Please Advise On Scanning

[inSohmniac]

So, I had nothing but issues with my original MetroX and Scanning (please see “Terrible Scans with MetroX and Revoscan 5.6.7 for MetroX”). Eventually I sent it back and ordered another. Due to all the frustration and wasted time trying to get that one to work I put it off and focused on other projects until I felt level-headed enough to try again. It’s been almost a year to get to that point. But here I am, again.

Firstly, I am getting MUCH improved scans compared to my last MetroX scanner. Still, I am having issues. I learned a lot last time, from my own trials and from you guys here on the forum. I have also learned a lot from other posts on here as well as the tutorial videos. And you and they make it all look SOOOO simple! But yet again, it has been anything but. Just to put away all those questions about how and what, here they are:

Lighting: Originally I was told on here that I needed to minimize the light, but I’ve watched the tutorials and watched them get perfect scans in a well lit room, so I don’t know which is actually true, so I tried them all. I have tried no light in a black nighttime room; I have tried light from behind, I have tried light from above; I have tried low light and bright light and no light at all and from angles that wouldn’t interfere with the lens gathering data. Lighting isn’t the issue.

Distance: I always make sure that I use one of the Preview windows in Revo Scan 5.8.6 to perfectly center my objects, and that the Distance Gauge always shows green in the “Excellent” section. I have MUCH better luck using the dual-axis table than by-hand scanning as I can maintain that “Excellent” distance more easily with the scanner being stationary, but trying to hold the scanner and manually scan leads to…

Scanning: …a wild “Pollock” style scan that makes ZERO sense, just like every scan with my prior MetroX scanner, as if the scanner thinks the object is constantly moving rather than stationary, creating an infinite mirrors meets Jackson Pollock on hallucinogens scan. One day someone will have to enlighten me how to overcome this issue because I’ve watched dozens of videos of people scanning manually and getting a perfectly stationary object scan. But, for now, because all my objects fit easily on the Turntable, I’d like to focus on that. I DO set up the Revo Scan software with “Feature Scan”, Auto Turntable, and General object type. I have played around with Color Scanning and not, removing the Base and not, Auto and manual Depth Exposure’s, and setting the turntable up to do a continuous scan or single image scans, also with a single rotation and multiple rotations with the different table angles, but I have the same issues.

Calibration: Yes. Again and again.

Finishing: I have tried the One Click feature and the individual steps manually, one at a time, but nothing comes close to an actual gear. At this point I’d be further ahead using calipers and a paper drawing to build the gears in 3D software, but I just don’t have the patience to learn how to design in 3 dimensions. And please don’t tell me how easy it is - I’ve tried, and much like programming, it makes me want to punch things. Everyone has their specialties, mine just lay elsewhere.

I am attempting to scan multiple gears for a product that is no longer in production and that I’d like to have scans of in case they break so I can 3D print replacements rather than trying to come up with $1400 for the latest model of the product when this one finally bites the dust. Which WILL happen because they use the wrong type of grease in the casing from the factory, which will break down the plastics over time. I replaced the grease, but the molecular damage has already been done. I believe it’s a feature rather than an oversight - built in self-destruction and obsolescence.

The gears are white plastic, but despite this they do not scan well without scanning spray. I have tried with nothing, with marker dots on the larger objects, and with randomly placed bits of adhesive putty around the gears prior to scanning to create non-uniformity. Even with the best scanning using the spray, I get endless “Not enough data points…” and “…More Feature Rich…” errors, especially after having scanned 360-degrees around, despite there being few fully-blue scanned areas after a full revolution of scans. I tried the random shaped and placed putty bits to offset the geometric shapes of the gears, and even placed a twisted and sprayed napkin next to it to give the scanner something to do, but it doesn’t help.

As you can see from the screenshot, the entire gear is “Seen” by the scanner, as shown in green, but past that first 360-degree turn the thin side edges turn orange and red. I’ve tried keeping the Rendered Frames in the 30’s, but despite how much green you see, the scan resulting dots are far and few between. Even with over 200 rendered frames and scans from multiple angles on the turntable there are huge gaps on the gear and tons of floating “Globs”. There are some identifiable divots on the flat sides, but with each pass the scanner puts those few divots all over the results rather than in the 3 places they actually are. Again, like it thinks the object is randomly moving all over the place rather than everything being stationary and only turning 10mm at a time between “Shots”.

