Mesh output?

[MattiaCasalegno-PRATT]

Hello, thank you for sharing this amazing project.
I was wondering what types of meshes can we export? And are those textured or vertex colored? I'm interested in exporting 3d reconstructed scenes into Blander or meshlab.
Thank you

[Angkitbharata]

Hello, thank you for sharing this amazing project.
I was wondering what types of meshes can we export? And are those textured or vertex colored? I'm interested in exporting 3d reconstructed scenes into Blander or meshlab.
Thank you

[MarvinChung]

There is a save_mesh argument in the scripts scripts/run.py which supports OBJ and PLY format.

First prepare a folder to save the model and mesh.

mkdir saved  

Then run the script

python3 ./scripts/run.py --mode nerf --scene ./data/nerf/fox/ --save_snapshot saved/fox.msgpack --save_mesh saved/fox.ply --gui --train

You can get the fox.ply in the saved directory and then export it to blender. It takes a while for me to export it to the blender.
My result looks like this:

[entropical]

Hi Marvin, thanks for this answer. Do you know how you would be able to include colour information in this? Ultimately I would like to convert it to a pointcloud, and render it with Potree. Some samples in Potree include colour information.

[INF800]

There is a save_mesh argument in the scripts scripts/run.py which supports OBJ and PLY format.

First prepare a folder to save the model and mesh.

mkdir saved  

Then run the script

python3 ./scripts/run.py --mode nerf --scene ./data/nerf/fox/ --save_snapshot saved/fox.msgpack --save_mesh saved/fox.ply --gui --train

You can get the fox.ply in the saved directory and then export it to blender. It takes a while for me to export it to the blender. My result looks like this:

@Tom94 haven't you mentioned exporting is not possible with current codebase? What exactly is the difference here with -save_mesh flag

[aceIsHere]

Hey man what will be the equivalent like for google colab for this notebook (https://colab.research.google.com/drive/10TgQ4gyVejlHiinrmm5XOvQQmgVziK3i?usp=sharing)

Basically after last line I want to export/save my file as a .obj or .ply format so that I can download and try it on my local blender environment

I tried to iterate your line for colab but it was not successful. I hope you can help me here. Thanks!

[MarvinChung]

Sorry, I have no idea. Maybe there is not enough disk space?

[aceIsHere]

Trying to figure out what line 295 is all about in the scripts/run.py, But still thanks for helping me out :)

[aceIsHere]

Issue solved refer to #427

[MarvinChung]

I see. I will also edit my solution to prevent confusion.

[LincolnG17]

