How to Make a GRWM Video with AI (Get Ready With Me)
The 4-step workflow to make a GRWM video with AI: identity locking for one consistent face, routine-step storyboard, lip-synced voiceover, and product integration.
You can make a GRWM (Get Ready With Me) video with AI in four steps: lock one avatar identity with Higgsfield Soul 2.0, storyboard the routine as a sequence of steps, generate each step with the same face, then layer voiceover and lip sync. A 20 to 30 second GRWM runs about $8 to $14 in model credits on the 8frame canvas, versus the $150 to $600 a beauty creator charges for one video. The entire format lives or dies on character consistency, so that is where most of this guide sits.
TL;DR
- Step 1: Generate one reference portrait in Nano Banana Pro, then lock it as your avatar in Higgsfield Soul 2.0
- Step 2: Storyboard the routine as 4 to 6 steps (cleanse, serum, makeup, hair, outfit, final look)
- Step 3: Generate each step clip with the same locked identity so the face never drifts
- Step 4: Add voiceover and lip sync, integrate the product beats, add captions
- Cost: $8 to $14 per finished GRWM versus $150 to $600 for a creator
GRWM as an ad format
GRWM works as a soft-sell ad because it embeds the product inside a routine instead of pitching it. The viewer watches someone get ready and the product appears as a natural step, which lowers the ad resistance that a direct claim triggers. For beauty, skincare, and fashion brands it is one of the highest-retention formats on TikTok and Reels because the "what's the final look" question pulls viewers to the end.
The catch is that GRWM is a single-character format across a whole sequence. The same person cleanses, applies serum, does makeup, and reveals the final look, and if her face changes between steps the illusion collapses. That is the one hard problem, and it is solvable.
The 4-step workflow
Step 1: Lock the avatar identity
Character consistency is the whole game. Start by generating a single clean reference portrait in Nano Banana Pro: front-facing, neutral to warm expression, even lighting, no heavy makeup so the routine has somewhere to go.
Portrait of a woman in her late 20s, natural bare skin, soft even light, front-facing, neutral relaxed expression, plain background, photorealistic
Upload that portrait into Higgsfield Soul 2.0 as your locked identity. Identity locking anchors the avatar's face, skin tone, and micro-expressions to that one reference so every step clip returns the same person. Do not change the reference between steps. If you generate across multiple sessions, reload the same portrait every time. This is the single most important discipline in the whole workflow; skip it and you get same-face fatigue in reverse, a face that shifts subtly and reads as uncanny.
Step 2: Storyboard the routine steps
A GRWM is a sequence, so plan the beats before generating. A tight beauty GRWM runs 4 to 6 steps:
- Bare-skin open, direct to camera hook line
- Cleanse or prep
- The hero product step (serum, foundation, whatever you are selling)
- Makeup or styling
- Hair or outfit
- Final look reveal, direct to camera CTA
Write the hook and the CTA first. The hook in the first 2 seconds decides whether the viewer stays: "Getting ready for a first date and my skin has to behave" is a scenario the target viewer recognizes. The final-look reveal is the payoff you promised in the hook.
Step 3: Generate each step with the locked face
Generate each storyboard beat as its own clip in Higgsfield Soul 2.0, all referencing the same locked portrait. Keep camera framing consistent, mostly medium and close shots, so the cuts feel like one continuous getting-ready session.
[Locked avatar] in a softly lit bathroom applies vitamin C serum with a dropper to her cheek, looking slightly down at the mirror, natural morning light, vertical 9:16, UGC handheld feel, 4 seconds
Generate three variants per step and select the most natural take. For the hero product step, drop in a Seedance 2.0 product-in-hand shot: upload the actual product photo as a reference so the serum bottle in frame is your real SKU, not a generic dropper. For fast b-roll and transition beats where the face is out of frame, Kling 3.0 at $0.28 to $0.40 per clip covers volume cheaply.
Step 4: Voiceover, lip sync, and product integration
GRWM usually carries a casual voiceover narrating the routine. Write it conversational and short. Generate or record the audio, then apply lip sync to the direct-to-camera clips so the mouth matches the words. Off-camera narration over b-roll steps does not need lip sync, which saves generation effort; reserve lip sync for the hook and CTA beats where the avatar faces the viewer.
Integrate the product at the hero step and again at the reveal. The pattern that converts: show the product being used mid-routine, then reference the result at the final look. "This serum is the only reason my skin looks like this" lands harder at the reveal than as an upfront claim.
Cost math versus a creator
| Line item | AI GRWM | Creator GRWM |
|---|---|---|
| Reference portrait (Nano Banana Pro) | $0.08 | included |
| 5 to 6 step clips (Higgsfield Soul 2.0) | $6 to $10 | included |
| Product-in-hand shot (Seedance 2.0) | $0.60 | included |
| Transitions and b-roll (Kling 3.0) | $1 to $2 | included |
| Total per finished cut | $8 to $14 | $150 to $600 |
| Turnaround | same day | 5 to 14 days |
A beauty creator GRWM runs $150 to $600 per video, plus 30 to 50% for usage rights on paid media. At $10 to $14 a cut you can produce a week of routine variations for the cost of a single creator booking, which is what makes AI GRWM a testing engine rather than a one-off.
Disclosure for sponsored GRWM
A GRWM that promotes a product for a brand is sponsored content, and an AI GRWM is also AI-generated. Both facts trigger disclosure. On paid placements, TikTok and Meta require the AI-generated content label applied at upload. For a sponsored routine you also owe a partnership disclosure under the platform's branded-content tools. Critically, the avatar is not a real customer sharing a genuine routine, so never frame the video as a real person's authentic experience; frame it as brand creative. Our guide on disclosing AI-generated ads covers the current requirements, and they update often.
FAQ
How do I keep the avatar's face consistent across every routine step?
Lock one reference portrait in Higgsfield Soul 2.0 and use that exact image for every step clip. Do not swap references between the cleanse step and the final reveal. If you work across multiple sessions, save the portrait and reload it each time, because the identity conditioning is anchored to that specific image. Keeping framing consistent (medium and close shots) also helps the cuts read as one continuous session.
Which model handles the product application shots?
Seedance 2.0 for any shot where the product is in hand or being applied, because its multi-reference conditioning lets you upload the real product so the bottle or compact in frame matches your SKU. Use Higgsfield Soul 2.0 for the face-forward routine steps and Kling 3.0 for fast transitions and b-roll where the face is out of frame.
Do GRWM videos need voiceover?
Most high-performing GRWMs carry a casual voiceover, but not always. A silent, caption-driven GRWM with strong on-screen text works when the visual routine is the story. If you do use voiceover, apply lip sync only to the direct-to-camera hook and CTA beats and leave the b-roll steps as narration, which saves generation effort.
How long should a GRWM ad be?
20 to 30 seconds for paid social. The format needs enough runtime to move through the routine steps and deliver the final-look payoff, but the hook has to promise that payoff in the first 2 seconds. Longer organic GRWMs exist, but for ad creative keep it tight and land the reveal before attention drops.
The workflow is four steps and $8 to $14 in compute: one locked identity, a storyboarded routine, per-step clips that never let the face drift, and a voiceover with lip sync on the face-forward beats.
Generate your reference portrait and lock it in Higgsfield Soul 2.0 at app.8frame.co, then build the routine one step at a time. For the broader UGC production playbook, see how to make a UGC ad with AI, and for the reveal-driven cousin of this format, how to make an unboxing video with AI.