How to create UGC ads with AI.

https://mcp.higgsfield.ai/mcp

I want to create an unboxing UGC ad for a fictional dog toy called PawPop.

PawPop is a colourful wobble toy: a weighted base, a short flexible stalk and a soft ball on top. It arrives flat-packed in three large pieces, so the owner assembles it in front of their dog, and the finished toy gives the dog something to nudge and chase.

Give me three or four different concepts for the ad first, one line each, and let me pick one before you write anything longer.

Once I choose, write the full script:
- 25 seconds, vertical 9:16, filmed like casual phone footage by the owner
- four beats, with the timing of each
- what happens on screen in each beat
- the exact words the creator says over each beat

Keep the speech natural and conversational, with dry restrained humour that still works when it is read by a synthetic voice. No slogans, no hype, no shouting.

Do not generate any images or video yet.

Create ONE photorealistic human identity reference sheet, NOT a storyboard and NOT a scene. Exactly three panels side by side on one horizontal image, separated by thin black dividers, on the SAME plain mid-grey seamless studio background: LEFT full-body front view standing, CENTRE full-body back view standing, RIGHT chest-up portrait facing camera. No words, labels, numbers or logos anywhere.

Lock identity across all three panels: one woman in her late twenties, average height and ordinary build, fair skin with visible freckles across the nose and cheeks, blue-grey eyes, light brown eyebrows, no makeup beyond a natural lip. Mid-brown hair worn up in a loose low bun with a few soft strands falling loose at both temples. Small thin gold hoop earrings, no other jewellery.

Wardrobe identical in every panel: an oversized cream washed-linen overshirt worn open, sleeves rolled twice to just below the elbow, one patch pocket on the wearer's right chest, plain cream buttons; a plain white ribbed cotton vest underneath; mid-blue straight-leg denim jeans with a slight crease at the knee, hem breaking at the ankle; cream canvas trainers with white soles.

Relaxed neutral posture, arms down at her sides, weight even, calm friendly closed smile. Feet fully visible in both full-body panels. Soft even studio daylight from the front left, gentle contact shadow on the floor, deep focus, neutral photographic colour, true skin texture with pores and fine hairs.

No props, no product, no dog, no furniture, no background scene, no second person, no cropping of head or feet, no beauty-filter smoothing, no bokeh, no cinematic grade, no watermark. This is the canonical creator identity to be reused exactly.

Create a photorealistic animal identity reference sheet, NOT a storyboard, with exactly three views of the SAME adult cream golden retriever on a plain warm off-white seamless ground: full-body standing left profile, full-body sitting front three-quarter, close-up head portrait. Three clean equal panels, no words or labels.

Lock identity across all views: medium-large adult golden retriever, healthy ordinary proportions, medium-length pale cream fur with slightly honey-coloured floppy ears, dark brown eyes, broad black nose, feathered tail, plain matte teal woven collar with one small grey buckle and no hanging tag. Calm curious expression, relaxed ears, natural gently open mouth on sitting view. Entire tail and four paws visible on standing view.

Real fur, daylight, neutral photographic colours, no cartoon, no breed mixing, no puppies, no costumes, no people, no toy, no box, no scene action, no watermark.

Create one photorealistic product identity reference image, NOT a storyboard or ad. Fictional PawPop dog wobble toy, unbranded. EXACTLY THREE components: 1) a broad matte teal rounded self-righting base, 20 cm diameter and 7 cm high, low domed upper surface with central recessed female screw socket and smoothly curved weighted underside; 2) a short matte coral flexible rubber stalk, smoothly tapered cylinder about 3 cm thick, exposed length 13 cm, integral thick male screw connector at its lower end and integral shorter male screw connector at its upper end, no separate adapters; 3) a large sunny-yellow 10 cm diameter rubber sphere with evenly spaced shallow rounded raised dots, recessed female threaded socket at its bottom. Assembled height about 30 cm.

