Here my two cents on improving the software. Not spectacular, but anyway.
It starts with the installation procedure; a dialogbox opens with Chinese language. My Chinese language isn’t great (read: I have none), but I can only guess. It’s pretty clear clicking on the pull-down box that I can select a language here, for me English is the way to go. But when selecting English, nothing changes, the dialogbox remains in Chinese, so I have to guess which button to select. 50% chance I select the correct one. So improvement is: translate to English
Starting the software it maximises to the entire screen. This is not a major issue, but reducing the size of the program window can only be done by the little ‘box’-button in the upper right corner of the window. Other Windows programs you can reposition/resize to previous size by dragging the title-bar of the window. This does not work in the Revoscan software. So improvement here: use the same Windows functionality/convention of handling program windows.
That’s it for now, maybe I’ll come up with other improvements.
Thanks for reaching out to the community. Here are some of my thoughts. I will try to break them up into categories of simple things and perhaps more complex. These thoughts are from a Windows 10-Pro user that has both a Pop2 and a Mini.
Simple:
Not sure if this is a multimonitor thing, or if this is due to having a taskbar hidden, but when I launch the application, it does not remember the last state that it was closed in and launches in maximized mode. It also blocks going to the edge of a screen to get the hidden taskbar to pop up. I propose that there be some sort of cached set of values that remember the launch window and state that it was in as well as make it so that it does not block any auto hiding taskbar
When at the home/launch page of the program and then initiating a new scan, the preview windows switch from being on the right side of the screen to the left side of the screen. These should stay in the same place. I am not too particular on which side they are on, but it would be nice that they stay in the same spot.
For the RGB camera, the value for manual or auto exposure is not sticky. It is better in the latest version that it will launch in auto mode, but if you cancel that scan and then go to another scan, it goes back to manual. This should be saved in a cache/settings file that can be read in between each scan.
There should be either tool tips or other helpers that are around each functional element of the UI. For example. I had to go searching through the forums to find out what red and blue meant in the depth camera preview. It would be nice to have documentation that gives a detail on this and what sets the criteria for these items. Things like point fusing and when you leave it to be a bigger number or smaller number and what the differences might be.
In place of asking for a file name at the start of a scan, ask for it at the end of a workflow. This is when the user goes to save the scan after being happy with the scan, then can then save it. To start the scan, the program can establish a scratch space to save interim scans at, and can even have some sort of autosave/crash recovery that saves the last user defined number of scans in an auto recovery location. This will cut down on the failed scans that do not get cleaned out.
Provide a zoom to cursor option. Right now it appears to just zoom to the centroid of a model, but most of the time I want to zoom to a specific feature.
It has already been mentioned on the FB pages that there is a manual control mode coming to the turntable, and this is good to see. But it would also be nice if while in the automated mode that you could slow it down or speed it up. Also being able to restart an automated mode would be nice for the cases where I stopped a scan and deleted the previous scan items or undid some of them, I cannot restart the turntable.
Provide a method to reset the coordinate system in the software. Allowing for options to select flat surfaces as planes for this would be great, but even selecting points for this and then doing a best fit of a 3D coordinate system would be fantastic.
Harder:
Issue #1:
The visual odometry portion of the program seems to be the biggest drawback of the current program. All the other items could be overlooked, but the poor visual odometry portion make scanning and getting a good scan for complex or larger objects a fine art. There are a couple of areas that this could be improved upon.
Potential Solutions:
The first is that if one is using the bluetooth, dual axis turntable, the program should be able to know where the turntable is at during all times. While it may be running in open-loop mode, it can report out the number of steps it has made since the last report as well as the table angle. Knowing these items apriori should help feed into the program for keeping tracking in place, especially with a scanner in a fixed position.
I believe that the scanners also have accelerometers/gyros, this should be feeding into the tracking algorithm as well so that as the scanner moves, it can understand that the camera might be moving with relation to the model and predict that movement.
Using the RGB cameras for visual odometry to also track movements. Fusing all of these items into a single solution to give approximate location to the target being scanned will help in maintaining location relatively in space with respect to the scanner regardless if the scanner is moving, if the object being scanned is moving, or if both are moving. It seems that this coupled with the actual returns from the scanner being combined in a Kalman filter would likely bring about some great things, but I know that there are a lot of other methods. Looking up “3d scanning visual odometry” on Google brought a lot of information talking about SLAM algorithms coupled with inertial data to help. I know that Robot Operating System has some of these algorithms in it as well.
Item #2:
Allow for multiple scanners to be used at the same time. The use of multiple scanners could allow for better tracking. Say for example using a Pop 2 with a Mini such that the Pop 2 is placed further away with a wider field of view with the mini scanning details. This would need to be thought through rather well, but I think could provide some unique capabilities. They could be mounted one on top of the other and calibrated together.
These are the thoughts that I have right now. I am going to go back and play with my Mini that just showed up this weekend.
I agree 100% with having an array. If they could have some kind of mounting bracket where they would absolutely know the relationship between the minis or pop2’s. Then if the software could somehow sync them increasing both the scanning area and the accuracy.