How Merlin Generates Ad Creatives From a Single Product Photo
Most AI marketing platforms claim they "generate ad creatives." When you look closer, that usually means they wrap an API call in a UI, charge you $5 for something that costs $0.03, and never tell you which model actually produced the image. Merlin does it differently: you see exactly which models run, you pay the provider directly at their price, and the output goes straight from generation to your ad account — no manual upload, no middleman markup.
This is a full breakdown of how Merlin's creative pipeline works, from a single product photo to a running ad. No black boxes.
The Pipeline: Product Photo to Running Ad
Merlin's creative generation follows a structured five-stage pipeline. Each stage uses a named, specific AI model — not a mystery "proprietary AI" that's actually someone else's API with a 100x markup.
Stage 1: Product Extraction (birefnet via fal.ai)
You upload a product photo — any photo, even shot on your phone against a kitchen table. Merlin sends it to fal.ai's birefnet model for background removal. birefnet is a state-of-the-art bilateral reference network specifically designed for high-resolution image segmentation. It doesn't just remove the background — it preserves fine details like fabric texture, transparent elements, and hair-thin edges that cheaper removal tools destroy.
Cost to you: roughly $0.01–0.03 per image, paid directly to fal.ai. No Merlin markup.
Stage 2: Ad Brief Construction
This is where Merlin's intelligence layer kicks in. Rather than sending a vague prompt like "make an ad for sweatpants," Merlin builds a structured AdBrief — a detailed creative document that specifies scene composition, lighting direction, color palette, model positioning, brand style, and emotional tone. The AdBrief is constructed from your brand guidelines, product attributes, target audience, and the specific platform the ad will run on (a Meta feed ad has different composition requirements than a TikTok full-screen video).
The AdBrief system is what separates Merlin from tools that just wrap a text-to-image API. A raw prompt produces random results. A structured brief produces ads that look like they came from a creative agency — because the brief encodes the same decisions a creative director would make.
Stage 3: Image Generation (Flux via fal.ai)
Merlin sends the AdBrief to fal.ai's Flux model for image generation. Flux is an open-weights model that produces photorealistic scenes at high resolution. For standard ad creatives, Merlin uses Flux to generate the scene — the environment, lighting, context, and any lifestyle elements — while keeping your actual product photo intact.
Cost to you: roughly $0.03–0.05 per generation. Again, paid directly to fal.ai at their published rate.
Stage 4: Composite Mode (Real Product + AI Scene)
This is Merlin's most important creative feature. Instead of asking the AI to reimagine your product (which always introduces inaccuracies — wrong stitching, different logo placement, altered proportions), Merlin composites your real product cutout onto the AI-generated scene. The product in the final ad is your actual product. The scene around it is AI-generated to maximize appeal for your target audience.
Why this matters: pure AI-generated product images fail in advertising because shoppers notice when the product doesn't match what arrives. Composite mode gives you 100% product accuracy with unlimited creative variety. You can generate 50 different lifestyle scenes for the same product in 10 minutes — beach, urban, cozy home, gym, studio — and the product looks identical in every one because it is the same photo.
Stage 5: Platform Upload and Ad Creation
The finished creative goes directly to your ad platform. For Meta, Merlin uploads the image to your ad account, creates the ad with your copy and targeting, and attaches it to the right campaign — testing, scaling, or retargeting. No downloading, no re-uploading, no manual steps.
Direct-to-Provider Pricing: The $0 Markup Model
Here's what a typical AI ad creative costs through different tools:
| Platform | Cost Per Image | What You're Actually Paying For |
|---|---|---|
| AdCreative.ai | $0.50–2.00 | Their markup on an API call |
| Synter | $1.00–5.00 (credits) | Credit system obscuring real cost |
| Jasper | $0.80–1.50 | Subscription + per-image fees |
| Merlin | $0.03–0.08 | Actual fal.ai provider cost |
Merlin uses your fal.ai API key. You sign up at fal.ai, get $10 in free credits (enough for ~200 ad creatives), and Merlin calls the API on your behalf. You see every charge on your fal.ai dashboard. There is no Merlin fee on top — the creative generation is included in Merlin's core functionality.
