I have been in a Reality Capture modelling-block for more than a year. I have been trying to capture an asymmetrical item and have two large data-sets.
Respectively a data set 1, shoot from the top halve and data set 2 from the bottom half.
I haven’t found a way to flawlessly merge the two meshes/datasets (haven’t tried mask yet). Shooting a new data set is not possible, so I must do with what I got.
With control points I managed to put the two halves together, but I got a slight offset in my final reconstruction. Probably due to having an organic object, with a slight move in one of the datasets.
Recently I have been trying to master Zbrush (2022).
Is it possible to reconstruct the two halves separately in RC, seamlessly put the mesh together and repair the offset with Zbrush and import back as a whole in Reality Capture for further texturing with the photo’s in the dataset?
If all above is possible, what would be the workflow?
To be able to combine models in any way in RealityCapture, it should be possible to combine components (sparse point clouds) which us not supported in RealityCapture. Without cameras looking at the model, you won’t be able to texture all of the models.
Try adjusting the control points weights to align more to their positions, and if possible, mask out the organic object.
Hi, everything is possible but this won’t be easy and the result might not be ideal.
First of all, I would personally try to get a single-component RC and try to minimize the offset. Try the mask workflow first and control points.
What you are asking with two halves is possible. What I see as the biggest problem will be when you try to import the hole-repaired mesh back into RC to match the camera positions for proper texturing because you will have two separate RC reconstructions in two different coordinate systems.
So after merging and repairing the mesh, also unwrap it in ZBrush and export. BTW I recommend exporting the model from RC as OBJ and the same from ZBrush.
Let’s say the exported mesh will match one of the RC reconstructions because you moved one half to the other. Now you need to find a way to match it to the second camera alignment. That could be possible with a tool like CloudCompare. Basically in Cloucmpare use the original exported mesh as a reference and match the repaired merged mesh to it.
So you should end up with two identical meshes with the same unwrap but differently rotated. Import the meshes to corresponding RC reconstructions and texture them. You will end up with two textures but since they have the same UV layout you can easily combine them in Photoshop for example.