iglict / deepfacevideoediting Goto Github PK

There is code for the release of the pytorch version. If so, when will it be released? I'm looking forward to it

I have tried several kinds of sketch, but I failed to make the person close his/her mouth.
Have you ever succeed in making it? If so, would you please share the sketch with me? If not, I wonder does the model have such ability?
Thanks a lot.

NVIDIA GPU RTX 4070 + CUDA 11.1 + cudnn 8.0.4+torch1.9.0
run python run_pti_stylegan3.py but failed:
Value 'compute_89' is not defined for option 'gpu-architecture'

ffhq_weights_stylegan3.pkl == jt_stylegan3_ffhq_weights_t.pkl ?

load this model error, RuntimeError: Ran out of input

i try to run the command(python run_editing.py) but it always can not generate the mask_warp picture successfully.

warp mask: 0/1
Traceback (most recent call last):
File "run_editing.py", line 188, in
predictions = make_mask_animation(source_image, mask_image, driving_video, generator, kp_detector, relative=False, adapt_movement_scale=False)
File "/home/xdy/code/DeepFaceVideoEditing-main/warp_mask.py", line 64, in make_mask_animation
mask = jt.array(mask_image[np.newaxis].astype(np.float32)).permute(0, 3, 1, 2)
File "/home/xdy/anaconda3/envs/jittor/lib/python3.8/site-packages/jittor/init.py", line 569, in transpose
return origin_transpose(x, dim)
RuntimeError: Wrong inputs arguments, Please refer to examples(help(jt.ops.transpose)).

Types of your inputs are:
self = module,
args = (Var, tuple, ),

The function declarations are:
VarHolder* transpose(VarHolder* x, NanoVector axes=NanoVector())

Failed reason:[f 0202 14:39:38.463771 60 transpose_op.cc:62] Check failed axes.size()(4) == xdim(3)