Create a more accurate 3D model

[rezalahmi]

Hi, thank you for this beautiful work.
I tried this model on several images. Why are the nasal structures of all exits the same? I have a bulge and deviation on my nose, but the nose model makes my nose look very beautiful. It's like I've had a nose job. Is there a way to improve the work and make more realistic models?
I think the model is not working correctly in determining the depth of the image. Therefore, it does not recognize the protrusion of the nose.

[mrsaleh]

What you are looking for is something like DECA, 3DDFA is very good for the first step of detailed face reconstruction.

[Zvyozdo4ka]

HEllo, @rezalahmi.

How did you run face reconstruction?

[rezalahmi]

Hi Julianna Yes, I did

…

On Mon, Oct 16, 2023 at 06:28 Julianna ***@***.***> wrote: HEllo, @rezalahmi <https://github.com/rezalahmi>. How did you run face reconstruction? — Reply to this email directly, view it on GitHub <#222 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AIE3H4WLV5V5GGZCDCDCS33X7SPFPANCNFSM6AAAAAASML7KCM> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[rezalahmi]

I don’t remember now, let me check and get back to you

…

On Mon, Oct 16, 2023 at 07:06 Reza Lahmi ***@***.***> wrote: Hi Julianna Yes, I did On Mon, Oct 16, 2023 at 06:28 Julianna ***@***.***> wrote: > HEllo, @rezalahmi <https://github.com/rezalahmi>. > > How did you run face reconstruction? > > — > Reply to this email directly, view it on GitHub > <#222 (comment)>, > or unsubscribe > <https://github.com/notifications/unsubscribe-auth/AIE3H4WLV5V5GGZCDCDCS33X7SPFPANCNFSM6AAAAAASML7KCM> > . > You are receiving this because you were mentioned.Message ID: > ***@***.***> >

[Zvyozdo4ka]

@rezalahmi - many thanks! I am looking forward to it

[ZHJNCUT]

Are you aware of this? The pose parameter of the 300W-LP dataset is 7 dimensional
However, in the implementation code of the 3DDFA paper, the 12 dimensions of regression are used directly
I don't know the connection, do you know this? There are many people asking this question on github, but I can't find the answer, can you help me

[Zvyozdo4ka]

Are you aware of this? The pose parameter of the 300W-LP dataset is 7 dimensional However, in the implementation code of the 3DDFA paper, the 12 dimensions of regression are used directly I don't know the connection, do you know this? There are many people asking this question on github, but I can't find the answer, can you help me

Unfortunately, i am not aware. I got inappropriate results, and the output did not resemble the aimed person.
And nobody actually gave any advice or hint, so i just gave up and found more accurate methods

[ZHJNCUT]

你知道这一点吗？300W-LP数据集的位姿参数是7维的，但是在3DDFA论文的实现代码中，直接使用了回归的12维，不知道有什么联系，你知道这个吗？github上有很多人问这个问题，但我找不到答案，你能帮我吗

不幸的是，我不知道。我得到了不适当的结果，并且输出与目标人不相似。 实际上没有人给出任何建议或提示，所以我只是放弃并找到更准确的方法

Ok, thank you. I saw that the code evaluation scheme in 3DDFA only has two-dimensional feature points (NME) error, there is no three-dimensional vertices (NME) error, I learned that the normalization of the calculation of three-dimensional vertices (NME) error is: the three-dimensional outer eye spacing of the face, you know how to find this, its index values are how much?