Double exposure in Single Shot mode

Support & Feedback Suggestions
1

Hey there, lately I’ve been trying out the newly added Single Shot mode and I really like it. Though while using it to scan a black and white object an idea came to my head - taking double or bracketed exposure scans!

Since we don’t move the scanner nor the object, it could technically take two or more scans from the same position at different exposures and merge them together, creating a HER (High Exposure Range) scan. Such a mode would greatly help in scanning shoes for example, as they often have high-contrasting elements.

It could work like this - when you begin scanning a configuration screen with cam preview/feedback will appear asking you to select minimal scanning exposure (white surfaces), and then maximal scanning exposure (black surfaces) or perhaps multiple exposure steps if needed, after the config scanning would progress normally in One Shot mode (but maybe with a timer so the scanner has time to take all the scans). And that’s pretty much the basic idea of the suggested scanning mode.

I think this would be a possible and a pretty welcome addition, though the point filtering would need to be configured somehow to decrease noise (current red/blue areas). So what are your thoughts?

3 Likes
2

I believe that’s precisely what happens in Dark Object Mode. The camera will do 2 exposures, one fast and one slow.

On the hardware side, it looks like a MINI 2 could even do up to 5 exposures per single frame trigger, but the RevoScan software doesn’t expose that functionality (yet).

3

Scanning objects with different color surfaces will alter the accuracy because black objects absorb more infrared than white . The object should be prepared with 3D scan for optimal results and accuracy as with any other scanner in the world .
So if you going for merge after , I recommend you use 3D spray on all objects parts .

4

That is not what happening while using dark mode , dark mode have just higher exposure range than normal mode and is only usable for hard to scan surfaces .

There is of course limitation what surfaces can be scanned because of physics and nature of the surface materials , black is not the problem here but the material itself , mostly materials with higher IOR are hard to scan due to absorption .

Creating double shots will only create overlaps what for this technology of scanning is really bad idea , 90% of that will be removed while fusing anyway .

So no, no double exposures are nesesery or needed and should be avoided for perfect clean scanning process .
Use proper 3D spray and you don’t need any of that .

I see you don’t take any notes what I am saying early . It is not a photogrammetry it is 3D structured light scanning technology that have own rules and specific workflow to be followed for best results .

Revopoint’s scanners have all the capturing scanning process happening inside the hardware , the software is just a control panel , nothing can be added to the software without changing the hardware or the firmware of the device .

5

I would find that very surprising because RevoScan 5 uses the 3DCamera.dll on Windows and here’s what the 3DCamera documentation says about the HDR mode for low-reflection:

“@~chinese 暗色模式适用于测深色物体,
会按照设定的曝光级数,增加较高曝光参数进行几次曝光后进行融合输出。曝光级数由用户设定
@~english suitable for dark object”

(copied from Revopoint’s official GitHub: 3DViewer/thirdparty/3DCamera/include/hpp/Types.hpp at master · Revopoint/3DViewer · GitHub )

My understanding is that the Chinese says that dark mode will do multiple exposures and then fuse them. “进行 to perform 输出 to merge 输出 output”

And that also matches my observation that if I capture a frame in dark mode, I can see the blue/UV line pattern play twice.

It appears that for a static camera and a static object, fusing multiple exposures works really well as a noise reduction technique. The official Revopoint SDK calls it “time-domain smoothing” and looking at the source code, they are effectively just capturing between 3 and 7 frames and then averaging them: 3DViewer/src/cscamera/process/depthprocessstrategy.cpp at 516ee9b64ed9a36f7db2b1303d4c9364e39ac8d7 · Revopoint/3DViewer · GitHub

Yeah, I have noticed that, too. In the regular scan mode, they camera will report 800x600 pixel depth maps. And RevoScan 5 keyframes are also those 800x600x 16 bit images.

But 800x600 is only 0.48 Megapixels. And since the website for the MINI very prominently says 2 Megapixel Depth Resolution:

image
image1402×926 55.9 KB

I’m guessing there must be a 2 MP hardware mode that RevoScan does NOT (yet) use. My guess would be that they had to limit the USB transfer rate into RevoScan or else it might not work on iOS/Android or old laptops.

And looking at the SDK, there appear to be many configuration options for the hardware that are hidden in RevoScan:

Among them is also a user-configurable multi-exposure mode, meaning that the hardware can already do the exposure bracketing that @Sheeter was asking for, it’s just that RevoScan does not have a user interface for configuring it yet.

4 Likes