Created 5 AI product photos and 1 video ad in 20 minutes for Under $10 total, replacing thousands in traditional production costs.
AI product photography has reached quality parity with professional shoots for ecommerce ad creative.
The prompt is everything: specific descriptions of camera, lighting, and materials produce photorealistic output.
Kling V3.0 Pro generates the most realistic 8-second video clips for UGC-style social ads.
Budget for 2-3 regenerations per video. Not every first attempt is perfect, but the iteration cost is minimal.
Speed advantage matters more than cost: testing 30 ad variations in one afternoon beats waiting 3 weeks for one set.
Look at the images below. They were created in under 20 minutes using a single AI tool. No photographer, no model, no studio, no post-production. Total cost: about 93 credits, roughly $9.
One of the AI-generated product images created in a single 20-minute session
The video version is even more impressive. An 8-second UGC-style video ad, complete with natural movement, realistic lighting, and a model that does not exist. This is what AI video ads look like in 2026, and this tutorial shows you exactly how to create them.
The final AI-generated video ad output. Created in under 3 minutes using Kling V3.0 Pro.
What Are AI Video Ads and Why They Matter
AI video ads are short-form video advertisements created using artificial intelligence instead of traditional production. Instead of hiring models, booking studios, and managing weeks of production, you describe what you want, select an AI avatar or scene, and generate the video in minutes. The technology has reached a point where most viewers cannot tell the difference between AI-generated UGC and content filmed by a real person.
The shift matters because creative testing velocity is now the biggest competitive advantage in paid advertising. Brands that can generate 30 ad variations in an afternoon and test them against each other will consistently outperform brands stuck in 3-week production cycles. AI video ads make that velocity possible at a fraction of the cost.
Tyler Ang
Digital Marketing Consultant
After consulting with 255+ businesses, Tyler discovered most do not need more traffic. They need someone to look at their business properly first. He built sportifate.com to 6,800+ organic users with zero ad spend, proving the research-first system works.
Every month you run ads, post content, or pay for SEO without knowing what is actually working is another month of budget leaking, and in one conversation I can pull up your Google Ads, Search Console, and Analytics to show you exactly where the hole is and which underutilised areas deserve your attention first.
What you get from a 30-minute strategy call:
Full Platform Audit
A full breakdown of your current numbers across Google Ads, Meta, Search Console, and Analytics, showing where your money is going, what it is returning, and which underutilised areas could produce results faster
Biggest Constraint
The single biggest constraint holding your business back right now, identified from your actual platform data rather than guesswork
Step 1: Creating AI Product Images
The first step is generating the product images that will serve as the visual foundation for your ad. The tool I used for this tutorial lets you describe the scene you want, select a model that matches your target demographic, and generate multiple variations in minutes.
The image creation interface showing multiple generated product photos with settings for model, ratio, and quality
Here is how the process works. You start by selecting the type of content you want to create (photo in this case). Then you describe the scene: the setting, the mood, the style of clothing or product placement, and the type of model. The AI generates multiple variations based on your description.
The prompt is everything. A vague description produces generic results. A specific, detailed prompt that describes lighting, camera angle, materials, and mood produces images that look like they came from a professional shoot. I will not share the exact prompts I used here, but the principle is: describe what you see in your mind as if you are briefing a photographer.
In this session, I generated 6 product images in about 10 minutes. The AI produced variations with different poses, angles, and compositions. I selected the 5 strongest images for the campaign.
Another variation from the same generation batch, different pose and composition
Step 2: Creating the AI Video Ad
Once you have your product images, the next step is generating the video. The same platform lets you switch from photo mode to video mode and use your generated images as the starting frame for a video.
Video creation settings: Kling V3.0 Pro model, 8-second duration, 9:16 ratio for vertical ads
The key settings for video generation:
Model: Kling V3.0 Pro (the most realistic option for UGC-style content)
Duration: 8 seconds (optimal for social ad hooks)
Ratio: 9:16 (vertical, optimized for TikTok, Reels, Shorts)
First frame: select one of your generated images as the starting point
Last frame (optional): set where you want the video to end for smooth transitions
You describe the motion you want in the video: a gentle head turn, wind moving through hair, the model reaching toward the camera. The AI generates an 8-second clip with realistic movement based on your description. The video took approximately 3 minutes to render.
Step 3: The Final Output
The result is a set of 5 product images and 1 video ad, all visually consistent, all featuring the same AI model in the same setting. The entire session took 20 minutes and used about 93 credits (roughly $9 on the Basic plan).