Base and stalk and ball all coaxial, ball rests atop stalk, no additional collars, no metal spring, no rope, no electronics, no buttons, no batteries, no fourth piece. Teal #238E91, coral #EB7867, yellow #F2C94C. Plausible matte moulded rubber and polymer textures, subtly visible moulding seams.

Show ONLY ONE fully assembled toy centred and complete on warm off-white seamless background, front three-quarter view slightly above floor level showing base shape and ball texture. All three components connected in exact order teal base, coral stalk, yellow ball. Soft neutral daylight, gentle contact shadow, accurate scale relationships, commercial product photography with honest materials.

No hands, people, dog, package, extra views, loose parts, exploded diagram, lettering, logos, annotations or watermark. This is the canonical assembled-product design to be reused exactly.

@Image1 is the canonical assembled PawPop toy. Create ONE matching photorealistic reference image of this EXACT product DISASSEMBLED into exactly THREE separate physical pieces on the same warm off-white seamless ground. This is a parts identity photograph, NOT a storyboard, NOT an assembly sequence, NOT a diagram.

Preserve its actual colours, shapes, surface textures and relative proportions from the reference, including the teal low-domed rounded base, slightly hourglass coral stalk and yellow raised-dot sphere.

Lay out the three pieces horizontally with generous clear space, all fully visible, consistent scale: LEFT teal base lying upright with its centred recessed female threaded socket visible; CENTRE coral stalk laying on a diagonal on its side, showing integral stout male screw extensions at both ends; RIGHT yellow ball tilted so its one recessed female screw socket on the underside is visible.

The integral connectors are part of the stalk, NOT extra loose pieces; ball and base each have exactly one receiving socket. The stalk's main body exactly matches the reference assembled silhouette, and its hidden threaded ends disappear fully into the two sockets after assembly. Do not redesign any surface visible in the assembled reference. Teal base remains widest object; yellow sphere about half base width.

No extra washers, adapters, screws, caps, rods or parts. No assembled duplicate toy. True matte moulded rubber/polymer photography, subtle moulding seams, neutral diffuse daylight, soft contact shadows, deep focus. No hands, people, dog, package, text, arrows, measurements, logos or watermark.

Create ONE photorealistic empty living-room location reference photo for a casual phone-shot dog-product unboxing demo. @Image1 is the approved creator reference ONLY for harmonious cream-and-denim everyday styling, do not put this person or any person in the image. This is a location reference, NOT storyboard and NOT an action scene.

Wide view from a seated person's eye height, normal phone lens without ultra-wide distortion. Warm off-white walls, low oatmeal linen sofa centred on back wall, two simple cream cushions, large plain light-beige low-pile rug occupying the open foreground, pale oak floor at rug edges. Tall window with sheer white curtain on LEFT, soft neutral daylight from left. One small light-oak side table to RIGHT of sofa, plain ceramic bowl on it, single modest green plant in far right corner. Doorway at rear right, natural route into room.

At least 2 metres of uncluttered rug space in front of sofa for sitting and dog play. Camera composition clearly establishes fixed positions of window, sofa, side table and doorway. Honest lived-in fabric and floor textures, subtle sofa creases, inviting but ordinary home, deep focus.

No coffee table, no people, no dog, no toy, no box, no text, no branding, no mirrors, no multiple panels, no dramatic cinematic grade or bokeh.

Create ONE horizontal storyboard sheet for a casual UGC iPhone unboxing ad, NOT a finished ad and NOT a single scene. Exactly FOUR vertical 9:16 panels of equal width, side by side in one row, separated by thin white gutters, read left to right as chronological shots. No captions, labels, numbers, arrows, titles, logos or watermark anywhere on the sheet.

@Image1 is the sole creator identity: the same woman in every panel, same face, same loose bun, same cream open overshirt with rolled sleeves, white top, blue jeans, cream trainers. @Image2 is the same single adult cream golden retriever with honey ears and a teal collar. @Image3 is the canonical assembled PawPop toy; @Image4 its matching THREE loose components; no other parts exist. @Image5 is the fixed living room: keep the sofa, rug, oak floor, window and sheer curtain on the LEFT, side table and plant on the RIGHT, and daylight from the left in every panel.

