How to Create AI UGC Ads With Plyry
By the Plyry editorial team · Published and reviewed October 15, 2026 · 12 minute read
TL;DR: Start with one product, three shots, and one clear benefit. Create three consistent images in Plyry Image, animate each approved image in Plyry Video, then download the clips and assemble a short ad in your video editor. Once that works, add a recurring character, a pet, more story beats, voiceover, or real product footage. Save the prompts that work in Prompt Studio so the next version starts from a useful template.
UGC means user-generated content. Here, AI UGC means an ad created with AI in the style of an everyday creator video.
You can build a useful first ad around a small, visible action: keys going into a desk organizer, a bag opening to reveal its compartments, or a cushion finding its place on a sofa. A few connected shots make the product understandable and give the viewer a reason to care.
This guide starts with a three-shot product ad, then builds up to a woman unboxing a sofa with her dog nearby. Use the models and reference modes available in your Plyry account; supported inputs, clip lengths, audio options, and prices depend on the selected model.
Start With a Three-Shot Ad
Keep the first project small. Choose one product, one location, one benefit, and one call to action. Leave dialogue, extra characters, and complex assembly for a later version.
For example, a 15-second desk-organizer ad could use this plan:
| Edit time | Shot | What the viewer learns |
|---|---|---|
| 0:00–0:04 | Wide view of a desk with scattered keys and earbuds beside the organizer | The everyday situation |
| 0:04–0:10 | Close-up of a hand placing the keys into one compartment | How the actual product is used |
| 0:10–0:15 | Wider view of the organized desk, with the same items visible | The finished setup and next step |
Add a short opening caption such as A place for the things you keep looking for, then close with See sizes and colors. Use those lines only when the product and destination page support them.
These are timings for the finished edit. Generate the clip length your chosen model supports, then trim the useful action to fit. Your first deliverable is three approved images and three accepted clips; rerenders may require additional paid generations.
Gather the References and Product Facts
Before generating, collect a clear product photo, a few detail or alternate-angle photos, the product dimensions, and the feature you want to show. Add the exact offer and the destination page. For software, use a real screen recording of the current interface.
Choose a benefit you can show. A compartment holding keys is visible. A claim that an organizer saves everyone ten minutes every morning needs separate evidence. Keep price, delivery, discounts, and compatibility aligned with the page people will visit.
Use photos you own or have permission to use. An existing creator ad can inspire the story structure, but build your own scene, character, and wording.
Create the First Image in Plyry
- Open Plyry Image and choose an image model that accepts the references you need.
- Select Image reference or Multiple image references, when offered. Upload the product photo or select it from the media library.
- Describe the first shot: product, setting, framing, lighting, and what must stay accurate.
- Choose a vertical composition for a vertical ad, using the available size or aspect controls. Check the displayed charge before generating.
- Review the result and keep an approved version in your library. Download a backup of important assets.
If the reference control is unavailable, select a model that supports image inputs. The model picker determines which reference modes and files the generator can accept.
First-image prompt
Vertical 9:16 photograph for a casual creator-style product video.
A desk organizer matching the attached product photo sits on a small
wooden desk beside scattered keys and wireless earbuds.
Eye-level medium-wide framing, soft window light from camera left,
ordinary lived-in room, natural colors, enough space to see the desk.
Preserve the product's exact shape, color, compartments, and materials.
Do not add features, printed text, captions, or a logo that is not
present in the product reference.
Approve the product before moving on. Check compartment count, proportions, labels, and materials against the real reference. A visually attractive image can still show the wrong object.
Build the Other Shots From the Approved Reference
Use the approved image as a reference for the next image edit. Keep the product photos available too, within the selected model's reference limit. Change the action and framing while preserving the room, light, and product.
Detail-shot prompt
Use the approved desk image for the room and lighting, and the product
photo for the organizer's exact design.
Create a closer view of the same organizer. One adult hand holds the
same keys just above an empty compartment, ready to put them down.
Keep the earbuds beside the organizer.
Same desk, product, daylight direction, colors, and materials.
Natural phone-camera appearance, vertical 9:16, no added text.
For the final still, show the keys and earbuds in their actual compartments on the same desk. Keep it possible to recognize the same objects across the sequence.
Review all three images together. Fix product changes and layout mistakes before animation. Continue using the approved product and scene references instead of relying only on a chain of increasingly altered outputs.
Animate Each Image in Plyry Video
Open Plyry Video, select a model with Image reference, and attach one approved shot image as its starting frame. Check the reference slot labels: an opening frame establishes the composition, while a separate reference mode may guide a subject without fixing the opening shot.
Write what should move during this clip. Keep the action small and the camera simple. For the organizer detail shot:
Starting from the supplied image, the hand gently lowers the keys
into the empty compartment and releases them. The keys settle.
Locked-off camera. Keep the organizer rigid and unchanged.
The desk, earbuds, lighting, and background remain consistent.
