search

AliJalalifar / Character_Animation

A Re-implementation of the paper "A Deep Learning Framework for Character Motion Synthesis and Editing"
113 stars 33 forks source link

About training data #2

Open HantingPan opened 5 years ago

HantingPan commented 5 years ago

Hi,AliJalalifar I want to try to retrain the data, so I want to ask about '/data_cmu2.npz'. Was it processed by Holden's code in “A deep learning framework for character motion synthesis and editing”? Does the training data contain only CMU?

karanahujax commented 5 years ago

http://theorangeduck.com/page/deep-learning-framework-character-motion-synthesis-and-editing

you can download that data from the above link. Look into the external folder there.

eastchun commented 4 years ago

I downloaded the data from the above link. But, I couldn't find the 'data_cmu2.npz' in any folders including 'external'. Meanwhile, there is 'data_cmu.npz' (not data_cmu2.npz) in the folder named 'processed'.

Is this the one we can use for training?

FNRocha commented 4 years ago

I downloaded the data from the above link. But, I couldn't find the 'data_cmu2.npz' in any folders including 'external'. Meanwhile, there is 'data_cmu.npz' (not data_cmu2.npz) in the folder named 'processed'.

Is this the one we can use for training?

I am having the same problem, if you use the file "data_cmu.npz" the input shape will be (240,73) instead of (240,70) as seen in the autoencoder.py script. I really would like to know what should be done to replicate the results, what should I do with these extra 3 degrees of freedom?

Jeffery-GH commented 1 year ago

I downloaded the data from the above link. But, I couldn't find the 'data_cmu2.npz' in any folders including 'external'. Meanwhile, there is 'data_cmu.npz' (not data_cmu2.npz) in the folder named 'processed'. Is this the one we can use for training?

I am having the same problem, if you use the file "data_cmu.npz" the input shape will be (240,73) instead of (240,70) as seen in the autoencoder.py script. I really would like to know what should be done to replicate the results, what should I do with these extra 3 degrees of freedom?

Now，Could you tell me How do you to solve this question in last? Please.

Jeffery-GH commented 1 year ago

http://theorangeduck.com/page/deep-learning-framework-character-motion-synthesis-and-editing

您可以从上面的链接下载该数据。查看那里的外部文件夹。

Hi，Dear [karanahujax]： I really want to know this answer of the question: if you use the file "data_cmu.npz" the input shape will be (240,73) but autoencoder.py script need (240,70) what should I do with these extra 3 degrees of freedom? Could I delete them? why these extra 3 degrees of freedom appear I wish you can answer me when you see this comment which is very important to me. Thankyou very much