
MagicTrick
The product behind Nimalthas Imaging: photo and video grading that runs in the browser, built to ask nothing of the user but the picture.
- TypeScript
- Next.js
- TensorFlow.js
- Supabase
- Stripe
Software · Computer science
C++ and Python pointed at image pipelines — super-resolution, artifact control, hardware acceleration — plus real-time AR and the web systems behind Nimalthas Imaging.
aadavn@nimalthas.com01 / Principal work

The product behind Nimalthas Imaging: photo and video grading that runs in the browser, built to ask nothing of the user but the picture.
A six-engine offline photo pipeline: auto-sorts, rebuilds structural defects, upscales with hybrid classical and AI methods, simulates depth of field, corrects artifacts, and compresses for delivery.

Real-time 3D hand skeleton reconstruction on-device, fusing Vision hand pose with ARKit depth through a modular RealityKit renderer.
02 / Research and tooling
A non-learning single-image SR pipeline posed as an inverse problem: progressive upscaling, edge-guided refinement and iterative back-projection.
A stable, explainable CLI upscaler built around artifact control — anti-ringing, edge protection, adaptive sharpening — rather than hallucinated detail.
03 / Technical
CMPT 125 and CMPT 225 at SFU, focused on OOP, recursion and complexity-aware implementation; then HCAISSR.
Introduced in CMPT 120, then expanded through self-directed tooling and research prototypes.
Self-taught through Swift Playgrounds, deepened via iOS and AR work with ARKit, RealityKit and Vision.
CMPT 225: linked structures, trees, sorting and searching, ADT design.
CMPT 225, analysing time and space complexity and validating tradeoffs under real constraints.
C and C++ implementation work, with emphasis on correctness, ownership and runtime behaviour.
04 / Background
Building local-first intelligent imaging tools. Shipped MagicTrick end to end — the grading model, the web application, subscription billing and accounts.
Photography, hybrid video and full live-event AV production for community events, weddings and broadcasts.
Currently focused on image sampling and upscaling research, in the territory staked out by AMD FSR and NVIDIA DLSS.
05 / Archive
06 / Next
If you are working on imaging, performance-sensitive systems or anything adjacent, I would like to hear about it. There is also an imaging and film side to what I do, and writing at Nimalthas Insights.