Comments (8)
Thank you for your attention, we will release the code soon.
from hfs.
Hi, I understand that you may be busy with other work, therefore I try to reproduce your work, since it seems not hard based on NeuS. Here summarize a few detail questions, and I hope you can give me some quickly reply :
1).Your paper did not mention about the background modeling as NeuS use a nerf++ for background modeling. I wonder how you deal with the background in the dataset as DTU at training time.
2).You have implement 2 MLP for SDFnetwork for displacement, I wonder how you deal with them when you implement feed them into the color network for volume rendering. Which density you choose for volume rendering? Or make some combination with both of them?
3). For the sampling strategy, As you mentioned in your paper "then adaptively update the weight according to the gain and sample an additional 64 points". Are these 64 points sampled as Neus did? eg. construct the pdf, and gradually sample 16*4 points near the surface. Does it mean you just have changed the Transparency weights using the eq.16 in the paper while other is the same as neus did.
4). In the eqution 14, why there is a scale term 4 at MLP_d, what will happen if it become larger or smaller, I guess you may try other scales, could you share the design about that?
Sorry for asking so many problems, Since I am following your work and try to extend it for Scene-level reconstruction, I wish you can give me some quick advice.
Looking forward to your kindly reply
from hfs.
Thanks for your interest,
1). We use the same way as NeuS to model background,
2). Eq. 14 is used for the combination of two SDFs.
3). Sec. 3.1 shows the difference between others.
4). The term 4 is applied for normalizing the maximum constraint to 1.
We are going to release the cleaned code next week. Cheers.
from hfs.
Thanks for your quick reply and your code releasing plan, (it seems some questions are just my misunderstanding), Any way your reply made me clear to understand, And now I will closs the issues, Thank you very much, and looking forward to your cleaned code next week!
from hfs.
Hi there,
Will your code be released this week?
Since my reproduced quality is not very well, I need your implementation for refrence, thank you !
Looking forward to your reply
from hfs.
Following your details in your paper, here is my reproduced results on blended MVS man.
Here is the results from NeuS:
The back of the head is not well reconstructed, I think it's some of my missunderstanding for your work, Therefore, I am looking forward to your implementation released. Or if you have meet the same issues, could you please simply point the possible problems?
from hfs.
Hi, today I am preparing some important materials to complete neurips requirements. I have cleaned up the code and will release the code tomorrow evening.
from hfs.
Thanks for your code releasing.
from hfs.
Related Issues (17)
- Question about the paper: 𝛼-composition and the case of multiple surfaces HOT 1
- no HOT 1
- Intensive Memory Usage
- blender dataset HOT 2
- Mesh without Displacement Function HOT 2
- Change to multi card training HOT 1
- Using pre trained models!!
- evaluating the mesh on NeRF synthetic dataset
- error
- Training Data HOT 7
- Training time and TQDM HOT 1
- Blendedmvs evaluation
- What's the difference between s and inv_s in function render_core? HOT 1
- Rendering is fine but mesh is empty HOT 1
- How to get water-tight mesh for evaluation? HOT 2
- Request for DTU training config with mask HOT 1
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from hfs.