Different scene variation from the same session, showing the versatility of AI-generated product content
These assets are ready to deploy directly into ad campaigns on Meta, TikTok, or Google. No post-production, no color grading, no editing software required. Export and upload.
What This Would Cost with Traditional Production
AI video ads vs traditional production cost comparison
Traditional
AI (This Tutorial)
Product Photos (5)
$1,500 - $4,000
~21 credits (~$2.03)
Video Ad (1)
$1,000 - $3,000
72 credits (~$6.98)
Model Fee
$500 - $2,000
$0
Studio/Location
$300 - $1,000
$0
Post-Production
$500 - $1,500
$0
Total
$3,800 - $11,500
~93 credits (~$9)
Turnaround
1-3 weeks
20 minutes
The speed advantage matters more than the cost advantage. When you can test 30 different ad variations in a single afternoon, you find winning creatives faster than competitors who wait 3 weeks for one set of assets. The brands that master AI ad creative testing velocity will have an unfair advantage in paid acquisition.
The 5 Kling prompts we used for the bus dress ad
Most AI video ad tutorials show the output and hide the inputs. This is the full prompt set we used on one of our best-performing Olivia's Lane dress ads, a 15-second K-beauty editorial in the style of Padini's viral campaign. Five clips, hard cuts, blur-focus pulls in CapCut. Garden setup, then four scenes inside an abandoned vintage bus. Every prompt below is the one we actually ran. Replace [DRESS] with your own product description and the workflow transfers.
Motion rules we follow on every clip
Never walk INTO camera. The model must not approach the lens, it breaks the editorial framing.
Walking shots are observer angles only. Side views or rear views, so the viewer sees how the dress looks in motion.
Show the dress moving. Walking, spinning, twirling. Static posing does not sell fabric.
No blinking. It looks uncanny and AI. Use chin tilts, gaze shifts, wind, or hand movements instead.
Tool settings we locked in
Image (Potion Ads): Nano Banana Pro, 2K, 4:5, Remix mode. Video (Kling): Pro tier, 4:5, 5 seconds per clip, uploading both first frame and last frame for interpolation. Clip 1 uses a Pinterest reference as Image 1. Clips 2 to 5 use the Clip 1 Potion output as Image 1 so the model face stays consistent across the whole ad.
Clip 1: Flower garden (front-facing, establishes the model)
This is the setup clip. The Potion output from this frame becomes Image 1 for every clip after, which is how we keep the same face in every shot.
Potion prompt (first frame):
Side view of young woman in [DRESS] walking through pink flowers in a garden. Mid-stride, dress swaying with movement. Soft dappled daylight. Flowers visible in foreground and background. 4:5.
Potion prompt (last frame):
Same side view among pink flowers, she has paused mid-step and is turning her body to face the other direction. Dress wrapping around her legs from the turn. Same dappled daylight. 4:5.
Kling motion prompt:
Woman walks slowly to the right in side profile, dress swaying with each step, then pauses and turns her upper body. One motion: walk + pause + turn. Stabilized side angle, camera static on tripod. Kling interpolates between first and last frame.
Clip 2: Abandoned bus, overhead angle
Scene change to the bus interior. Bird's eye framing so the dress is the star of the frame. The flower pluck motion connects into Clip 3 narratively.
Potion prompt (first frame):
Zoomed-in bird's eye of young woman in [DRESS] lying on worn leather seat of an abandoned vintage bus. Tropical flowers and vines visible through broken windows. One hand resting near her hair. Focus on her and the dress. Warm golden light from above. 4:5.
Potion prompt (last frame):
Same zoomed-in bird's eye, fingers now plucking a small flower from the vine. Same abandoned bus, same warm golden light. 4:5.
Kling motion prompt:
Fingers trace toward vine and gently pluck a small flower. One motion: slow pluck. Static overhead camera. Kling interpolates between first and last frame.
Clip 3: Close-up with flowers, still in the bus
Tighter medium shot inside the same bus. Shows the dress fabric in the air as she walks through the aisle and half-spins, which connects into Clip 4.
Potion prompt (first frame):
Medium shot of young woman in [DRESS] inside the abandoned bus, standing in the aisle between seats. She is mid-step walking through the bus, one hand trailing along the vine-covered seat backs. Flowers framing the scene. Warm golden light from the windows. 4:5.