PANEL 1, MEDIUM-WIDE eye-level: the creator kneeling on the rug, both hands on opposite sides of a SEALED plain brown cardboard box she is lowering to the floor, top flaps shut with intact paper tape. The dog stands beside her sniffing the box corner. No product visible.
PANEL 2, TIGHT OVERHEAD looking straight down: the same box on the rug, her left hand steadying it, her right hand pulling the paper tear strip along the top seam. Box still closed. No product visible.
PANEL 3, OVERHEAD MEDIUM: the opened box and its moulded tray at the top edge of frame, the teal base and coral stalk already lying on the rug, her right hand lifting the yellow ball clear of the tray. Exactly three components in frame, no duplicates.
PANEL 4, CLOSE-UP low to the rug: her left hand steadying the teal base, her right hand turning the coral stalk clockwise into the base socket. The yellow ball rests loose to the right, not attached.

Consistent scale across panels: assembled toy about 30 cm tall, base about 20 cm across. Casual handheld phone framing, neutral daylight, deep focus, honest domestic materials. No extra people, no second dog, no duplicated toys, no text, no split panels within a panel, no bokeh, no cinematic grade.

Create ONE horizontal storyboard sheet continuing the same UGC unboxing ad, matching the first sheet exactly in layout and treatment. Exactly FOUR vertical 9:16 panels of equal width, side by side in one row, thin white gutters, read left to right. No captions, labels, numbers, arrows, titles, logos or watermark.

Same references and same rules as sheet one: @Image1 sole creator identity, @Image2 the same golden retriever, @Image3 the assembled PawPop, @Image4 its three components, @Image5 the fixed living room with daylight from the left. No box, tray or packaging in any panel.

PANEL 1, TIGHT CLOSE-UP low to the rug: the coral stalk already screwed upright into the teal base, her left hand holding the base steady, her right hand lowering the yellow ball onto the top thread and turning it clockwise.
PANEL 2, MEDIUM: the creator kneeling behind the fully assembled toy on the rug, the dog sitting beside her looking between her and the toy, her right hand resting lightly on the base.
PANEL 3, LOW DOG-HEIGHT CLOSE-MEDIUM: her hands withdrawn into her lap, the dog nudging the yellow ball once with its nose, the coral stalk bending modestly away and the base rocking a few degrees while staying on the floor.
PANEL 4, WIDE THREE-QUARTER: the dog in a playful bow facing the wobbling toy, front paws down and hindquarters raised, tail up, the creator kneeling behind watching with her hands on her thighs.

Exactly one assembled toy, intact, in every panel. Same scale as sheet one. Casual handheld phone framing, neutral daylight, deep focus. No extra people, no second dog, no duplicated toys, no packaging, no text, no bokeh, no cinematic grade.

Create the video from the approved storyboard.

Generate it in two clips, the first covering sheet 1 and the second covering sheet 2, then join them.

For each clip, write the full prompt yourself and show it to me before you render:
- attach the matching storyboard sheet and all five references
- use the storyboard as the running order and staging
- one static locked camera per shot, hard cuts only, no slow motion or morph transitions
- the exact second each cut lands
- what each hand is doing in every shot
- the exact spoken words over each shot, taken from the approved script
- one consistent voice, described the same way word for word in both clips
- the product mechanics spelled out: the stalk screws into the base, the ball screws onto the stalk, there are no other parts, only the stalk bends, and the ball never comes off
- no on-screen text, subtitles, watermarks, logos or music

Vertical 9:16, authentic casual phone footage, natural daylight, deep focus, real skin and fur texture.

@Image1 is my product, a vitamin C face serum in an amber dropper bottle.

Create a before-and-after UGC ad concept for it: one creator, filmed on her own phone at home, fifteen seconds, vertical.

