4K Skin Realism: Full Breakdown
The exact 3-step pipeline for hyper-real AI humans: a Higgsfield Soul 2 character, a skin-imperfection pass, then 4K Seedance 2.0 animation.
The line between real and AI just disappeared. Here is the exact 3-step pipeline.
Step 1: Character (Higgsfield Soul 2)
Soul 2 is the strongest model for believable human identity. Generate your character here first, then swap the details in the prompt to taste.
Prompt 1: the character still
Candid iPhone selfie of a young woman in her early 20s, straight brown hair with a centre parting, natural groomed brows, brown eyes, soft natural makeup, gold hoop earrings, lying on a bed in a bright minimal bedroom, soft natural window light, relaxed neutral expression, shot on a front-facing phone camera, photoreal, shallow depth of field, true-to-life skin tones, no retouching.
Keep it grounded and unglamorous: "candid", "no retouching", "front-facing phone camera". The more it reads like a real phone photo and less like a render, the better everything downstream works.
Step 2: Skin enhancer (imperfection = realism)
Run the Soul 2 output through the skin enhancer. This is the step most people skip and it is the whole game: you are adding imperfection, not removing it. Pores, fine freckles, peach fuzz, slight unevenness. Perfect skin is the #1 AI tell, texture and flaws are what convince the eye.
Push the enhancement enough to bring detail out, but stop before it looks crunchy or over-sharpened. You want "real human in good light", not "HDR".
Step 3: Animate (4K Seedance 2.0 on Higgsfield)
Upload the enhanced still as the start image, then use these settings.
• Model: Seedance 2.0
• Resolution: 4K (mode: std, required for 4K)
• Aspect ratio: 9:16
• Duration: 10s
• Bitrate: high (keeps the texture from getting crushed in compression)
• Audio: on
Prompt 2: the silent-macro animation
Handheld vertical iPhone front-camera selfie video, photoreal, shot like a real casual clip, not staged. A young woman with straight brown hair lies on a bed in soft natural window light, chin on her hand. She does NOT speak for most of the video. She silently raises the phone very close to her face and slowly moves it across different areas of her skin in extreme macro detail: super close on the pores and fine freckles across her nose and cheeks, then the soft peach fuzz catching the light, then the delicate texture and faint vertical lines of her lips, then the individual lashes and the brown of her iris. Really, really close, skin filling the frame, every pore, every tiny detail visible. Only soft natural breathing and quiet ambient room tone, no talking, no dialogue during these close-ups. Then, in the final moment, she lowers the phone back to normal selfie distance, settles her chin back on her hand, looks straight down the lens with a small knowing smirk and says one short casual line: 'Four K C-Dance on higgs field, babe.' Then a quick playful wink. Natural micro-movements throughout: slow blinks, a stray hair shifting, tiny handheld camera wobble. Soft diffused daylight, warm natural skin tones, shallow depth of field, organic handheld motion. Realistic, understated, not performative.