About PerfectVector
We're two former research scientists building a tool that turns images into editable vectors.
Our background is in deep learning and computer vision. Our focus is the work you can make with it.
Meet the two of usThe people
We build PerfectVector's engine and interface ourselves, using our research and product experience to refine the vectors you scale and edit.
Brian brings his experience as a product lead at a NYSE-listed tech company and a tech unicorn to PerfectVector's engine and user experience. His background includes deep learning research published at top-tier international conferences and contributions to image analysis patents.
Jay builds PerfectVector's models and software with experience as a researcher at a tech unicorn and a software engineer at a global company. He has authored several top-tier international conference papers.
The research
We develop the core models and algorithms that turn images into editable shapes.
Develop
We combine deep learning and classical computer vision in the core vectorization models and algorithms we develop in-house.
Examine
We review how the engine reconstructs shapes, edges, colors, and fine features, and use those results to guide further work.
Refine
Improving the conversion engine is an ongoing part of building PerfectVector. The product and the research develop together.
The work
A vector needs to hold together when you scale it up and start editing. The way shapes meet, the edges of a letter, and the smaller details all matter.
A closer look
Compare a lower-resolution PNG preview with the SVG, then inspect its edges and anchors.
The colors in this SVG
Results depend on the source image. You can vectorize and inspect protected previews for free, with no usage limit, before deciding to unlock a download.
Try your own imageGet in touch
Questions, feedback, or a result that needs a closer look? We'd like to hear about it.
