Introducing Magic Mirror

TLDR

Magic Mirror is now live on Fiddl.art, offering a streamlined AI headshot generator that turns selfies into professional portraits. The guided workflow includes capture, training, look selection, generation, and optional animation. Key features include a large template library, privacy by default, and marketing tools like QR promo codes. It’s built on our custom AI models stack for reliable, high-quality results.

Meet Magic Mirror

Magic Mirror is a guided portrait flow that transforms a quick selfie into styled images using gendered templates and your subject details. It simplifies the entire process—from capture and training to look selection and generation—making it ideal for creating consistent, high-quality AI headshots without prompt engineering. Built on Fiddl.art’s custom AI models, it ensures sharp, personalized results every time. Try it now at app.fiddl.art/magicMirror.

The end-to-end flow

The process is designed for ease and efficiency:

1. Capture: Use the live camera with device switching on desktop and mobile, or upload up to 10 selfies from your gallery. The interface includes a mirrored preview, clear feedback, and graceful permission handling.
2. Train: Add subject gender and an optional description for better fidelity. Clear progress indicators keep you informed during training.
3. Pick looks: Choose exactly three templates from a curated grid, with gender-appropriate previews and a sticky confirm bar for smooth selection.
4. Generate: Batch async creation processes your three looks simultaneously with consistent parameters, including negative prompts to avoid common artifacts.
5. Results: View your images in a mosaic gallery on desktop or a streamlined list on mobile. Options include downloading, sharing, generating more variations, or animating favorites.
6. Animate: Turn still portraits into short videos with estimated credit costs shown upfront. Videos default to private and 9:16 aspect ratio for vertical viewing.

Templates that stay on model

Magic Mirror features a extensive template library designed for portrait consistency and stylistic range. Each template includes camera notes, composition, lighting, mood, and robust negative prompts to ensure high-quality outputs. Categories include everyday and editorial looks (e.g., LinkedIn headshots, streetwear), culture and genre themes (e.g., cyberpunk, wizard), and lifestyle sets (e.g., beach traveler, café creator). All templates have male and female variants for tailored results. This system allows you to create AI art with custom models without constant tweaking.

Under the hood

Technical enhancements power Magic Mirror’s reliability:

• Template APIs: Filterable by tags and kind, with consistent slots and OpenAPI documentation.
• Batch async creation: Single calls schedule multiple images with clear status tracking.
• Training metadata: New fields for subject gender, category, and description enable richer customization.
• Model support: Wider model list for images and video, with LoRA scale adjusted to 0.98 for stronger customizations in Flux-based models.
• Privacy defaults: New images and animations are private unless set to public.
• Shareable links: Short base64url IDs replace long UUIDs for cleaner sharing.
• Edge safety: Centralized error handling and structured logs improve reliability.

Camera and gallery polish

The capture experience is refined with live device switching, stable iOS playback, and a clean file picker for gallery uploads. The media viewer prevents accidental closes and ensures smoother HD loading.

Marketing and ops

Promo codes, QR, and print sheets

For events and promotions:

• QR batch generator: Create hundreds of unique claim codes with CSV/JSON manifests.
• A4 sheet builder: Automatic 3x3 layouts at 300 DPI for print-ready assets.
• Promo claim UX: Streamlined code detection, one-time claims, and auth-aware flows.
• Admin tools: Copy standard or Magic Mirror-embedded claim links easily.

We tested this at Nas Summit Thailand, distributing 100 trial codes to creators.

Growth and comms

• SEO-friendly landing: Fast edge delivery with Open Graph and JSON-LD for Magic Mirror.
• Email reliability: All user emails sync to SendGrid for transactional and marketing sends.
• Live chat: Tawk integration includes user context like points and avatar for better support.
• SEO tuning: Updated robots.txt and sitemap for key pages, plus crawler guidance.

Other improvements

• Custom video uploads: Use your images as video start points with a new gallery picker.
• Veo3 image input: Support for starting images in Veo3 requests with 720p or 1080p output.
• Model discovery: Enhanced filtering on the Models Page and faster loading for single-model pages.
• Profile revamp: Unified profile and account with clearer creation capabilities.
• Pricing clarity: Updated points pricing and cost estimates for animations.

Who is this for

Magic Mirror is ideal for creators seeking fast, consistent portraits without prompt fatigue, teams managing events with promo codes, and anyone exploring text-to-image AI through a guided path. If you use Forge for custom model training, it doubles as a personal face generator for style experiments. For broader creative tools, see our guide on how to create AI art with Fiddl.artRead blog post.

Notes and small fixes

• Stronger error handling in async flows with explicit exit codes.
• Deduplication of identical video queries to prevent duplicate loads.
• Smoother login prompts and redirects in Magic Mirror.
• Typography and layout refinements for galleries and action bars.

Getting started

1. Visit Magic Mirror and capture or upload selfies.
2. Add a subject description for extra guidance (optional).
3. Pick three looks and confirm to start training and generation.
4. Download, share, generate more, or animate your results.

Follow our progress on Trello for the latest updates.

FAQ

How does Magic Mirror ensure privacy?

All generated images and animations are private by default. You can choose to make them public later if desired.

Can I use Magic Mirror for custom model training?

Yes, Magic Mirror integrates with Fiddl.art’s Forge for training custom models, making it ideal for personalized style experiments.

What templates are available?

The library includes everyday, editorial, cultural, genre, and lifestyle templates, all with male and female variants for consistent portraits.

How do promo codes work?

Promo codes can be generated in batches with QR codes and print sheets. Users claim them one-time via a streamlined UX.

Is there a cost for animations?

Yes, animations estimate credit costs based on duration and model pricing before you commit, ensuring transparency.