Improving Small Object Scan Results on the Mini 2?

[kayembi]

Hello all,

I got a Revopoint Mini 2 a few days ago, and have been learning how to scan small objects with it, following the various YouTube tutorials and going through the threads and help materials here. So far, I’m mostly happy with the results, but I’m wondering if I can do better.

Short version: I’m trying to scan the head of an action figure and am looking for tips on making the details sharper and clearer. Pictures of the object and my results are below.

Long version:

I’ve been testing by scanning the head of an action figure, since it’s one of the smallest objects I’m likely to scan. My results are decent, but lack clarity and sharpness, with features in the scan appearing too soft.

At 14 x 14 x 12mm, the action figure head is close to the Mini’s 10mm lower limit. I therefore expect I’m just hitting the scanner’s limits. If so, that’s fine - the results I’m getting are okay, and I’ll be retopologising in Blender anyway. But I am interested to find out if I can get even more detail/clarity out of it, so would be grateful for any tips on how to improve my scans.

Here’s the object I’m scanning:

What I’m doing:

To aid with tracking, after trying a number of things and seeing Steamkraft’s amazing results in his Showcase, I created my own scanning helper base copied from his:

I also put random pieces of Blu-Tack around the base.

I’m using 3D scanning spray so that the red and silver areas are picked up. (In fact, I found that both Aesub Blue and Revopoint spray can damage some paint on action figures, so I switched to using dry shampoo, seeing no difference in the results.)

I have the turntable set to a low speed, the scanner placed at a distance Revo Scan 6 determines as excellent, and exposure set so that there is barely any red or blue (the “1” setting, basically). I’m using feature tracking with high accuracy. And I’m only doing one full rotation for the scan.

Once the scan is complete, I fuse with a point distance of 0.05mm, remove the base, then do isolation at 80% and overlap at different values, all of which is based on stuff I’ve read here. I mesh at 0.05mm to match the fused point distance (I tried 0.02mm but it makes no visible difference).

Here’s one of my best scans after fusing:

And here are a couple of my best results after meshing, exported to STL and imported into Blender:

I also tried scanning from four different angles and fusing the four scans, but that turned out pretty horribly:

Do these results look about the best that could be expected given the size of the object I’m scanning and the Mini’s limits, or can I hope to sharpen things up a bit? Any tips or suggestions would be very much appreciated.

Incidentally, one thing I noticed is that, when I go to fuse the point cloud in Revo Scan 6, I am limited to a point distance of 0.05 - 0.2mm, but I have seen posts in this forum talking about fusing at 0.02mm for the Mini. Is that actually possible? I assume 0.05mm is the lowest it can go, given that the accuracy is listed as 0.05mm. (I’m using Revo Scan 6 on macOS, in case that makes a difference.)

Also, I’ve read on these forums that the best scanning distance is around 10-12cm. But if I try to place the scanner any closer than around 25cm, the object isn’t detected and Revo Scan tells me the distance is too near. I’m not sure if that is a relevant factor.

Thanks for reading!

[PUTV]

Hi @kayembi

Try to use single shot mode around 36-4O frames per 360 full rotation , remove everything that don’t belongs to the scan before fusing , it will give you already better results .

Try to download Revo Scan 5 , it still have fusing setting of 0.02mm for Mini 2

Dey shampoo will erase the hard edges and corners , for this small type of scans I use 3D Liquid spray , you can also made it yourself using Zinc Oxide and low % alkohol. Then use brush to paint just the colors you want in place of spray everything .

[kayembi]

Hi @PUTV,

Many thanks for the helpful reply! I’ve been testing out your suggestions, but unfortunately haven’t had much luck so far; maybe I’m doing something wrong.

@PUTV

Try to use single shot mode around 36-4O frames per 360 full rotation

I set everything up as before (high accuracy, feature mode, excellent distance, my custom base to help with tracking etc), and this time used single-shot mode, rotating the object manually and taking around 40 shots for the full rotation, as you suggested. But my result was worse than before. I tried several times, but this was the best I could get:

In one test I tried taking several shots for each angle, as suggested in the YouTube video on using single-shot mode, but that didn’t help.

(Incidentally, there seems to be a nasty bug in single-shot mode in the macOS version of Revo Scan 6. If it loses tracking and you hit Undo, you nearly always lose all the scan data and have to start again.)

@PUTV

remove everything that don’t belongs to the scan before fusing

Sorry, but could you please explain what you mean by this? As far as I can see, it’s not possible to edit the point cloud before fusing, so I don’t see any way of removing the support base or bits of blu-tack that get picked up by the scan before fusing. The polygon and selection tools are all disabled until after fusing, unless I’m missing something.

@PUTV

Try to download Revo Scan 5 , it still have fusing setting of 0.02mm for Mini 2

I tried Revo Scan 5 at 0.02mm (which I notice is only available in “standard” fusion mode - “advanced” fusion mode only goes down to 0.05), but the results were very grainy:

When I launched Revo Scan 5, it recommended that I recalibrate the scanner’s IMU - I ignored that, though, as it’s a new scanner. Should I try recalibrating?

Thanks for the tip on the zinc oxide - I’ll order some and give that a go.

[PUTV]

@kayembi

Sorry, but could you please explain what you mean by this? As far as I can see, it’s not possible to edit the point cloud before fusing, so I don’t see any way of removing the support base or bits of blu-tack that get picked up by the scan before fusing. The polygon and selection tools are all disabled until after fusing, unless I’m missing something.

