Browse categories
Explore
Fiverr Pro
English
$
USD
I will train a custom cartoon filter ai model for your app
Arif
About this gig
Hi I train custom cartoon-filter AI models for apps.
If you want your users to tap a button and see themselves as a cartoon, anime character, or any illustrated style, you have two options:
- Call a cloud API every time. Slow, expensive per image, needs internet.
- Ship a trained model inside your app. Fast, offline, one flat cost.
This gig delivers option 2. You send me 1020 reference images in the style you want (your own artwork, a mood board, screenshots anything consistent). I train a dedicated image-to-image model on that exact style using a FLUX-based pipeline that generates thousands of training pairs matched to your look.
You get back a packaged model file that runs on CPU, Android, iOS or web no Stable Diffusion install, no FLUX, no GPU required on your users' side. Model sizes: 8 MB (anime-style tiny) to 700 MB (high-detail portrait).
I've shipped similar models into a photo-kiosk product that runs across theme parks in Southeast Asia, so the pipeline is battle-tested not a notebook demo.
Message me first with your reference images and target device (iOS / Android / web / Pi / etc.) so I can confirm the fit before you order.
Get to know Arif
Arif
Senior AI and Software Engineer
- FromBangladesh
- Member sinceApr 2026
- Avg. response time1 hour
Languages
English
Hey, I'm Arif. After years deep inside the AI and LLM space, I went independent to work with people who actually need things built, not just talked about in meetings.
I work with Python, large language models, and cloud infra daily. Fine-tuning models for production, designing AI pipelines that scale, or building full systems end to end — that's my thing.
Most of what I do lives between AI engineering and real-world software. I don't just play with models, I ship products. If your project needs someone who knows how these systems work under the hood, let's talk.
My Portfolio
Other AI Development Services I Offer
FAQ
What exactly do I get at the end?
A packaged AI model file (ONNX by default, plus CoreML / TFLite if you ordered the Standard or Premium tier) and a usage guide. You load the file inside your app and feed it a photo; it returns a cartoonified version. No cloud calls, no runtime dependencies on Stable Diffusion or FLUX.
How many reference images do you need from me?
0–20 is the sweet spot. They should share one consistent look (same artist, same filter, same aesthetic). I can also work from a verbal brief + 2–3 anchor images if you don't have a full set; message me first.
Can it do full-body shots, or portraits only?
Starter is portrait-only (face fills the frame). App-Ready and Production tiers include full-body training with a face-focused refinement pass, so group shots and full-body photos work too.
What if the output doesn't look right?
Each tier includes revisions (1 / 2 / 3). I'll re-train or adjust the style and redeliver. If the reference images turn out to be too inconsistent to train on, I refund in full and tell you what would work.