Potion prompt (last frame):
Same medium shot inside bus, she has walked further down the aisle and is doing a small half-spin, dress fabric catching the air. Hair swinging with the turn. Same warm golden light. 4:5.
Kling motion prompt:
Woman walks slowly through the bus aisle, one hand trailing along vine-covered seats, then does a graceful half-spin. Dress fabric fans outward with the turn. One motion: walk + half-spin. Stabilized medium shot, camera static on tripod. Warm golden light from bus windows consistent throughout. Preserve the dress fabric, flower vines, and bus interior exactly as shown in the frames.
Clip 4: Side profile walk, still in the bus
Rear three-quarter angle so viewers see the dress silhouette from behind. The shoulder glance at the end sets up the spin in Clip 5.
Potion prompt (first frame):
Side profile of young woman in [DRESS] walking through the abandoned bus. Rear three-quarter angle so viewers can see how the dress falls and moves from behind. Vines and flowers framing the bus windows. Warm golden backlight through windows. 4:5.
Potion prompt (last frame):
Same side profile angle, she has stopped walking and is looking over her shoulder back toward camera. Dress settled around her legs. Hair caught by breeze from broken window. Same warm golden backlight. 4:5.
Kling motion prompt:
Woman walks away in rear three-quarter view, dress swaying with each step, then pauses and looks over her shoulder. Hair lifts in breeze from broken window. One motion: walk away + shoulder glance. Stabilized medium shot, camera static on tripod. Warm golden backlight from bus windows. Preserve the dress fabric, flower vines, and bus interior exactly as shown in the frames.
Clip 5: The spin, full motion inside the bus
The hero clip. First frame is Clip 4's last frame (she is already looking over her shoulder, so Kling picks up mid-motion). Last frame is a fresh Potion render of the dress mid-spin.
Waist-up shot of young woman in yellow [DRESS] mid-spin inside the abandoned bus. Dress fabric fanning out around her, hair flowing down and swinging with the turn. One hand extended, the other near her chest. Flower petals caught in the movement. Warm golden light from bus windows. Focus on her and the spinning dress. 4:5.
Kling motion prompt:
Woman turns her body into a slow graceful spin to the left. Her hair falls and swings with the rotation as the dress fabric fans outward around her waist. Arms extend naturally with the turn. Smooth, slow, continuous rotation over the full 5 seconds. Stabilized medium shot, camera completely static on tripod. Warm golden light from bus windows consistent throughout. Preserve the yellow dress fabric, flower vines, and bus interior exactly as shown in the frames.
The tool I used for this tutorial is Potion Ads, which is built specifically for ecommerce ad creative. But the AI video ad space has several strong options depending on your use case. I wrote a detailed comparison of Higgsfield AI, Arcads, and Potion Ads that breaks down pricing, features, and which tool fits which workflow.
AI video ad tools comparison
Tool
Best For
Starting Price
Key Strength
Potion Ads
eCommerce UGC + product photos
Free (7 credits)
Demographic avatars, product photography, competitor ad copying
Arcads
Most realistic talking heads
$110/mo
300+ real actor avatars, best lip-sync
Higgsfield AI
Cinematic content
Free (10 credits/day)
15+ AI models, 70+ camera presets
Creatify
URL-to-video
$29/mo
Paste product URL, get video ad
For a detailed comparison of Arcads vs Potion Ads, including pricing tiers and feature breakdowns, read my full Arcads review.
Try It Yourself (15% Off Inside)
The 5 product photos and 1 video ad in this tutorial cost $9 total. A traditional photoshoot in Singapore runs $500 to $2,000 per session before you even add video. If you are testing ad creative for an ecommerce brand, the math is not close.
Use this link to get 15% off Potion Ads. Start with the free tier to test output quality on your own products. If the results look good, upgrade and you will still spend less on a month of AI creative than one afternoon with a photographer.
This tutorial shows the workflow and settings. The prompts are where the real value lives, because a good prompt produces professional-grade output while a generic prompt produces obvious AI content. The difference is in the details: camera specifications, lighting descriptions, material textures, and compositional direction.
If you want the exact prompt frameworks I use for AI product photography and video ads, book a consultation. I will walk you through the full system, customized to your product and target market.
Clear Next Step
A clear next step: the lowest-hanging fruit that will move the needle fastest
Honest Assessment
An honest assessment, because if you do not need me I will say so. This is a strategy session, not a sales pitch
