Skip to content

Latest commit

 

History

History
77 lines (57 loc) · 3.65 KB

README.md

File metadata and controls

77 lines (57 loc) · 3.65 KB
Raw

#1

furniture-renderer

Recruitment task - create a Django project with endpoint /projection accepting a list of geometries and plane to which they should be projected.

Requirements

  • Docker

Installation

Build the image:

$ docker build --tag svg_renderer:latest .

Run the container:

$ docker run -p 8000:8000 svg_renderer:latest

Assumptions / simplifications

I assumed that geometry described with a pair of two points (x1, y1, z1, x2, y2, z2) has each of its sides always parallel to one of XY, YZ, or XZ plane. With that assumption, I will render SVG on which only one side of given geometry may be visible.

Additional features not required by specification

I was curious how the furniture will look from the back. To check that I decided to implement additional "reverse planes".

This leads me to the conclusion that furniture like this from the example input, when cast to 2D would look exactly the same from both front and back because the same geometries are visible and are almost symmetric. I decided to use depth (the z-axis perpendicular to the given plane) that could be easily calculated and use it to make the closer figures bright, and those which are from the back - darker.

Front (XY plane) Back (-XY plane)
XY_plane XY_rev_plane

When looking at the furniture from the top (XZ plane) - only one figure is visible, and the rest is obscured. Therefore there is no need to render each geometry from the input. I implemented a simple mechanism to remove figures fully obscured by another one.

For described case app produces svg with single object::

<?xml version="1.0" encoding="UTF-8"?>
<svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink"
     width="1680" height="920" viewBox="0 -460 840 460">
<defs>
</defs>
<rect x="70" y="-390" width="700" height="320" fill="#c8c8c8" stroke="black" stroke-width="1" />
</svg>

I am aware of the case when figures could be fully obscured by multiple others, but those would be harder to implement in a hurry. The Idea might be to sort figures and stack them on each other, adding the collective area of them to the condition used to determine whenever further figures are obstructed.

To sum it up, additional features are:

  • removing obscured geometries,
  • shading of the geometry based on its z-index,
  • negated planes allowing to see the furniture from the other side.

Example usage:

curl -XPOST -H "Content-type: application/json" -d '{
  "geometry": [{"x1": -207, "x2": -332, "y1": 9, "y2": 191, "z1": 0, "z2": 18}, {"x1": -207, "x2": -332, "y1": 209, "y2": 391, "z1": 0, "z2": 18}, {"x1": 207, "x2": 332, "y1": 9, "y2": 191, "z1": 0, "z2": 18}, {"x1": 207, "x2": 332, "y1": 209, "y2": 391, "z1": 0, "z2": 18}, {"x1": -8, "x2": 10, "y1": 9, "y2": 191, "z1": 0, "z2": 320}, {"x1": -8, "x2": 10, "y1": 209, "y2": 391, "z1": 0, "z2": 320}, {"x1": -350, "x2": -332, "y1": 9, "y2": 191, "z1": 0, "z2": 320}, {"x1": -350, "x2": -332, "y1": 209, "y2": 391, "z1": 0, "z2": 320}, {"x1": 332, "x2": 350, "y1": 9, "y2": 191, "z1": 0, "z2": 320}, {"x1": 332, "x2": 350, "y1": 209, "y2": 391, "z1": 0, "z2": 320}, {"x1": -350, "x2": 350, "y1": 391, "y2": 409, "z1": 0, "z2": 320}, {"x1": -350, "x2": 350, "y1": 191, "y2": 209, "z1": 0, "z2": 320}, {"x1": -350, "x2": 350, "y1": -9, "y2": 9, "z1": 0, "z2": 320}],
  "projection_plane": "XY"
}' 'http://localhost:8000/projection' > output.svg

Allowed "projection_plane" are: XY, YZ, XZ, -XY, -YZ, -XZ