Step-by-Step Guide

How UPretty Works

Transform your photos into stunning AI-generated artwork in just a few simple steps. No design skills required — our AI handles everything for you.

STEP 1

Browse & Choose a Theme

Explore our curated collection of 100+ professionally designed themes and styles. From elegant portraits to creative fantasy scenes, cinematic movie posters to artistic illustrations — find the perfect theme that matches your vision.

  • Filter themes by category (Portrait, Fantasy, Cinematic, Artistic, etc.)
  • Filter by gender suitability (Male, Female, Unisex)
  • Preview sample outputs to see what each theme produces
  • New themes are added regularly by our creative team
STEP 2

Upload Your Photo

Upload a clear photo of yourself or the subject you want to transform. Our AI works best with well-lit, front-facing photos where the face is clearly visible. You can use any photo from your phone, camera, or existing collection.

  • Supported formats: JPG, PNG, WebP
  • Best results with clear, well-lit face photos
  • Front-facing photos produce the most accurate results
  • Your uploaded photo is processed in real-time and never stored permanently
STEP 3

AI Generates Your Image

Once you select a theme and upload your photo, our AI powered by Google's Gemini API goes to work. The AI analyzes your photo, understands the facial features and expressions, and intelligently merges your likeness into the chosen theme to create a completely new, unique image.

  • Processing takes approximately 5-10 seconds
  • AI preserves your facial features and expressions
  • Each generation counts toward your free daily limit (10 per day)
  • Results are unique — no two generations are exactly alike
STEP 4

Download & Share

Your generated image is ready! Preview the result in full quality, download it to your device, and share it on your social media profiles, use it as a profile picture, or send it to friends and family. All generated images are available in your gallery for up to 30 days.

  • Download in high-quality PNG format
  • View all your generations in the Gallery tab
  • Images are stored for 30 days for re-downloading
  • Share directly to social media platforms

The Technology Behind UPretty

UPretty leverages Google's state-of-the-art Gemini API for image generation. Gemini is one of the most advanced multimodal AI models available, capable of understanding both text and images with remarkable accuracy. But the model alone is only part of the story — what you actually experience as "a generation" is the result of several careful engineering layers working together.

Multimodal Understanding

Earlier generations of image AI could only understand text prompts — you described what you wanted in words, and the model imagined the rest. Gemini is different. It is a multimodal model, meaning it can reason about your uploaded photo and a written theme description simultaneously. That is why UPretty can preserve the actual you — your face shape, your skin tone, your expression — instead of generating a generic person who happens to share your prompt.

Prompt Engineering, Done For You

The single biggest differentiator between mediocre and stunning AI output is the prompt. A short prompt like "make me a movie poster" produces something cliché. A 400-word prompt that specifies lighting direction, lens choice, mood, colour grading, era, composition rules, and negative constraints produces something cinematic. Every UPretty theme is backed by a hand-crafted prompt that we have iterated on dozens of times. You never see it — that work happens entirely on our backend, so you only ever see polished results.

Facial Preservation Pipeline

One of the hardest problems in stylised image generation is maintaining likeness. The model needs to apply heavy artistic transformation while keeping the subject recognisable. We tackle this with a combination of careful reference-image weighting, theme-specific guidance, and post-generation quality checks. If a generation drifts too far from the input photo, our system flags it so we can refine the underlying prompt.

Theme Curation

Anyone can plug into the Gemini API. What makes UPretty different is the library of themes — each one developed and tested by our team across many input photos to ensure consistent results regardless of who uploads. We add new themes weekly, often inspired by user requests, seasonal events, and cultural moments. Themes that consistently underperform are removed; the ones that delight users are kept and refined.

Infrastructure

UPretty runs on modern cloud infrastructure with auto-scaling, so generation times stay snappy whether you are generating one image at 3 AM or thousands of users are generating simultaneously during a viral moment. Authentication is handled by Auth0 (enterprise-grade identity), and image delivery via a global CDN so downloads are quick from anywhere in the world.

Behind the Scenes: 10 Seconds, Step by Step

Ever wondered what is actually happening between clicking "Generate" and seeing your image? Here is the timeline.

  1. 0.0s

    Upload received and validated

    Your photo is checked for format, size, and quality. Files are kept in encrypted memory only — never written to disk.

  2. 0.3s

    Theme prompt assembled

    The hand-crafted theme prompt is fetched and combined with safety constraints. This is the secret sauce that makes each theme look distinct.

  3. 0.5s

    Request sent to Gemini

    Your photo and the assembled prompt are sent securely to the Gemini API. Network latency depends on your region but is usually well under a second.

  4. 1s – 8s

    AI generation in progress

    Gemini processes the multimodal input and generates the new image. This is the heavy-compute step and the bulk of the wait time.

  5. 8.5s

    Quality and safety checks

    The generated image is automatically scanned to confirm it meets safety standards and matches the theme. Failed generations do not count toward your daily limit.

  6. 9s

    Image saved to your gallery

    The result is uploaded to our CDN and added to your private gallery. From here it is yours to download, share, or discard.

  7. 10s

    Image delivered to your screen

    The full-resolution image streams down to your browser. You see the result and can download it instantly.

Tips for Best Results

Use High-Quality Photos

The better your input photo quality, the better the AI output. Use well-lit photos with good resolution for the best transformations.

Clear Face Visibility

Make sure your face is clearly visible and not obscured by sunglasses, masks, or heavy shadows. Front-facing photos work best.

Simple Backgrounds

Photos with simple, uncluttered backgrounds allow the AI to focus on your face and produce cleaner results.

Experiment with Themes

Don't be afraid to try different themes! Sometimes unexpected combinations produce the most creative and impressive results.

Ready to Transform Your Photos?

Join thousands of users who are already creating stunning AI-generated images with UPretty. Start creating today — 10 free generations every day, no credit card required.

Start Generating