Hi @kayembi

You can edit the raw scan before fusing , just click the 3 dots next to the scan for the menu and function if edit , after you remove everything it will ask you to save before moving to fusion .

When you use different software always calibrate it , no matter new scanner or not the IMU is for the giro not the actual calibration , I think that may be the problem .

I was able to scan wings if insect with Mini 2 and get full details of it . So you must be able to do that scan without issues .

When you scan using single shot use the Dual Axis Turntable for automatic process if you own it .

If not you can move the object manually just small angles at once.

After full calibration you should get perfect result using fusing in the Advanced mode , the standard mode include too much noise data and need to be cleaned , the advanced mode gives you the best result , after fusing use the same GRID settings as the fusing , if it was at 0.05 use that settings on the GRID while meshing to archive best meshing .

I also scanned small coins so the results was great , I am 100% sure you get there with little practice .

When you merging scans use the best scan as first in the list to keep the best accuracy , by selecting it as first .

Make sure you clean the both scans before merging and after merging to remove all overlapped points before meshing.

[kayembi]

Many thanks again for the help and advice - much appreciated. I have managed to improve my results a fraction, but there are still some areas I’m struggling with, so clearly have plenty to learn.

@PUTV

You can edit the raw scan before fusing , just click the 3 dots next to the scan for the menu and function if edit , after you remove everything it will ask you to save before moving to fusion .

Thanks. I had difficulty finding this using RevoScan 6, but looking at RevoScan 5, I assume that’s the version you are referring to, and “Key Frames Edit” is what I’m after? I managed to find the same feature in RevoScan 6 via the play/keyframes button next to the “Raw Data” in the current scan area.

I tried removing the Blu-tack and base from the point cloud in the key frames editor, but I didn’t see any noticeable difference when fusing, unfortunately.

@PUTV

When you use different software always calibrate it , no matter new scanner or not the IMU is for the giro not the actual calibration , I think that may be the problem .

Following your advice, I calibrated, and interestingly the scanner did fail the accuracy test before calibrating, and it seems to have helped a bit. This is one of my best results (ignoring where some scan spray rubbed off, and the fact that the top of the head wasn’t scanned):

This was done by scanning in feature mode, letting the turntable rotate once, then moving the scanner up and letting it go through another rotation, fused and meshed at 0.05mm. This is actually a very good result, so I have no complaints if it’s the best it gets. (I do need to figure out how to get a scan of the whole thing, though, without gaps at the top. If I try to capture too many angles, I seem to get artefacts.) I am just curious about whether I can squeeze more out of it, especially given what I’ve seen in the showcase areas of the forum.

@PUTV

I was able to scan wings if insect with Mini 2 and get full details of it . So you must be able to do that scan without issues .

I saw your insect and it looks amazing. Is there a good tutorial anywhere on how to get results that good for similar objects? I read that you used single-shot mode, but I get very rough results when I try it. (I don’t have the dual axis turntable, so I have to manually rotate.)

Here is a single-shot scan I did using feature mode, a full rotation in 77 frames in RevoScan 6, fused and meshed at 0.05:

And here is another done with 44 frames for a 360 rotation in RevoScan 5:

(As I mentioned above, single-shot can be extra awkward because Undo can cause all the data to be lost, at least on the Mac version.)

How did you manage to get your insect so clear and perfect?

@PUTV

When you merging scans use the best scan as first in the list to keep the best accuracy , by selecting it as first .

Make sure you clean the both scans before merging and after merging to remove all overlapped points before meshing.

Thanks. I tried all of this but I still haven’t got the hang of merging - I end up with a lot of noise. Here are the fused point clouds for two scans taken from different angles (not using single-shot mode, scanned while the turntable does one rotation, isolation and overlap performed to clean things up):

And here is the resulting merge, meshed:

As you can see, it’s a bit of a mess compared to the un-merged version in my first screenshot.

Thanks again!

[PUTV]

Hi @kayembi

The results definitely improved already !
I scan my insect by manually rotating it very slowly and used single shot mode , not merged scan.

When you merging your scan try to not overlap completely the whole area , merge sides only that did not appear in other scans angles leaving only 20-30% overlap.

To be honest I would not merge that had scan , I would scan it all at once starting from the top , then after each 360 rotation I would move down again and again until I capture the whole object .

You should not have issue with that and that would be best solution here .

The calibration improved things as well , so just practice .

[kayembi]

Thanks again for the help. I’m having some problems with calibration and accuracy getting worse again but will drop an email to support about that if I can’t fix it. I would like to master merging because the head is difficult to capture in one scan, but I will start ask a specific question about that in another thread so as not to go off-topic. I’ll get there in the end.

[PUTV]

Ok no problem , but seriously you could capture the head in one scan.

[Asylum]

@kayembi

Thanks again for the help. I’m having some problems with calibration and accuracy getting worse again but will drop an email to support about that if I can’t fix it. I would like to master merging because the head is difficult to capture in one scan, but I will start ask a specific question about that in another thread so as not to go off-topic. I’ll get there in the end.

You’re very welcome! It sounds like you’re making good progress. Hopefully support can help you get the calibration and accuracy issues sorted. And yes, starting a separate thread for merging makes perfect sense. Capturing the head can definitely be tricky, but with a bit of practice I’m sure you’ll get the hang of it. Good luck, and I’m looking forward to seeing your results!