search

emreaksan / deepwriting

Code for `DeepWriting: Making Digital Ink Editable via Deep Generative Modeling` paper
MIT License
101 stars 28 forks source link

use as generative model (text -> svg) #12

Closed benswift closed 2 years ago

benswift commented 5 years ago

I've got the visualise_hw.py script working correctly (as described in the readme). What's the best/easiest way to use the pre-trained model to generate the data (either as a svg, or even better as <key, value> pair stroke data.

Thanks for the help.

Yashwanthv21 commented 5 years ago

@benswift did the code run as it is or did you change something. I am trying to run the code.

benswift commented 5 years ago

I made a few small tweaks to get it to install with pipenv, you can see them in my fork.

After that, it worked as described in the readme, e.g. this will generate "handwriting" svgs for the first 3 phrases in the dictionary (index 0, 1 and 2).

PYTHONPATH=$PYTHONPATH:./source pipenv run python visualize_hw.py -D ./data/deepwriting_validation.npz -O ./data_images -S 0 1 2