So, please, for the Love of Pete, can someone explain why I am having such a hard time getting the results that everyone else makes look easy? $900 is a pricey bit for something that twice now hasn’t accomplished what it’s designed to do, so either I’m entirely missing something that isn’t covered anywhere yet, or all the reviews and videos have been faked somehow. I’m guessing that latter isn’t it or I’d have heard about it by now.

Here are a couple screenshots showing an actual scan on the turntable, and the resultant gear with weird additions and floating globs. Hopefully it shows what you need to know if my explanations aren’t clear.

Thanks for any help!

[ivan]

Hi

First of all, kudos to you for being so persistent and not giving up, because MetroX is a great tool for what you need.!

Second: I’ll be completely honest: you’ve made some fundamental mistakes. You were on the right track, but then you took a wrong turn. But don’t worry, we’re here to help.

My procedure actually doesn’t different from this official tutorial:

Revopoint MetroX 3D Scanner Tutorial: Getting to Know Revo Scan 5 MetroX

The first decision you need to make: Marker Mode vs. Feature Mode. In this case, Marker Mode actually makes a lot more sense. Why? Gears consist of repeating patterns, and for that reason alone, you’ll end up with a “Pollock-style” mess Feature Mode, as explained, is intended for objects with many features and distinguishable characteristics.

So do yourself a favor and try marker mode first. UNLESS you place differently shaped objects next to or below your main object. If you do that—for example, with lumps of putty, as you tried — you’ll need to make those pieces look very different - aka easily distiguishable - from each other. Also, remember to adjust the scanning distance. People too often use the turntable with Feature Mode and place it close to a stationary background. However, this is also taken into account during tracking and messes up the tracking even when object itself is feature rich. That’s Unless you reduce the scanning distance slider so that the background isn’t getting scanned. I often put the turntable near the Edge of a table so there is greater distance to background stuff.

But back to marker mode: since the scanner needs to see all 5 markers at the same time, depending on the scanning angle, this requirement might not always be met. That’s why it’s a good idea to have these marker aids/jigs. So if you have a 3D printer or access to one, print these out. Of course, make sure, they don’t move while scanning.
Here’s an example, but there are countless more on printables.com alone if you search for “scanning markers”
3D Scanning Marker by firstgizmo | Download free STL model | Printables.com.

White object and laser-mode: indoor ambient lightning shouldn’t be much of problem. Proper brightness and exposure settings matter a lot, many users go too high - yes, one can overdo it - and the object doesn’t get scan anymore, @PUTV explained it e.g. great here

Metro Y Pro: Poor cross line scanning performance

The settings are based on the material surface , common sense in 3D scanning objects that are made from materials that absorbs in this case blue light needs higher laser settings and slightly higher exposure . Example , White objects Depth 1 Laser Brightness 5. Gray objects Depth 2 Laser brightness 7 And how darker it gets how higher the settings . Remember red color is nothing else but middle gray so it will require much higher settings . Black plastic is made from oil , oil absorbi…

Scanning distance with laser mode: Distance indicator shows the distance Level (poor-excewllet) for all parts that are being picked, ad depending the scanning scneario you see it indicating more for something more far away than the main object itself. That’s why while moving the scanner at same distance I check in the preview if more and more of object itself or the parts of it that I want to scan appear scanned, otherwise I back up with a scanner a Little bit. Maintaining the optimal distance aka on the same distance “plane” i move the metrox in circle motion and adjust this circle motion “plane” when changing scanning angle.

3. Mostly you Need about 10000 Frames for each individual scan. Mostly you Need 3 scans for proper merging after.

4. Postprocessing: One-Click SOMETIMES works great, but often it doesn use optimal settings and scans look worse than they would using Manual steps. That’s why I almost never use it for optimal results and Manual processing is very easy. Those few extra clicks can make a difference betweeen okeyish scan and great scan

5. Fusion: Software proposes optimale point-distance for fusing automatically, depending on the Quality of your scan. Normally the more frames you scanned, the lower point distance (“Resolution”) it proposes.

