In recent years, cloud-based photo editors powered by AI have mastered the art of stripping away layers of clothing from user photos — automatically erasing the “fabric” pixels and reconstructing a seamless, photorealistic image of the body. At first glance, it looks like just another beauty filter, but these algorithms deliver results so convincing you’d swear it’s an untouched snapshot.
In this case study, we’ll demystify how this technology works, explore the audiences it hooks, and reveal just how viable it is to launch an ad campaign in this decidedly unconventional niche.
Key Points
Offer: AI Non-Mainstream bot
GEO: US, GB, DE, FR, IT
Traffic: Non‑mainstream High & Medium
Ad Format: Popunder mobile
Ad campaign period: 25 May – 31 May (7 days)
ROI: ~150% (monthly 300%)
What are photo editors focused on non-mainstream transformations?
Recent breakthroughs in artificial intelligence have given rise to so-called “nudification” tools (DeepNude), which you can now easily find via a simple search. These AI non-mainstream bots, mobile apps, and web services can whip up photorealistic non-mainstream images of real people — everything from erasing the pixels that represent clothing to animating the final result.
AI algorithms learn from massive datasets — predominantly featuring women — because they tend to perform best on the female form. Experts estimate that about 99% of these “unofficial” creations online involve female subjects.
Today, generative-AI technology is more accessible than ever, fueling the rapid spread of these services. No technical know-how or Photoshop skills are required: just upload a photo, and the system automatically generates an artificial non-mainstream image. Unlike traditional retouching, AI tools deliver unparalleled speed, realism, and ease of use — all thanks to powerful neural networks.
Modern AI non-mainstream services — whether standalone websites or messenger bots — rely on deep learning and cutting-edge neural architectures. They’re trained on massive photo datasets to distinguish clothing, skin, shadows, and body contours. Once trained, the pipeline unfolds in three stages:
Image Analysis
The model segments the uploaded photo, pinpointing garments, background, and the subject’s outline.
Fabric Removal
Leveraging inpainting techniques or GAN-based networks, the system “paints over” areas where clothing once was — seamlessly blending in realistic skin textures and fine-tuning lighting.
Final Render
The neural network synthesizes a fully photorealistic, clothing-free image and hands it back to the user.
Output quality varies with the breadth and diversity of the training set — some services deliver jaw-dropping realism, while others reveal telltale artifacts.
Start advertising your offers with HilltopAds!
Types of AI Non-mainstream Services
Telegram Bots
Lightweight apps living inside the Telegram ecosystem. You send a photo in chat, and the bot replies with your transformed image. Thanks to “related channels” and in-app recommendations, some bots attract millions of monthly users — though actual usage stats fluctuate wildly: casual testers versus power users running hundreds of edits.
Web Applications (Cloud-SaaS)
Upload your photo to a website; the heavy lifting happens on cloud servers. Within seconds you get a blurred demo, and after payment you download the full-resolution, clothing-free version. Processing speed depends on server capacity and current load.
Mobile Apps
Native Android and iOS apps that split computation between on-device (edge-AI) and cloud APIs. This hybrid approach lets you preview results instantly — even offline — though high-end hardware is a plus.
Browser Extensions
Plugins for Chrome or Firefox that “snatch” any image on a page. A right-click sends it off to the cloud for processing, returning your new image without ever leaving your browser.
Partner API Services
RESTful endpoints designed for integration into third-party platforms — from niche non-mainstream sites to custom chatbots. Partners pay per request or via subscription, and build their own frontend around the undress API.
Each format comes with its own business model (free trials, subscriptions, one-time fees), privacy guarantees, and speed profile. When selecting an offer or partner tool, consider the interface (messenger bot vs. web portal), response time (edge-AI vs. cloud), and pricing structure (single API keys vs. enterprise-scale packages).
Subscription & Monetization Models in AI Non-mainstream Services
All AI non-mainstream services and bots operate as commercial ventures: a limited demo is free (usually delivering a heavily blurred or watermarked image), but you pay for the full, unfiltered result. Three main payment schemes dominate the market:
Tiered Subscription (Feature-Scaling)
Users choose from multiple subscription tiers — higher plans grant better resolution, faster processing, and priority in the generation queue. Eight out of twenty major platforms use this model; the average cost per image sits around $0.30–$0.34, with top-tier subscribers enjoying even steeper per-image discounts.
Credit Packs (Pro-Credit)
Here, users buy bundles of “credits” that are consumed as they generate images. Seven out of twenty sites favor this approach, pricing each credit at roughly $0.30.
Flat-Rate Access (All-for-One)
A handful of platforms offer unlimited access under a single monthly fee—on average $16/month, which breaks down to about $0.35 per generation.
On top of these, many services provide enterprise/API plans for integrators: white-label solutions or direct API access, typically priced between $20 and $299 per month. As for payment methods, 17 out of 20 platforms accept cryptocurrency (five of them exclusively), alongside traditional options like credit cards and PayPal.
In the US and across Europe, users tend to lean on global payment channels — but whether it’s buying token packs or signing up for a monthly plan, the underlying models remain the same worldwide.
Ultimately, this diverse subscription landscape ensures there’s a plan for every kind of user — from the casual tester to the power user who can’t get enough of AI-powered photo transformations.
Demand and Popularity by Region (US, GB, DE, FR, IT)
Global interest in AI non-mainstream services — both Telegram bots and web platforms — is staggering: according to WIRED, the top 50 bots draw about 4 million unique users per month, while 16 leading AI non-mainstream sites collectively amassed 200 million visits in the first half of 2024. Individual platforms routinely see tens of millions of visits every month.
Regionally, the picture is much the same: roughly 45% of traffic originates from the United States, but English-speaking audiences in the GB and users in Germany, France, and Italy also show strong engagement. Most rely on the web versions and Telegram bots, since official mobile apps rarely make it into the App Store or Google Play.
Niche ad networks — think HilltopAds — dominate the promotion of these non-mainstream offers, yet pricing, features, and monetization models remain virtually identical across borders. The only real difference from one country to the next? Scale of demand — everyone’s using the same services, just in varying amounts.
Start working with HilltopAds!
There is high-quality traffic and a wide variety of GEO
Motivation & Market Demand – Target Audience
When it comes to who’s using AI non-mainstream bots, it’s overwhelmingly young men — after all, these tools primarily process images of women. Among teens and young adults (roughly 15 – 25), a full 69.4% of guys who’ve heard of these apps have actually tried them, compared to 45.4% of young women.
Broadly speaking, the typical user is a tech-savvy 18 – 35 year-old with a taste for cutting-edge IT tools and non-mainstream content. They live online — constantly scrolling through social feeds, chatting on Telegram and Discord, and exploring niche corners of the web — and they can’t resist trying out the newest AI-powered tricks.
Now let’s dive into the campaign setup on HilltopAds. Our advertiser from Bulgaria has been working with HilltopAds since October 2024, and we’ll walk you through how he crafted his ad and fine-tuned it over the week-long run for peak performance.
General settings for an advertising campaign on the HilltopAds platform
Before you can start your advertising campaign on HilltopAds, you’ll need to sign up as an advertiser. You can register through this link.
Once you’re all set up, here’s how to create your campaign:
- Head over to the Manage Campaigns section.
- Hit the Add Campaign button.
- In the campaign creation area, choose the Popunder mobile ad format.
- In the Traffic Channels section, pick Non-Mainstream High and Medium.

