New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Can provide some dataset to test the demo (like default Candi Banyunibo data set)? #7
Comments
seems I go to fast I read now: "add the option to drag&drop your own point clouds after vacation. We will also add a test data set by then." |
I'm not sure how long this will be around for, but: http://5.9.65.151/mschuetz/temporary/MorroBay.zip From this tweet: https://twitter.com/m_schuetz/status/1526923601433268225 |
I'll be up for a while, the temporary folder is just the place I start removing stuff when I run out of storage. |
@m-schuetz for example copied from demo:
console shows errors (lots of OPENGL DEBUG CALLBACK: GL_INVALID_VALUE error generated. + is greater than the value of BUFFER_SIZE for the buffer bound to .):
|
Might be an issue with GL_ARB_sparse_buffer and/or that GPUs older than the Pascal series are not supported. I'm using a sparse buffer that allocates 16GB virtual memory at startup, but only physically allocates as much memory as needed to load the point cloud. Since it's 16GB virtual and not actual physical memory, I was under the assumption that it should be working on GPUs with 4GB. I'll give it a try on my GTX1060 3GB when I get to it. If it works there, then the issue might be that something is not working on GPUs older than that. |
Btw, this earlier branch/tag doesn't use sparse buffers so maybe it could work there: https://github.com/m-schuetz/compute_rasterizer/tree/compute_rasterizer_2022 But it also doesn't support drag&drop so you'll have to modify main.cpp to load your own data set, and it also only supports loading a single point cloud. The reason for using sparse buffers is that it makes it easy to put several different point cloud data sets into one huge large buffer, which makes implementing stuff a whole lot easier. |
Loading 100M points works on my 1060 3GB (it's already close to the limit, though). I can only assume that there might be issues with GPU generations before Pascal. |
Hi @m-schuetz thanks for all info and sorry for delay in answering..
|
Ah, then the or one of the problems is the hardcoded maximum size (2048*2048) of the framebuffer (actually a regular storage buffer where points are rasterized into with atomicMin): https://github.com/m-schuetz/compute_rasterizer/blob/master/modules/compute_loop_las/compute_loop_las.h#L96 Thanks for the hint, I'll have to fix that. |
Fixed - now supports larger window sizes. |
thanks! will try soon.. |
Hi,
just found this paper & project via graphics weekly news..
just compiled the demo, and seems uses that dataset by default (banyunibo_inside_morton)..
anyway to obtain that dataset?
if not can you provide some download link to some huge & equivalent data set used by the demo like retz,eclepens,etc..
are this datasets under non "open" licenses?
just wanted to test performance on my Titan V compared to a 3090..
thanks..
The text was updated successfully, but these errors were encountered: