Comments (4)
@MondRubberduck So the video shows a correctly generated object, and only the mesh exporting fails? Could you check the printed log? It should say something like this:
==> Saving mesh to trial/mesh
[INFO] running xatlas to unwrap UVs for mesh: v=(129072, 3) f=(257500, 3)
[INFO] writing obj mesh to trial/mesh/mesh.obj
[INFO] writing vertices (129072, 3)
[INFO] writing vertices texture coords (236023, 2)
[INFO] writing faces (257500, 3)
==> Finished saving mesh.
from stable-dreamfusion.
@ashawkey !Thank you for the reply! It actually doesn´t
[INFO] Trainer: df | 2022-10-11_11-16-57 | cuda | fp16 | trial [INFO] #parameters: 12249015 [INFO] Loading latest checkpoint ... [WARN] No checkpoint found, model randomly initialized. ==> Start Test, save results to trial\results 97% 97/100 [00:03<00:00, 53.96it/s]==> Finished Test. ==> Saving mesh to trial\mesh [INFO] running xatlas to unwrap UVs for mesh: v=(0, 3) f=(0, 3) 100% 100/100 [00:04<00:00, 20.19it/s] Traceback (most recent call last):
Then it goes through some files and the last bit:
RuntimeError: instance mode - pos must have shape [>0, >0, 4]
It warns me that there are no checkpoints. However, I do have my checkpoints in the default files structure ./stable-dreamfusion/mytest/checkpoints
Couple of large PTH files. -> df_ep0100.pth (for example)
from stable-dreamfusion.
Update: I literally didn´t change anything. Deleted all the checkpoints /results/df´s etc and ran another test with:
python main.py --text "a high quality, detailed photo of a mercedes benz" --workspace mytest -O --iters 5000 --albedo_iters 6000 --lambda_entropy 1e-5
Looks like shit (well duh. These settings were for a quick result), but at least I got a mesh from it.
from stable-dreamfusion.
@ashawkey If I may steal your time for one more second: What parameters have the biggest impact on quality? Thanks
from stable-dreamfusion.
Related Issues (20)
- "Overhead" prompt for elevation angle HOT 1
- solve 'NaN or Inf found in input tensor' with disable --fp16 HOT 1
- Omit the diffusion gradient term via graph detaching instead of SpecifyGradient hack HOT 5
- Any plans to support SDXL? HOT 1
- Unable to save meshes, everything else worked. HOT 1
- Question about zero123 azimuth angle
- Error running DeepFloyd-IF only
- Zero123 on multiple images HOT 3
- Assertion Error Torch not compiled with Cuda HOT 2
- Possible bug when "as_latent" is used?
- Inaccurate grad scale with the new sds loss HOT 1
- TypeError: grid_encode_forward(): incompatible function arguments. HOT 1
- TypeError: forward() takes 2 positional arguments but 4 were given HOT 3
- Colab error: ModuleNotFoundError: No module named '_gridencoder' HOT 2
- Question about replace the training dataset
- AttributeError: module 'pymeshlab' has no attribute 'Percentage' when export mesh.
- Replace Text with audio+Text
- Colab install dependencies error HOT 3
- Tutorial about the project HOT 1
- Getting Bus Error
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 stable-dreamfusion.