6. after Fusion delete unwanted parts of the fused scan using Isolation tab and selection Tools.

7. Move to overlay: for proper cleaning this can take some clicks, because you should adjust the slider for different distances, at which it detects overlaying point to delete, not only the first distance it proposes.

8. “mesh” → ALWAYS use the SAME setting for grid size (=point distance) you did for the fusion! If it doesn’t let you choose exactly the same value , use the next higher value. When using lower value the software hat to calculate for artificial points, which can make the surface more noisy.

Notice this : Learn the Difference: Point Distance (Resolution) vs. Accuracy

So, please try these steps and let us know, if you get better results.

[Xile6]

That was a long read.

Use laser cross or parallel line mode. Place markers around the small gears and manually hand scan it.

Auto turnable mode will work, but you need to use maker mode since its a symmetrical item.

As for lighting, if its not to strong of UV or on the blue side, then it doesnt matter. The problem you are most likly having is those gears are reflecting the light which is why you have to use spray. Again using laser mode you would not have this problem.

[ivan]

I missed the last one:

[Xile6]

Lol. Was talking about the OP.

Your has some good information in it. Just felt like the original dragged on for a little bit. Probably because it see it happen often.

The cant get anything to scan but tried XYZ and still no luck.

[ivan]

OP was a bit longer,indeed, but it was the details t hatmade me understand better why despite- IMHO - very clear manual and tutorials sometimes still some mistakes are being made. I hope we could free another user from the frustrations

[PUTV]

Hi @ivan thank you , I have nothing else to add to it .

[inSohmniac]

First, Thank you Ivan! That was a very helpful response. I am beginning my 3D printed geometric blocks right after this.

As to long reads, well, yeah, I can be wordy. I don’t physically talk much, but after 20-years in I.T. I learned it was much better to be wordy than not. I worked over those years with a breadth of people, starting out doing Windows 95 phone support then being the entire I.T. department for a lumber brokerage, then worked for an environmental engineering firm, and finally for an electronics manufacturing company of 5,000 employees replete with scientists and hardcore engineers who daily worked with things on the microscopic level that blew my mind. And I learned a very simple truth - those people could function easily within wildly complex applications and never blink, but tell them to click the Start button on a Windows machine and they’re entirely lost and shortly showing their temper. People literally know next to nothing about the technology they use daily. In fact, I’ve sat through enough meetings to know that most people believe tech runs on magic and not logic, which is SO bizarre to me given how deeply embedded it is in most peoples lives. So, when I write anything on any tech forum I am completely aware that a large swathe of people will pass on reading my posts because of the length, but I also know that somebody will have the same or similar issue, and outlining everything I’ve done to that point to try and resolve my issue before asking for help may help someone else recognize the same steps or steps they missed, as well as defining what I’ve already tried to those willing to help lead me from there. We had a self-help portal for I.T. related issues at the last 2 of those employments that we asked people to look through before contacting the Help Desk in case their issue was common and easy, and I was the person the other tech team members would send their articles to for proofreading. And every single time those “Articles” started in the middle, as if the end user isn’t as mystified by tech as they actually are, and then leave the “Article” with no conclusion. Just “Turn this off/on and close/restart”. WHERE is “This”? Restart WHAT? The average end-user needs every single step, every click, every menu, submenu, context menu, the number of items in a menu and where that one listing sits in that order, and on and on. WITH pictures, or they call the Help Desk anyway. It’s like Linux. Linux is this irritating catch 22. I wanted to host a website locally with an AI and a database, and after research, Linux was my best solution. But my Linux is basic. It’s not as widely used in most companies as people tend to think. So I set out to learn. EVERYTHING out there starts in the middle with no conclusion. Ask a question on a forum and someone gives a cryptic, single sentence answer, assuming you know. But if you DON’T know, no one will talk to you. NO ONE. It’s a weird boys club, and you have to know the handshake, but no one will ever TEACH you the handshake. So to know Linux you have to learn it yourself through trial and error and hair pulling and gritted teeth screaming at your impassive monitor. So, yeah, I’m wordy. But I USUALLY cover everything.

