3DTopia / 3DTopia/MaterialAnything

Unexpected result running demo

Open
#13 3 comments 0 reactions 0 assignees View on GitHub
Dominant language
Python
Stars
367
Forks
18
PR merge metrics
No merged PRs in 30d

Description

I run the bash/test.sh on the demo.obj.
After the first material estimation step,I get the following results,whose albedo,metallic and roughness are unexpected...

Image

After running all the test script,I get the results like this,they are also unexpected,with the white albedo

Image

Image

I just changed the path of pretrained models to mine, and change the initial texture from unfounded texture_kd.png to this:
init_texture = Image.new('RGB', (args.uv_size, args.uv_size), color=(255, 255, 255))

The settings are as follows:

CUDA_VISIBLE_DEVICES=2 python scripts/generate_texture_pbr_3d.py \
--image2materials_model ./pretrained_models/material_estimator \
--uvrefine_model ./pretrained_models/material_refiner \
--input_dir $mesh_dir \
--output_dir $output_dir \
--obj_name mesh \
--obj_file mesh.obj \
--prompt "a rusty orange robot, ultra realistic" \
--add_view_to_prompt \
--ddim_steps 50 \
--new_strength 1 \
--update_strength 0.5 \
--view_threshold 0.1 \
--blend 0 \
--dist 1.2 \
--num_viewpoints 36 \
--viewpoint_mode predefined \
--use_principle \
--update_steps 0 \
--update_mode heuristic \
--seed 42 \
--post_process \
--device 2080 \
--use_objaverse # assume the mesh is normalized with y-axis as up

Contributor guide

No contributing guide indexed for this repository

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.