This isn't a loss leader or a temporary promotion. It's the architecture. Merlin is the orchestration layer — it decides what to generate, how to brief it, and where to deploy it. The generation itself happens at the provider, at the provider's price. We don't resell API calls. We orchestrate them.
Why Most "AI Creative Tools" Don't Show You What's Under the Hood
Because the margin is the product. When a platform charges $2 per image and the underlying API call costs $0.04, the 50x markup is the business model. Revealing the model name would let you call it directly. Revealing the cost would make the markup obvious. So they say "proprietary AI" and hope you don't look.
Merlin takes the opposite approach: we name every model, we show you the provider, and we encourage you to understand what you're paying for. Our business model isn't API markup — it's being the best AI marketing brain that connects to your existing tools and accounts.
Creative Quality: Why the AdBrief System Outperforms Generic Prompting
The difference between a good AI ad creative and a bad one is almost never the model — it's the prompt. Two marketers using the same model will get wildly different results based on how they describe what they want. Merlin's AdBrief system encodes creative direction expertise into every generation:
- Platform-specific composition — Meta feed ads are square with product centered; Stories are vertical with product in the lower third; TikTok is full-bleed with bold scene context
- Brand consistency — color palette, typography style, and visual tone carry across every generated creative
- Audience-aware scene selection — a streetwear brand targeting 18–24 gets urban night scenes; a wellness brand targeting 35–50 gets clean, bright, aspirational spaces
- Conversion-optimized layout — product visibility, negative space for text overlay, and visual hierarchy follow patterns proven to drive clicks
This is why Merlin's composite mode produces creatives that perform on par with — and often outperform — professional product photography in A/B tests. The product is real. The scene is optimized by AI that knows what converts for your specific audience on your specific platform.
How to Generate Your First Ad Creative in Merlin
- Upload a product photo (any quality — phone photos work fine)
- Tell Merlin your target audience and platform ("Meta feed ad for 25–34 women interested in fitness")
- Merlin builds the AdBrief, generates the creative, and shows you the result
- Approve it — Merlin uploads to your ad account and creates the ad
- Total time: under 2 minutes. Total cost: under $0.10.
No Canva. No Photoshop. No waiting 3 days for a freelancer. No downloading and re-uploading. Product photo to running ad in one conversation.
Try it at merlingotme.com — bring any product photo and see the output in under 60 seconds.
FAQ
What if I don't have a fal.ai account?
Merlin walks you through setup in under 2 minutes. You visit fal.ai, create a free account, generate an API key, and paste it into Merlin. New accounts get $10 in free credits — enough for approximately 200 ad creatives. The entire process happens inside Merlin's onboarding flow.
Can I use my own images instead of AI-generated scenes?
Yes. If you have professional product photography you love, Merlin can use those images directly — no generation needed. It will still handle cropping, formatting for each platform, uploading, and ad creation. The AI generation is there when you need unlimited creative variety at near-zero cost.
How does composite mode handle products with transparent or reflective surfaces?
birefnet excels at complex edges and transparency — it's specifically trained on high-resolution segmentation tasks. Glass bottles, translucent fabrics, and reflective surfaces are handled cleanly. For very complex products (like sunglasses with transparent lenses), Merlin generates multiple cutout options and lets you choose the best one before compositing.
What happens if I don't like the generated creative?
Tell Merlin what to change — "make the background warmer," "more dramatic lighting," "add more negative space on the left for text." Merlin adjusts the AdBrief and regenerates. Each iteration costs $0.03–0.05. You can generate 20 variations for less than $1. Compare that to asking a designer for revisions.
Ready to put your marketing on autopilot?
Try Merlin Free →