Give me the idea in one line, then a shot-by-shot script with timings and the exact words she says. The visible result is the whole ad, so the change in her skin has to carry it. Plan one hidden cut in the middle, hidden behind her hand passing across the lens, so the before and after are the same person in the same place with the same light.

Keep the writing plain and spoken, no slogans, no hype. Do not generate any images yet.

Create ONE photorealistic human identity reference sheet, NOT a storyboard and NOT a scene. Exactly three panels side by side on one horizontal image, thin black dividers, same plain light-grey seamless studio background: LEFT full-body front standing, CENTRE full-body back standing, RIGHT chest-up portrait facing camera. No words, labels or logos.

Lock identity across all three panels: one woman in her late twenties, warm light-olive skin, dark brown eyes, full defined dark eyebrows, natural glossy pink lip, soft everyday makeup. Shoulder-length mid-brown hair with subtle lighter ends, centre parting, loose natural waves tucked behind one ear.

Jewellery identical in every panel: small gold hoop earrings, one fine gold chain necklace with a small round gold coin pendant sitting at the collarbone, a second shorter delicate gold chain, two thin gold rings on the right hand and one on the left.

Wardrobe identical in every panel: a cropped teal-green boucle tweed jacket with a collarless round neck, flecks of pink and cream in the weave, gold-toned buttons, worn open; a bright red fine-ribbed knit top underneath; plain dark denim straight jeans.

Relaxed neutral posture, arms at her sides, calm friendly closed smile. Soft even studio daylight from the front left, deep focus, neutral photographic colour, true skin texture with visible pores and fine hairs.

No props, no product, no second person, no background scene, no cropped head or feet, no beauty-filter smoothing, no bokeh, no cinematic grade, no watermark. This is the canonical creator identity to be reused exactly.

Create ONE photorealistic empty location reference photo for a casual phone-shot skincare demo at home. This is a location reference, NOT a storyboard and NOT an action scene. No people, no product, no hands.

A bright upstairs hallway landing in an ordinary home. Off-white matte walls. A round wall mirror with a thin brushed-gold frame hangs on the back wall at head height. An open doorway to the LEFT lets in soft neutral daylight, which is the only light source. Pale wood floor, one small plant on the floor in the corner, nothing else.

Camera at standing eye height, normal phone lens, no ultra-wide distortion, framed so there is clear space in front of the mirror for one person to stand and film themselves at arm's length. Honest lived-in surfaces, faint wall texture, deep focus, neutral colour.

No people, no reflection of a person in the mirror, no product, no bathroom fittings, no text, no branding, no multiple panels, no bokeh, no cinematic grade, no watermark. This is the fixed location for every shot of this ad.

Create ONE horizontal storyboard sheet for a casual UGC iPhone skincare ad, NOT a finished ad. Exactly FOUR vertical 9:16 panels of equal width, side by side in one row, thin white gutters, read left to right as chronological shots. No captions, labels, numbers, arrows, logos or watermark.

@Image1 is the sole creator identity: same woman in every panel, same face, same hair, same teal boucle jacket, red ribbed top and gold jewellery. @Image2 is the exact product, unchanged in bottle shape, collar and label artwork. @Image3 is the fixed location: same hallway, same round gold mirror behind her, same daylight from the left in every panel.

All four panels are handheld selfie framing, chest-up, arm's length, same distance and same eye height throughout.
PANEL 1: BEFORE. Her bare skin looks flat, dull and tired, slightly uneven tone, a little shine on the forehead. She points two fingers at her own cheek.
PANEL 2: she holds the open bottle up beside her face in her right hand, the glass pipette and amber oil clearly visible.
PANEL 3: AFTER. Identical framing, identical location and light, identical hair, jacket and jewellery, identical face shape and features. Only the skin changes: even tone, soft natural dewy sheen on the cheekbones and nose, rested. She tilts her face slightly into the light.
PANEL 4: she holds the closed bottle beside her face, label flat and square to camera, reading correctly left to right.

True skin texture with pores in every panel, flat natural daylight, deep focus. No second person, no duplicated bottle, no text, no split screen, no bokeh, no beauty filter, no cinematic grade.

