The gaugan2-renderer is a Python script that automatically uploads an image sequence to http://gaugan.org/gaugan2/ as a segmentation map, generates the output and creates a video from it.

A example video created by Gaugan2 Renderer. On the left the input video on the right the output video.

1. clone this repository and import the Gaugan2Renderer from gaugan2_renderer or use the example in main.py
2. install the dependencies via pip install -r requirements.txt
3. create a folder named input_folder and add there the segmentation maps as PNGs you want to render

from gaugan2_renderer import Gaugan2Renderer # import the gaugan2_renderer class from gaugan2_renderer.py

renderer = Gaugan2Renderer() # create an instance of the gaugan2_renderer class
renderer.run("./input_folder", "./output_folder") # run the renderer
renderer.create_video("./output.mp4") # create a video of the created images

gaugan2_renderer.run(input_folder, output_folder)

• input_folder the folder with the segmentation maps that should be rendered, PNGs required
• output_folder the folder with the rendered images - if it does not exist, it will be created automatically

gaugan2_renderer.create_video(output_path)

• output_path the path to the rendered video

• PNG format - required
• input images with the size of 1024 px x 1024 px
• the exact segmentation map colors
• no anti aliasing (every pixel should have a color value specified in Semgmentation Map Colors)

• bridge: #5e5bc5
• bush: #606e32
• clouds: #696969
• dirt: #6e6e28
• fence: #706419
• flower: #760000
• fog: #77ba1d
• grass: #7bc800
• gravel: #7c32c8
• ground-other: #7d3054
• hill: #7ec864
• house: #7f4502
• mountain: #869664
• mud: #87716f
• pavement: #8b3027
• platform: #8f2a91
• river: #9364c8
• road: #946e28
• rock: #956432
• roof: #9600b1
• sand: #999900
• sea: #9ac6da
• sky: #9ceedd
• snow: #9e9eaa
• stone: #a1a164
• straw: #a2a3eb
• tree: #a8c832
• wall-brick: #aad16a
• wall-stone: #ae2974
• wall-wood: #b0c1c3
• water: #b1c8ff
• wood: #b57b00