akko/diffvg
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Merge pull request #49 from daniel347x/background_image

Browse Source 
Resolve crash in 'backward' when a background image with only 3 channels is passed to 'forward'
This commit is contained in:
Tzu-Mao Li
2023-01-09 12:17:05 -05:00
committed by GitHub
parent 63db510588 af48db06cd
commit adb04d04ee
1 changed files with 1 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
pydiffvg/render_pytorch.py
View File

@@ -384,9 +384,7 @@ class RenderFunction(torch.autograd.Function):
if background_image is not None: if background_image is not None:
background_image = background_image.to(pydiffvg.get_device()) background_image = background_image.to(pydiffvg.get_device())
if background_image.shape[2] == 3: if background_image.shape[2] == 3:
background_image = torch.cat((\ raise NotImplementedError('Background image must have 4 channels, not 3. Add a fourth channel with all ones via torch.ones().')
background_image, torch.ones(background_image.shape[0], background_image.shape[1], 1,
device = background_image.device)), dim = 2)
background_image = background_image.contiguous() background_image = background_image.contiguous()
assert(background_image.shape[0] == rendered_image.shape[0]) assert(background_image.shape[0] == rendered_image.shape[0])
assert(background_image.shape[1] == rendered_image.shape[1]) assert(background_image.shape[1] == rendered_image.shape[1])