. This lamp, flea market, five bucks." On "Yes, it's unmade" she shrugs, unbothered, and grins on "I'm not sorry". On "Fairy lights" she points up at the lights above the bed and wiggles her fingers. She walks over to the desk, the camera following her face, and on "This lamp" she taps the warm desk lamp, proud. The camera moves with her as she walks, the framing bobs and sways a little with her steps like a real handheld phone, and the room passes behind her. Her face, hair and clothes stay exactly as in the video, and it is the same room as in the video: the unmade bed with fairy lights and photos above it, the desk with the warm lamp, the big leafy plant, the window with city lights at night, the clothes rack with jackets. Warm light from the lamp and the fairy lights, soft bokeh. One continuous shot, no cuts, no music, no text on screen.

A vertical 9:16 selfie room tour: the woman films herself with the phone's front camera held at arm's length in one hand; the phone itself is never seen. Audio 1 is her voice: she says exactly these words, her lips exactly in sync with Audio 1: "Best five bucks I ever spent. And this guy, this is Gerald. Gerald is a plant. He's been dying for three years, and he just refuses to actually do it. Respect." On "Best five bucks" she pats the lamp. She turns to the big leafy plant and holds one of its leaves next to her face on "this is Gerald". She says "Gerald is a plant" deadpan, lowering her voice, raises her eyebrows on "refuses to actually do it", then laughs and gives the plant a tiny salute on "Respect". The camera moves with her as she walks, the framing bobs and sways a little with her steps like a real handheld phone, and the room passes behind her. Her face, hair and clothes stay exactly as in the video, and it is the same room as in the video: the unmade bed with fairy lights and photos above it, the desk with the warm lamp, the big leafy plant, the window with city lights at night, the clothes rack with jackets. Warm light from the lamp and the fairy lights, soft bokeh. One continuous shot, no cuts, no music, no text on screen.

A vertical 9:16 selfie room tour: the woman films herself with the phone's front camera held at arm's length in one hand; the phone itself is never seen. Audio 1 is her voice: she says exactly these words, her lips exactly in sync with Audio 1: "Okay, the rack. Every jacket I own lives here. And yes, they're all basically the same jacket. Don't judge me. And the best part? The window." She walks to the clothes rack. On "Every jacket I own" she runs her hand along the hanging jackets; on "the same jacket" she lifts one sleeve and holds it next to the jacket she is wearing, both the same. She sighs and rolls her eyes on "Don't judge me", then on "The window" she walks toward the window, her eyes lighting up. The camera moves with her as she walks, the framing bobs and sways a little with her steps like a real handheld phone, and the room passes behind her. Her face, hair and clothes stay exactly as in the video, and it is the same room as in the video: the unmade bed with fairy lights and photos above it, the desk with the warm lamp, the big leafy plant, the window with city lights at night, the clothes rack with jackets. Warm light from the lamp and the fairy lights, soft bokeh. One continuous shot, no cuts, no music, no text on screen.

A vertical 9:16 selfie room tour: the woman films herself with the phone's front camera held at arm's length in one hand; the phone itself is never seen. Audio 1 is her voice: she says exactly these words, her lips exactly in sync with Audio 1: "At night, the whole city just glows. This is where I think. So yeah, that's my room. Small, messy, home. See you tomorrow." She stands by the window, the city lights glowing behind her, calmer and softer now. On "the whole city just glows" she glances out of the window and back to the lens. She leans her shoulder on the window frame on "This is where I think". A warm smile on "Small, messy, home", and on "See you tomorrow" a little wave of her free hand, then she holds the smile. The camera moves with her as she walks, the framing bobs and sways a little with her steps like a real handheld phone, and the room passes behind her. Her face, hair and clothes stay exactly as in the video, and it is the same room as in the video: the unmade bed with fairy lights and photos above it, the desk with the warm lamp, the big leafy plant, the window with city lights at night, the clothes rack with jackets. Warm light from the lamp and the fairy lights, soft bokeh. One continuous shot, no cuts, no music, no text on screen.

ffmpeg -ss <stub end − 0.1> -i extended.mp4 -c:v libx264 -crf 16 -pix_fmt yuv420p -c:a aac -b:a 192k -movflags +faststart clip.mp4

Picture 1 is a character sheet of one <man/woman/person>. A horizontal 16:9 live-action film still of the same <man/woman/person> hosting <their> <mood> video podcast in {studio}: <they> sit(s) at a desk behind a black podcast microphone on a boom arm, <the lights and the things behind them>. A medium shot from the waist up, the camera at eye level on a tripod, <they> sit(s) a little left of center, the microphone in front of <them> below <their> chin so <their> mouth is clearly visible, <they> look(s) straight into the lens with <a soft calm smile>, about to speak. <Their> face, hair and clothes exactly as in picture 1. Natural film look, shallow depth of field, no text.

The same <man/woman/person> <starts/keeps> talking straight into the camera, calm and warm, like hosting <their> quiet podcast. Audio 1 is <their> voice: <they> say(s) exactly these words, <their> lips exactly in sync with Audio 1: "<the piece's words>" <acting beats tied to words>. <They> stay(s) seated at the desk behind the microphone. The camera holds still on a tripod. <Their> face, hair and clothes, the studio and the warm lamp light stay exactly as in the video. One continuous shot, no cuts, no music, no text on screen.