To analyze ad campaign performance, we recommend you to set up Postback. To learn more, please check out our Full guide:
Next, we set the necessary targeting settings:
- GEO – US, GB, DE, FR, IT
- Device type – mobile/smartphone and tablet

You can also configure campaign filters and allow/disallow Proxy and WebView traffic from the campaign. In this case, we enabled the WebView filter while disabling Proxy traffic. To learn more about WebView and Proxy traffic, please check out this guide.

You can set both daily and total budget limits, but keep in mind the daily limit shouldn’t fall below $20. During testing, we used a daily budget of $300 for the campaign.

The last step before going live is setting your CPM — in our case, $3.12 — to balance reach and budget. Although the campaign has been running since November 2024 and continues today, we focused on a seven-day window from May 25 to May 31, 2025, during which this rate yielded 248 conversions, giving us a clear, consistent dataset to measure campaign effectiveness.
Advice from HilltopAds: As a starting point to test a new offer, use the suggested CPM. This will indicate if the offer is performing and if the campaign should be optimized.

Optimization & Tracking
To filter out unprofitable ad zones from your campaign, you can apply the Auto Optimization feature. This will automatically detect and exclude low-performing traffic sources, and include the non-profitable zones in your BlackList to make your performance more efficient.
We switched on automatic optimization from day one, using a 48-hour evaluation window with thresholds set at 1,500 impressions and at least 1 conversion. This longer timeframe ensures we gather a more representative dataset for each traffic source, while the minimum thresholds prevent any channel from being dropped over random noise. As a result, we quickly cut out underperforming placements and focused our budget on audience segments that consistently drove at least one conversion within 2 days.

To learn more about the capabilities of automatic optimization with HilltopAds, you can check out the article on the website.
Results
We just wrapped up a one‑week push on HilltopAds (mobile pop‑under) across the US, UK, Germany, France, and Italy—and the numbers speak for themselves. From May 25 – 31, we served roughly 673,078 impressions at a $3.12 CPM, driving 248 conversions on a $2,100 spend — an eCPA of just $8.4677. In other words, every $3.12 invested yielded a conversion — an impressively efficient result for such an non-mainstream vertical.
Official data from our advertiser confirms a weekly ROI in the 120–150% range, yielding a total profit of approximately $2,646 during our 25–31 May test. And thanks to the subscription‑based model and automatic renewals, that ROI compounds: over a full month, returns can climb as high as 300%.
The advertiser declined to reveal exact revenue figures, but did share that this performance aligns with — or even outpaces — other successful HilltopAds campaigns. For reference, a recent utilities Popunder saw a 150% ROI, while a non‑mainstream dating offer hit 183%. Clearly, well‑optimized Popunder traffic can be incredibly profitable. In short, our AI non‑mainstream bot test didn’t just break even — it generated serious profit, both week‑to‑week and as subscriptions roll on.
Key Campaign Metrics:
- Impressions: 673,078 total (≈100,000 per day)
- Conversions: 990
- eCPA: $2.12
- Total Spend: $2,100
- Total Profit: $2,646
- Weekly ROI: 126 % (confirmed by advertiser)
Conclusion
This campaign was a clear win. Even in a non-mainstream niche, we achieved strong efficiency and profitability. The key takeaway is that targeted Popunder mobile traffic can convert well when the offer and audience align. With an eCPA of $2.12 and a confirmed 126 % ROI, we more than recouped our investment. The advertiser was pleased, and these numbers give us confidence to scale up. In practical terms, it means we can roll out the AI non-mainstream bot offer to other geos with similar audience profiles (think other Tier-1 countries with comparable user behavior) and expect similarly solid performance. In sum, this unconventional experiment paid off and sets the stage for expanding into new markets with a proven winning formula.
Advertise your offer with the best ad network for advertisers HilltopAds!
- Advanced targeting options
- Direct traffic sources
- Self-serve platform
- Fully-managed service
- Postback tracking