Dear Revopoint-Team, Hello Guys, Hello Ladys and Hello to all non binary’s
I noticed, that my computer with an 6core 4,5gHz CPU is 100% in use while capturing when scanning with POP2 and it is loosing the track. If I set it to Fast Scan or use POP1 the CPU is 80% used and it is harder to loose the track.
Maybe you could add a mode, where you capture only what happens and record it on the harddrive/sd-card, not into RAM, without processing it. Then after the raw-capture is done, the processing and track following stuff can be done slowly and most precisely or with manual adjustments.
Would that be possible? My smartphone is always running out of RAM. It would be nice to simply capture the scan (doesn’t matter if it takes a few gigabytes), send it to my computer and then process it.
There is very little processing during Revo Scan that does not have to be done. The Fuse and Mesh operations can even be held off until you move your files into Revo Studio.
But the Point Cloud still has to be formed in real time and that takes processing power, either in the POP or in the computer/smart phone.
But routing the data to hard drive instead of to RAM would be very much slower (unless you have an NVMe drive, and even then it would still be slower, just a little, not a lot).
On computer you don’t need to fuse the cloud points after scanning , you can do it later with batch processing using Revo Scan .
But if you don’t fuse the point cloud on your smartphone after scanning there is no way to do it later , as the cache data is removed if you leave the fussing/mesh/texture panel in Revo Scan .
We discusses this smartphone issue with @Revopoint-Cassie so she is already aware of the lost data and maybe we get some updates soon and have an option to erase or keep the cache data to use later on computer.
But once you fuse the point cloud after scanning you will be able to open it in Revo Studio and mesh it , the only thing that Revo Studio is missing is Texture Mapping function , so you will have to use secondary 3D software to extract the color data into textures like Zbrush or Blender etc…
I think about capturing raw 3d Data without tracking the orientation, or generating the model. That can be done after the data is collected. Maybe a better example is the turn table. I would need to adjust the speed of the turntable for a high-quality scan, which is not possible. At the moment there seems to be lost frames, and the track is lost, which means I need to start over, or undo. I know when it rotated 360°, and could stop capturing raw data. Then the process of tracking and generating the 3D object could be done frame by frame, where the processing of each frame can take as long as it needs.
My new turntable is 10 sec faster than the original one ( 20 sec per rotation) , I can change the speed but how slower it goes how worse the scan data , overlapped points and mess . If you do not capture color data the speed of the tunable really no matters , normally you will need to capture just 9 frames of the model in one turn 360 degree where on turntable it captures 300 frames in 30 sec . It all depends how fast the scan is , and since it is 10fps you don’t need 300 frames captured in 30 seconds .
I captured the shoe in 10 seconds around 100 frames … no issues with tracking .
POP2 uses visual tracking by feature , objects with a lot of features that are fully visible with the Depth camera sensor normally never lose the tracking , that why it is important to choice the proper mode for any objects .
I tried to capture frame by frame , but the minimum frames for generating point cloud is around 5 , so the object can be scanned with 360 degree turn with minimal 45 frames in around 4-5 seconds .