No additional objects, cuts, dialogue, captions, or music.
Use one main action per clip. For the opening shot, a subtle camera move may be enough. For the last shot, keep the scene still or use a gentle move toward the finished setup. Motion instructions are requests, so inspect the output for extra movement or unwanted sound.
Generate a short test where the model offers a suitable lower-cost option. Watch the full clip: hands should release objects, compartments should remain rigid, and nothing should appear or disappear. Regenerate a failed shot instead of rebuilding accepted shots.
Assemble the First Ad
Download the accepted clips from your Plyry library and put them on a timeline in your video editor. For this workflow, Plyry supplies the generated assets and reusable prompts; final cutting, caption placement, and the assembled export happen in the editor you choose.
- Put the situation first, the useful action second, and the finished setup last.
- Trim waiting time before and after each action. Cut when the viewer understands the beat.
- Add a readable opening caption, product name, and one call to action.
- Add suitable sound effects and commercially licensed music if they help.
- Preview on a phone, with sound and muted. Keep essential text away from interface overlays and screen edges.
Keep an editable master so you can change captions, music, and the closing offer for different placements. Add text and logos in the editor for reliable spelling and placement.
Add a Story: The Sofa and the Dog
Once the three-shot workflow works, expand the story. A woman receives a sofa, unpacks it, assembles it in stages, and sits down with her dog. The product has a visible role in every beat, and the ending provides a small emotional payoff.
Plan the states between cuts. Each image should establish a believable stage of the process. Animate one action within that stage, then cut to the next approved stage. Follow the real product's packaging and assembly instructions.
| Edit time | Frame to create | Action to animate |
|---|---|---|
| 0:00–0:03 | Woman beside the delivery boxes; dog nearby | Dog sniffs one box |
| 0:03–0:06 | Same room, packaging partly opened | Woman pulls back one piece of wrapping |
| 0:06–0:10 | Components arranged in the correct assembly order | Woman positions one component |
| 0:10–0:14 | Sofa frame assembled; one cushion ready | Woman slides that cushion into place |
| 0:14–0:18 | Finished sofa in the same room | Woman sits down |
| 0:18–0:22 | Woman and dog settled on the sofa | Small relaxed movement; closing caption added in editing |
These are six separate shot images, not a single collage to animate. A contact sheet can help review continuity; each video clip needs its own correctly framed input image.
Keep the Person, Pet, Product, and Room Consistent
The first image establishes the look; a reference set helps preserve it. Keep approved views of the recurring subjects and a short text description of the details that must stay fixed.
| Subject | Details to preserve |
|---|---|
| Person | Face, adult age range, hair, clothing, accessories, proportions |
| Dog | Size, coat color, distinctive markings, ears, collar |
| Sofa | Shape, module count, fabric, seams, legs, cushions, scale |
| Room | Window position, wall color, floor, surrounding furniture, light direction |
For a recurring character, generate a neutral front and three-quarter view in Plyry Image. Use the approved first image as a visual reference. Keep lighting simple so the sheet describes identity clearly. The prompt guide explains reference sheets in more detail.
An image-edit instruction for a later sofa shot could be:
Use the attached approved references for the same adult woman,
the same dog, the exact sofa, and the same living room.
New shot: medium side view. The sofa is assembled except for the
last seat cushion. The woman holds that cushion just above its place.
The dog lies on the floor beside her, away from the assembly area.
Keep face, hair, outfit, dog markings, collar, sofa fabric and seams,
window position, and daylight direction unchanged. Vertical 9:16.
If the model accepts fewer references than you need, simplify the shot or create and inspect a compact reference sheet. More subjects create more details to check. Keep the dog relaxed in a fixed place while the woman performs the main action.
Save the Workflow as Reusable Prompts
Prompt Studio lets you save the parts you will use again. Start with a product description, a scene description, and a shot template. Add a character and pet block when the story needs them.
## UGC Product Shot
[product] in [setting].
[action]
Natural window light, [framing], casual phone-camera appearance,
vertical 9:16. Match the attached product reference accurately.
Keep the established colors, materials, and room layout.
After saving the template, reference it using its actual identifier. The code below is illustrative; replace it with the identifier Plyry assigns to your saved prompt:
@12345("UGC Product Shot", "[product]"=>"the desk organizer",
"[setting]"=>"the same small wooden desk",
"[action]"=>"A hand holds keys above the empty compartment.",
"[framing]"=>"close-up, locked-off camera")
The first ## comment names the prompt. References use the unique code, and replacements use =>. Keep the approved visual references attached through the selected generator's reference controls alongside the written identity instructions.
