-
Notifications
You must be signed in to change notification settings - Fork 2
Ways of creating models #15
Copy link
Copy link
Open
Labels
discussionNot to be attached to a specific PRs/implementation. Each concrete feature should have its own issueNot to be attached to a specific PRs/implementation. Each concrete feature should have its own issue
Metadata
Metadata
Assignees
Labels
discussionNot to be attached to a specific PRs/implementation. Each concrete feature should have its own issueNot to be attached to a specific PRs/implementation. Each concrete feature should have its own issue
Type
Fields
Give feedbackNo fields configured for issues without a type.
I tried creating models of my room from 236 pictures taken using my iPhone 6S (12MP). It took 4.5 hours for execution of MeshRoom on Google Colab which is free.
Pikachu and friends:

Piles of books:

Entire view:

Pictures and outputs (.obj + .stl + textures) are available here. Snapshots above show that surfaces such as walls and desks are kind of rough. 100+ images aren't enough? On the other hand small stuff like stuffed toys and laptops look nice.
I'm going to compare several methods to create models, which are MeshRoom + Colab, MVE + my laptops, MVE + Colab, 3DF Zephyr etc... It is helpful that somebody tries to create models using materials provided in the link above by their own GPUs or environments.