vrn
vrn copied to clipboard
AaronJackson
Improve the result
Hello Aaron, good job! I try to use your implementation, but the result that I get is worse than in the demo page. I use it from python, what can I do to improve the result?
thanks!
Hi, the built in render script is not very good. You might want to try visualising the obj file with something else. Meshlab supports per-vertex colouring, although I can't remember how to enable it.
@MaskeZen Hello, I'm a bit new up here. Can you please tell me if you managed to improve the result of image processing? I have the same problem as you. At the moment I am trying to solve it, would be very grateful for the help. Thank you
@MaskeZen Hello, I'm a bit new up here. Can you please tell me if you managed to improve the result of image processing? I have the same problem as you. At the moment I am trying to solve it, would be very grateful for the help. Thank you
Hello @sergey1513, yes in the way that @AaronJackson say. I use the file raw2obj.py to get a .obj file, then with the meshlab the result is a lot better.
Now I am working in a script to use with blender, to automate the process.
Write if you need more detail or if you want share others approaches, good luck!
@MaskeZen I'd love to chat with you about your implementation of vrn / this projection pipeline using MeshLab and potentially help out with some of the scripting/chain this with a pipeline that I have built to combine faces with head assets and output fully-textured bald heads :+1:
@MaskeZen thanks for sharing your results. I am not a coder but more of a graphic artist with some basic hacking skills. I have played around with vrn but my results are very smooth. The voxels are clearly seen. I am very interested in learning your method and also taking to Blender. Thanks.
Hello @richardtwindom @umeshukla, I took notes of the environment that I use to make the test. I passed these notes to a gist to share with you.
In the next days I translate the spanish coments to english, if you like I can chat (with my bad english) from 18:00 to 21:00 UTC.
You can find me in the mail box maskeee on the mail of google.
see you!
@MaskeZen Hello, great job. Could you share with us the steps you followed to get this rendered smoothly? Many thanks.
@MaskeZen Hello, I am a designer looking to follow your notes and before getting started, just wanted to understand whether a nvidia driver is a prerequisite or is it simply to speed up the calculations? I do have a gaming laptop with Win10 and a good RTX card but and I cant have Nvidia access in a centos7 via a VM.
Therefore to make Aaron's VRN script work, do you suggest:
- to try to do it all in Win10
- CentOS as VM but with no graphics acceleration
- CentOS as a dual boot/flash drive
Do I even have a choice?
Thank you for your feedback!
Hello Pierre-Henri, first of all, sorry for if my english is bad, I speak in spanish, I'm from Uruguay. If you have Windows 10, may be you can try with the Linux Subsystem for Windows. (NVIDIA Drivers for CUDA on WSL, including DirectML Support)
I made the test in Ubuntu, the cuda installation can be a little tricky, don't worry and be patient. If this doesn't work may be you can try with a dual boot installation. Is not necessary CentOS, I use Ubuntu with a Docker image of CentOS.
If I can help in any way please let me know.