In the past I have used image layers for texturing lidar data and it’s allowed me to remove unwanted parts of the scanner image ( erasing people and vehicles). With Reality scan it seems to render the alpha channel as black instead of not using that part of the image. Did image layers alpha change?
Hello @LRewerts1
Do you mean the 2D view of the mask or that that the render is on mesh? Can you show some image what are you getting?
Hi, do you have also the mask layer in your dataset?
The provided screen shot shows texture layer:
and therefore the black part of the view is considered as a colored one, not the mask. To use masks, you need to have separated mask layer.
I’ve never used a mask layer previously for the texturing, but I’ve added one and the result is still showing black in the texture.
What are your mask usage settings?
Is your mask BW image?
Which version were you using before?
My mask is a black and white image. I’m not at work right now but I don’t think I have those mask options when working with lidar data. I’m not exactly sure what version of RealityCapture I last used image layer alpha but I may have some examples at work I can find.
As this is a laser scan, is it somehow modified? As there is not a cube rig after laser scan import…
In that case it is possible that the masks aren’t working, as for laser the layers don’t work (if I remember correctly).
Maybe you can try to use the texture layer as separated images for texturing aligned with the laser scans and use those masks on the texture images…
This is using the new Arial lidar and I’m using images priors to get the locations. This is vehicle lidar data so I have the image positions and rotations but the images themselves don’t align very well. We also use faro scan data and I can render images from the panoramas to use as texture layers and in the past I could just erase the parts of the image I didn’t want to show up in the texture. I never had to use masks.
OK, just for the info, the aerial lidar functionality wasn’t developed for the vehicle lidars or other movable ones. Basically, only the aerial data can be used. It can work also with other types, but the functionality is not guaranteed.
Would it be possible to share your data for some internal testing?
When I get to work I can put together a small sample. I wouldn’t think aerial lidar would work differently than regular lidar. They both create .lsp files. It seems the way image layers are used when texturing has changed. Everything else seems to work the same and produces a good result.
This looks different as the examples above… You have all data aligned in one component, the layer has different color…
Probably the best option will be to share your data for some internal test. When you’ll find some sharable, I can send you the invitation for the data upload.
In my initial example my _texture01 layer had images in the upwards direction that didn’t have points so no .lsp was made for it. When I added the folder those images are loaded as photos instead of image layers so I just disabled them. The color of the alpha channel I did realize depends on how it’s created. In photoshop it’s white, when I created my initial images in 3ds max my background was black so the alpha looks black. I can send you a small section of my latest data.
This is a small section from my initial data. I’ve created the alpha in Photoshop to show how it’s now displayed in white.
This is the data from the latest images.
RC_Data_sm.zip (19.6 MB)
Thank you for your data. I’ll create a bug for it.
Thank you.