alexandre01 / deepsvg

[NeurIPS 2020] Official code for the paper "DeepSVG: A Hierarchical Generative Network for Vector Graphics Animation". Includes a PyTorch library for deep learning with SVG data.
https://www.reshot.ai
MIT License
981 stars 99 forks source link

Error is returned when SVG is processed #46

Open paczyg opened 1 month ago

paczyg commented 1 month ago

Dear All,

I have been trying to use the pretrained DeepSVG, right now to simply have the original SVG returned. For a simple SVG, it works well.

However, when I try to use a simple SVG returned by potrace, I receive an error. The error: ` 195 res[arg] = torch.stack([t.get_relative_args() for t in tlist]) 196 if arg == "args": --> 197 res[arg] = torch.stack([t.args() for t in t_list]) 199 if "filling" in model_args: 200 res["filling"] = torch.stack([torch.tensor(t.filling) for t in t_sep]).unsqueeze(-1)

RuntimeError: stack expects each tensor to be equal size, but got [72, 11] at entry 0 and [67, 11] at entry 1`

Here is the code of the SVG (with bad xml tags to have it showed well here:

<.?xml version="1.0" standalone="no"?> <.!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 20010904//EN" "http://www.w3.org/TR/2001/REC-SVG-20010904/DTD/svg10.dtd"> <.svg version="1.0" xmlns="http://www.w3.org/2000/svg" width="256.000000pt" height="256.000000pt" viewBox="0 0 256.000000 256.000000" preserveAspectRatio="xMidYMid meet"> <.metadata> Created by potrace 1.16, written by Peter Selinger 2001-2019 <./metadata> <.g transform="translate(0.000000,256.000000) scale(0.100000,-0.100000)" fill="#000000" stroke="none"> <.path d="M995 2228 c-3 -7 -4 -440 -3 -963 l3 -950 313 -3 312 -2 0 965 0 965 -310 0 c-239 0 -312 -3 -315 -12z m613 -955 l-3 -948 -297 -3 -298 -2 0 950 0 950 300 0 300 0 -2 -947z"/> <./g> <./svg>

Can you help me with this issue?