issues
search
Boese0601
/
MagicDance
[ICML 2024] MagicPose(also known as MagicDance): Realistic Human Poses and Facial Expressions Retargeting with Identity-aware Diffusion
https://boese0601.github.io/magicdance/
Other
615
stars
52
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
请问对于提高手部生成质量有什么思路吗？
#38
BugsMaker0513
opened
5 days ago
1
模型权重的一些疑问
#37
dengfenglai321
opened
6 days ago
0
Cuda memory
#36
Caettano
opened
1 week ago
0
The quality is not the same as shown in the examples
#35
adityapandey9
opened
2 weeks ago
0
Appearance Control Model during inference
#34
burnquiet
closed
2 weeks ago
0
Questions about Input Latent for Appearance Control Model
#33
CrispyFeSo4
opened
3 weeks ago
4
FaceID in Generated gif does not match the person in the original image
#32
songzr0
opened
1 month ago
0
What are your training resources?
#30
dengfenglai321
closed
1 month ago
0
some questions
#29
BJQ123456
opened
1 month ago
0
数据集下载地址
#28
dengfenglai321
closed
1 month ago
0
motion module
#27
dwanfun
opened
1 month ago
0
the dependencies cannot be installed in the Windows
#26
mikeyimer
opened
1 month ago
1
hdfs 文件下载
#25
dwanfun
opened
1 month ago
1
The verification results are poor, can you give some advice? Thank you
#24
goalinshi
opened
1 month ago
3
How to improve the face cos? Any suggestions? Thanks!
#23
BugsMaker0513
opened
1 month ago
1
How to fine-tune Temporal module(motion module)?
#22
haohantianchen
opened
2 months ago
1
T_T cldm_v15_reference_only_temporal_pose.yaml and cldm_v15_reference_only_temporal.yaml and cldm_v15_reference_only_mask.yaml could not be found.
#21
zhjj666
opened
2 months ago
0
the code of computing PSNR in the Disco repository is wrong
#20
Delicious-Bitter-Melon
opened
3 months ago
2
The more details about computing fid-vid.
#19
nehcoah
closed
3 months ago
2
I've reproduced your code but the results are inconsistent with the effects of the weights you've open-sourced. I'm wondering if your model_state-10000.th was only trained on the TikTok dataset? Also, could I possibly get the optimizer_state_latest.th that corresponds to your first-stage weight model_state-10000.th?Thank you in advance.
#18
Jeff-Fudan
closed
4 months ago
9
I want to train the whole body with dance video data, any suggestions?
#17
wujiafu007
closed
4 months ago
0
During training, Crop trained, so how to reason about the whole body map during reasoning?
#16
wujiafu007
closed
4 months ago
0
I want to infer the results of the first stage, how should I proceed?
#15
Jeff-Fudan
closed
4 months ago
3
It seems that the Dataloader is performing slower than expected. When I train using Moore's open-source AnimateAnyone code, I have noticed that the GPU utilization reaches around 90%. However, when using MagicPose, the GPU utilization is only around 20%. I have a dataset of approximately 3K videos.
#14
Jeff-Fudan
closed
4 months ago
3
I am very interested in the Multi-Source Self-Attention mentioned in the paper. Can you tell me in which script of the open-source code it can be found?
#13
Jeff-Fudan
closed
4 months ago
4
RuntimeError: shape '[13, 77, 8, 40]' is invalid for input of size 788480
#12
Jeff-Fudan
closed
4 months ago
3
The training script you provided is for a single GPU, how can I switch it to 8 GPUs?
#11
Jeff-Fudan
closed
4 months ago
1
How can I train my own video data? Great job, looking forward to your reply!
#10
Jeff-Fudan
closed
4 months ago
1
Update mask.py
#9
eltociear
opened
4 months ago
0
ask for help
#8
zcdliuwei
closed
4 months ago
3
Unexpected key(s) in state_dict: "cond_stage_model.transformer.text_model.embeddings.position_ids".
#7
sooxp
closed
4 months ago
5
any updates on the code release?
#6
james-310110
closed
4 months ago
0
wait for real open source
#5
forechoandlook
closed
4 months ago
0
waiting for the code
#4
1179021477
closed
4 months ago
0
Testing data for TikTok
#3
xianrui-luo
closed
6 months ago
2
how to detail with cross-attension in SD
#2
ShineChen1024
closed
4 months ago
0
Can't wait to try your model
#1
hyfu0583
closed
4 months ago
3