Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Requirement: What kind of object do you want me to generate? #18

Open
ARuiChen opened this issue Jun 17, 2023 · 6 comments
Open

Requirement: What kind of object do you want me to generate? #18

ARuiChen opened this issue Jun 17, 2023 · 6 comments
Labels
documentation Improvements or additions to documentation

Comments

@ARuiChen
Copy link
Collaborator

Leave your requirement here. I will take some time to implement and upload the corresponding configuration files for you to reproduce.

@ARuiChen ARuiChen added the documentation Improvements or additions to documentation label Jun 17, 2023
@Thundros
Copy link

I would like the ability to have high quality model with normals, proper lighting, and proper texturing.

@ARuiChen
Copy link
Collaborator Author

Sorry, I may not have expressed myself clearly. What I mean is, when you can't adjust it yourself, I can try to help you adjust the parameters. For example, if you want to generate 'A blue tup' but fail, I can try to help you adjust the parameters and upload the relevant configuration files.

@ARuiChen ARuiChen pinned this issue Jun 17, 2023
@bennyguo
Copy link

Prompt: "a hamburger", "an astronaut riding a horse"
I would like to run on a single GPU, thank you!

@ARuiChen
Copy link
Collaborator Author

Hi, @bennyguo ! I have preliminarily generated it under the resolution of 128, and tomorrow I will release a more detailed object under the resolution of 256.
horse

@AmberCheng
Copy link

Prompt: "A metal five-pointed star decoration"
I have tried this prompt with configs like the car demo. However, I always get the result like
img_geometry_modeling_000030
But what I want is like this (I tried the prompt in Stable Diffusion Text2Img)
00001-878225224
How can I adjust the config to get a better result?

@oindrilasaha
Copy link

I am trying "A northern cardinal with wings spread out" starting with ellipsoid but after 500 iterations, the outputs turn from ellipsoid to pure noise. I am training on 3 A100s with a total of 12 batch size (4*3). This is the config I am using:
{
"mode": "geometry_modeling",
"sdf_init_shape": "ellipsoid",
"sdf_init_shape_scale": [0.5, 0.5, 0.5],
"translation_y": 0,
"random_textures": true,
"iter": 3000,
"coarse_iter": 1500,
"save_interval": 100,
"train_res": [512, 512],
"batch": 4,
"dmtet_grid" : 256,
"display": [{"bsdf" : "normal"}],
"train_background": "black",
"validate" : true,
"mesh_scale" : 2.6,
"out_dir": "cardinal_geometry",
"text" : "A northern cardinal with wings spread out",
"seed" : 0,
"add_directional_text": true,
"camera_random_jitter": 0.45,
"fovy_range": [25.71, 45],
"elevation_range": [-10, 45],
"guidance_weight": 100
}

Please let me know if this can be fixed. Thank you!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
documentation Improvements or additions to documentation
Projects
None yet
Development

No branches or pull requests

5 participants