Add More Possibilities One at a Time
| Addition | How to try it | What to check |
|---|---|---|
| Voiceover | Write short narration, generate suitable speech in Plyry Audio, then place it over the clips | Pronunciation, pacing, truthful wording, commercial rights |
| Speaking character | Select a video model that supports the required speech/audio options and test one sentence | Mouth movement, voice, timing, identity |
| Music | Create a suitable track in Plyry Music or use properly licensed music | Paid-ad rights and voice intelligibility |
| Start and end frames | When the chosen model exposes both slots, provide approved starting and ending states | A plausible path between the states |
| Multiple references | Use the mode and file types offered by the selected model | Which references control identity, objects, and scene |
| Real footage | Put real product footage or screen recordings between generated lifestyle shots | The demonstration matches the actual product |
| More formats | Try unboxing, how-to, day-in-the-life, or an objection answer | One clear reason to buy per version |
| More placements | Reframe and edit for TikTok, Reels, Meta feeds, or YouTube | Crops, safe zones, captions, labels, and destination-specific music rights |
For a simple voiceover, describe what is visibly happening: One place for keys and earbuds. Check the dimensions and choose a color. Synthetic narration does not make a character's mouth match automatically; use it over product footage unless you have checked the speaking output.
For software ads, preserve the real interface in a screen recording. Use generated scenes to introduce the situation or close the story without inventing buttons, results, or product behavior.
Make Variations That Answer a Question
Keep your accepted body clips and create a few alternate openings:
- Situation: scattered everyday items, then the organizer.
- Product detail: a close-up of the most useful compartment.
- Use first: the keys landing in their place immediately.
Change the opening image, caption, or narration deliberately. Then try a different benefit or story format as a separate concept. Record which prompt and assets belong to each version.
Judge early attention alongside clicks and actual purchases or qualified leads. A video that holds attention can still create the wrong expectation. Keep the offer and landing page consistent while comparing versions, and allow for the normal delay before conversions appear.
Keep Generation Costs Under Control
Approve images before buying motion. Fixing the product or face in a still prevents paying repeatedly to animate a mistake. Reuse accepted clips, keep actions focused, and check the charge whenever you change the model, reference mode, duration, or quality.
The price shown in Plyry is the price to use for planning. Available models and settings change; an old tutorial's price or input limit may not match your account. Test quality, where offered, is a separate generation and does not guarantee that a later full-quality result will be identical.
Keep product photos, approved images, exported clips, and the final edit backed up. Browser-library storage is useful for creation, but clearing browser data can remove locally delivered media.
Troubleshoot the Shot That Failed
| Problem | Next adjustment |
|---|---|
| Face or dog markings drift | Return to approved references; reduce pose changes and background activity |
| Product shape changes | Add a clear product angle, tighten framing, and reduce motion |
| Hands merge with the product | Simplify the interaction, choose an easier starting pose, or use real footage |
| Assembly jumps make no sense | Correct the still-image stages against the real assembly sequence |
| Camera motion overwhelms the scene | Use a locked-off camera and one subject action |
| Generated text is wrong | Remove it from generation and add it in the editor |
| A storyboard grid becomes the video | Export or create an individual shot image before animation |
| Reference upload is unavailable | Select a model with the required reference mode and accepted file type |
Publish a Clear and Truthful Ad
An AI-generated actor can present a product or appear in a fictional scene. Do not invent a customer purchase, personal experience, review, or result and present it as real. The FTC's avatar and testimonial guidance explains why false underlying testimonials remain a problem.
Use faces, voices, product assets, and music you have permission to use commercially. Label synthetic content as required by the destination and target market. For example, TikTok's advertising policy requires disclosure for significantly AI-generated or edited media. Check the current upload settings and the final displayed ad.
Frequently Asked Questions
Do I need a talking avatar?
No. Product actions, captions, sound, and a clear sequence are enough for a first version. Add a speaker when it improves the explanation or story.
Can I generate the entire ad in one request?
Some models support longer or multi-shot outputs. Start with separate shots so you can approve each product state and replace one failed clip. Try a longer request after you have a sequence you trust.
Do I need a particular image or video model?
Choose by the job: image generation/editing for consistent frames, image-to-video for animating them, and appropriate audio support for speech. Use the capabilities shown in Plyry's model and reference controls.
Should I animate a whole reference sheet?
Use the sheet to guide identity. Create a clean, individual scene image for the opening frame of each clip. A sheet with multiple panels can otherwise appear as a grid in the video.
Can the same workflow work for services or apps?
Yes. Show a real deliverable, process, or recorded app action in the middle. Build the surrounding scenes around the customer situation and a specific next step.
Keep Creating
Start with one product and three shots. Open Plyry Image to approve the frames, animate them in Plyry Video, and keep the reusable directions in Prompt Studio. Once the sequence is clear, add the person, pet, voice, or extra story beat that makes the next version more useful.
For more detail, read How to Write Better AI Media Prompts, compare free daily video-generation options, or explore Plyry workflows.
Start with three shots: open Plyry Image to create your first frame, then animate the approved images in the video studio. Have a question or an idea for this guide? Use the feedback button.