Create the video from the approved storyboard.

Fifteen seconds, vertical 9:16, one continuous piece with a single hidden cut. Write the full prompt yourself and show it to me before you render:
- attach the storyboard, the creator sheet, the product photo and the location
- handheld selfie framing throughout, same distance and same light before and after
- the before and after are the same woman in the same place: only her skin changes, never her face shape, features, hair, clothing or jewellery
- hide the change behind her free hand sweeping across the lens, filling the frame for a few frames. No dissolve, no morph, no visible cut underneath
- the exact spoken words from the approved script, one consistent voice
- in the closing shot she holds the bottle still and square to the lens, label reading correctly left to right
- no on-screen text, subtitles, watermarks, logos or music

Authentic casual phone footage, natural daylight, deep focus, true skin texture with pores.

The product label in the final shot is mirrored, so the brand name reads backwards.

Regenerate the video with everything else identical: same creator, same location, same light, same shots, same cuts, same spoken words.

Add these three rules to the final shot and change nothing else:
1. The label reads correctly left to right, exactly as it appears in the product reference image. Never mirrored, flipped, warped or re-spelled.
2. She holds the bottle STILL and SQUARE to the lens, label facing camera, at chest height below her chin, for at least one full second, with no rotation and no hand movement.
3. If the lettering cannot be rendered legibly, frame the bottle so the label is partly out of shot rather than showing incorrect text.

@Image1 is my product, a toy water blaster. @Image2 is the creator: same face, hair and clothing in every shot.

Create a how-to UGC ad that teaches someone to use it in four steps: open the cap, add water, close the cap, aim and spray. Write the script first, one short spoken line per step, plain and instructional, no hype.

Then create ONE horizontal storyboard sheet: exactly FOUR vertical 9:16 panels of equal width, side by side in one row, thin white gutters, read left to right, one panel per step. BAKE a short step caption into the top of each panel in clean white sans-serif text, two lines maximum, reading "Step 1 — Open Cap", "Step 2 — Add Water", "Step 3 — Close Cap", "Step 4 — Aim & Spray". No other text anywhere.

Panels 1 to 3 are close, table-height shots in a garden: the product resting on a wooden table, her hands working on it, her face visible above. Panel 4 steps back to a standing three-quarter shot in the same garden with the product held in both hands.

Same creator, same product and same daylight in every panel. Product unbranded. Casual phone framing, true skin texture, deep focus. No extra people, no duplicated product, no arrows, no logos, no watermark, no bokeh, no cinematic grade.

KEEP EXACTLY the framing, composition, slot layout, camera distances, poses, subjects, product AND any on-frame step captions of this horizontal storyboard sheet and every one of its side-by-side vertical slots — no reframe, no zoom, no crop, no re-layout, no change to the scene, to any person's face / hair / body, to the product design, or to existing on-frame text.

CHANGE ONLY micro-realism, applied identically in every slot: true-to-life pore-level skin with natural texture and fine vellus hair, real material detail, even natural daytime light with gentle highlight roll-off and faint true sensor noise, a flat authentic iPhone photo, deep focus.

PRESERVE each face's exact shape / width / proportions 1:1 — do NOT squeeze / narrow / slim / stretch any face.

AVOID AI-slop: waxy plastic skin, airbrushed poreless skin, beauty-filter smoothing, over-saturation, HDR glow / bloom / halos, oversharpening, teal-orange grade, shallow depth of field, bokeh, cinematic / DSLR look.

Keep the product blank / unbranded, no NEW added text, no watermark.

Create the video from the approved storyboard, and keep the step captions exactly as they are baked into each panel — same wording, same position, same size, no new text.

Write the full prompt yourself and show it to me before you render. One static camera per shot, hard cuts between steps, the exact spoken line over each step, one consistent voice, vertical 9:16, authentic casual phone footage in natural daylight.

No extra on-screen text, subtitles, watermarks, logos or music.

Use the attached creator and product, take it from concept and script all the way to a finished vertical video.