With that said, however, I had tried so many things I kind of glossed over some others.

When I mentioned rendered frames being in the 30’s or under 200, it wasn’t that I tried to stop there - the app simply wouldn’t process any more, and those were the average maxes I could get depending on what I tried. I did play with the distance settings bar, but had the same issues. I moved the scanner closer and further and every 10mm in between on the other end of distancing, to no avail. I tried a matte black poster board for the background, and then left the background alone but as far from other background objects as I can get (my shop is 12’x12’ and is crammed with tons of tools, a large C02 laser, a diode laser, 2 fiber lasers, a CNC, a large volume 3D printer, and lots of consumable materials for each, from print spools to wood to slate to acrylics to metal and more, so “Distance” is relative). I tried Feature and Marker scans, with and without the turntable, utilizing the different blue-light output modes and on the tripod and manually scanning (hence the infinite mirrors, Pollok, hallucinogenic trip outputs). I’ve tried it all. One issue, however, is that many of my gears are very small, so the marker dots don’t fit. AND, I HAD watched that Revopoint tutorial video you linked multiple times before last night, as well as all the others. And I watched it again last night. But I watched it on my Roku TV, so in YouTube I had to search for that exact video title. Which had the fortunate side effect of pulling up many similarly named videos.

The tutorial videos from Revopoint are good starting points, and they cover most everything, but only surface level in most places. And in some areas they kind of do that thing I try to overcome by being wordy - they start in the middle and don’t offer much of a conclusion on some very important topics. No offense meant by that. Unfortunately it’s a very common theme that when someone knows a thing inside and out they take for granted that many/most people know NOTHING of that thing and rely an detailed information to get us through the often many features offered by that tech. Like the Linux world (I still laugh when I see the endless videos in my YouTube feed with titles like “The easiest alternative to Windows” or “So much easier than Windows”, because every person I suggest try Linux on a bootable thumb drive winds up yelling at me with phrases like "Easier my @$!). But last night, after watching that video again for the hundredth time, I looked at all the other videos from my search and landed on this one:

Metrox 3D Scanning - Marker and Feature Tracking

I look at my gears and I think how complex they look. But KenDoes3D really explains it well. And now I honestly understand WHY I’m having these issues, about scanning and landmarks and creating referenceable points in as many places as possible so the scanner can hold that object your scanning in one fixed place. Seriously, anyone having similar issues should watch this video - it’s laid out well on an otherwise kind of glossed over area of knowledge elsewhere.

So, now I think I honestly understand better. You guys covered this stuff, too, and thank you. But SEEING it helped it truly to stick.

And now I must away! I have some 3D printing to do and some reflective marker dots to apply randomly across those prints so that I might help create a fixed point in space for my scanner to hold onto…

Thanks so much for the lengthy response. I honestly appreciate that breadth of knowledge and the willingness to share it. Should this somehow fail to resolve my problem or creates new ones, I shall return. But unless or until then, thank you once more.

P.S. Sooooo… When the scanner is “Seeing” the object to be scanned and giving it a representation in green, an incredibly DETAILED representation, why isn’t that data incorporated into the scan? The scanner clearly “See’s” it, so can’t that green data be used as an additional level of the scan, filling in where the little blue dots failed to recognize? I know there’s an answer you’re going to provide, but I also assume that with some additional programming that “Green” data COULD be used as part of the processing information, and could likely be toggled on and off. It would be incredibly useful for those items that don’t have a ton of referenceable data without adding all the other geometric shapes. Maybe I’M the delusional one this time seeing magic where only logic lay and the 2 are incompatible, but I kinda feel not. Anyway, I hope someone will help hip me to the truth of that question, or, less likely, have an aHA! moment and change the landscape of 3D scanning. Wouldn’t THAT be cool?