I can't use this command either. I have tried both my own nerf and the default fox:
(base) PS C:\ngp\instant-ngp> python ./scripts/run.py --mode nerf --scene ./data/nerf/fox/ --save_snapshot saved/fox.msgpack --save_mesh saved/fox.ply --gui --train
Warning: the '--mode' argument is no longer in use. It has no effect. The mode is automatically chosen based on the scene.
12:28:12 SUCCESS Initialized CUDA. Active GPU is #0: NVIDIA GeForce RTX 3060 Ti [86]
12:28:12 INFO Loading NeRF dataset from
12:28:12 INFO .\data\nerf\fox\transforms.json
12:28:12 SUCCESS Loaded 50 images after 0s
12:28:12 INFO cam_aabb=[min=[1.0229,-1.33309,-0.378748], max=[2.46175,1.00721,1.41295]]
12:28:14 SUCCESS Initialized Vulkan and NGX on GPU #0: NVIDIA GeForce RTX 3060 Ti
Training: 0%| | 0/35000 [00:00<?, ?step/s]12:28:14 INFO Loading network config from: .\configs\nerf\base.json
12:28:14 INFO GridEncoding: Nmin=16 b=1.51572 F=2 T=2^19 L=16
12:28:14 INFO Density model: 3--[HashGrid]-->32--[FullyFusedMLP(neurons=64,layers=3)]-->1
12:28:14 INFO Color model: 3--[Composite]-->16+16--[FullyFusedMLP(neurons=64,layers=4)]-->3
12:28:14 INFO total_encoding_params=13074912 total_network_params=10240
Training: 27%|█████████████▋ | 9557/35000 [01:57<04:57, 85.45step/s, loss=0.000416]12:30:12 WARNING Vulkan error: Validation Error: [ VUID-VkImageCreateInfo-extent-00944 ] Object 0: handle = 0x1e6baecb790, type = VK_OBJECT_TYPE_DEVICE; | MessageID = 0x8869da08 | vkCreateImage: parameter pCreateInfo->extent.width (= 0) is not greater than 0 The Vulkan spec states: extent.width must be greater than 0 (https://vulkan.lunarg.com/doc/view/1.3.239.0/windows/1.3-extensions/vkspec.html#VUID-VkImageCreateInfo-extent-00944)
12:30:12 WARNING Vulkan error: Validation Error: [ VUID-VkImageCreateInfo-extent-00945 ] Object 0: handle = 0x1e6baecb790, type = VK_OBJECT_TYPE_DEVICE; | MessageID = 0xeaa98d2c | vkCreateImage: parameter pCreateInfo->extent.height (= 0) is not greater than 0 The Vulkan spec states: extent.height must be greater than 0 (https://vulkan.lunarg.com/doc/view/1.3.239.0/windows/1.3-extensions/vkspec.html#VUID-VkImageCreateInfo-extent-00945)
Training: 27%|█████████████▋ | 9557/35000 [01:57<05:12, 81.44step/s, loss=0.000416]
Traceback (most recent call last):
File "C:\ngp\instant-ngp\scripts\run.py", line 179, in
while testbed.frame():
RuntimeError: C:\ngp\instant-ngp\src\dlss.cu:657 cudaGetMipmappedArrayLevel(&first_level_array, m_cuda_mipmapped_array, 0) failed with error invalid resource handle
12:30:12 WARNING Vulkan error: Validation Error: [ VUID-vkDestroyDevice-device-00378 ] Object 0: handle = 0x1e6baecb790, type = VK_OBJECT_TYPE_DEVICE; Object 1: handle = 0xa9fdf000000014c, type = VK_OBJECT_TYPE_IMAGE; | MessageID = 0x71500fba | OBJ ERROR : For VkDevice 0x1e6baecb790[], VkImage 0xa9fdf000000014c[] has not been destroyed. The Vulkan spec states: All child objects created on device must have been destroyed prior to destroying device (https://vulkan.lunarg.com/doc/view/1.3.239.0/windows/1.3-extensions/vkspec.html#VUID-vkDestroyDevice-device-00378)
12:30:12 WARNING Vulkan error: Validation Error: [ VUID-vkDestroyDevice-device-00378 ] Object 0: handle = 0x1e6baecb790, type = VK_OBJECT_TYPE_DEVICE; Object 1: handle = 0xe8fb46000000014d, type = VK_OBJECT_TYPE_DEVICE_MEMORY; | MessageID = 0x71500fba | OBJ ERROR : For VkDevice 0x1e6baecb790[], VkDeviceMemory 0xe8fb46000000014d[] has not been destroyed. The Vulkan spec states: All child objects created on device must have been destroyed prior to destroying device (https://vulkan.lunarg.com/doc/view/1.3.239.0/windows/1.3-extensions/vkspec.html#VUID-vkDestroyDevice-device-00378)
12:30:12 WARNING Vulkan error: Validation Error: [ VUID-vkDestroyDevice-device-00378 ] Object 0: handle = 0x1e6baecb790, type = VK_OBJECT_TYPE_DEVICE; Object 1: handle = 0xd406ad000000014e, type = VK_OBJECT_TYPE_IMAGE_VIEW; | MessageID = 0x71500fba | OBJ ERROR : For VkDevice 0x1e6baecb790[], VkImageView 0xd406ad000000014e[] has not been destroyed. The Vulkan spec states: All child objects created on device must have been destroyed prior to destroying device (https://vulkan.lunarg.com/doc/view/1.3.239.0/windows/1.3-extensions/vkspec.html#VUID-vkDestroyDevice-device-00378)

[Angkitbharata]

So thankyou Pada tanggal Sel, 5 Apr 2022 01.28, aceIsHere ***@***.***> menulis:

…

Still Thanks for helping me out :) — Reply to this email directly, view it on GitHub <#412 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AX7H3SDGYBEDOQODHID56X3VDMRENANCNFSM5SOLARFQ> . You are receiving this because you commented.Message ID: ***@***.***>

[barakmam]

For the color you can use MeshLab software

[tempdeltavalue]

Hi, how to add color inside blender (not Meshlab) ?

[lucasew]

Just use meshlab, its way easier and its free

[tempdeltavalue]

try to launch camera view , blender doesn't render color in "object/edit" mode 🤷‍♂️ (F12)