Engineering scans - Modception

Hi @lajoosh check the Grid under meshing , if the meshing can’t be set at higher quality than you fused meaning having higher resolution will not produce more details and you reached the maximum according to your fusing settings .

For example you fused s big mesh at 0.20 mm but the Meshing Grid only allows 0.26 maximum this means you can get more details meshing it outside in other software … but that don’t happen often , mostly when using Range in my case .

Thing is that you only can get as maximum details as your fusing point pitch distance …over meshing it will only induce noises and not details as that is technically impossible . Noise sometimes create the sharper effect but in reality it don’t add more accuracy or details .

Sometimes the grid is too small , for that reason try to clean the point cloud in Raw mode and remove everything you don’t want to be fused , it can improve the meshing settings and give you higher values to play with .

I can see slightly improving on your examples but I will have to see the grid setting in RS5 to see if there was not enough level for meshing .

1 Like

Yes that is exactly what happens, I fuse all my scans at 0.2, the most dense point cloud possible, this example gets a ~0.4 grid at quality 6 and if it is a larger scan, it gets even larger grids that I cannot get any smaller because the slider maxes out where ever it decides, even if I try to type in a custom value, so theres no other choice but to export it and do the meshing in something else. I had much wilder cases before, where the difference would be much more obvious. This example is not the most descriptive because the revoscan mesh is also pretty good as it is. fusing and meshing grid settings were maxed out for this differential unit scan, yet meshing it in geomagic gave the result better defined edges and sharper corners. The thing is, I like to work with the high resolution data first and only after I’m done with all the post processing, I do the decimation with a method that really keeps the small details intact in the model while overall the scan loses up to 90% of its polycount but still looking excellent.
Anyway, theres the workaround for this, so personally I’m fine, but I guess it would be nice to have this option right in the revoscan software.


2 Likes

Yes in this case I do the same , you losing 2 levels of quality here fusing it at that settings .

Same here , I prefer to get maximum possible according to pointcloud and simplify it according to my needs .

I like to Simplify my models in Zbrush as I have wide level of settings while doing it like keep the edges, UVs or other important settings .
RS5 Simplifier is not that bad at all , you can still reduce an object from 1 milion to 6OK and still have full details …

3 Likes

I find the dialogue between both of you highly valuable. I’ve pinned this thread to ensure that more users can benefit and learn from the discussion.

3 Likes

It is really great that you guys listen! This small change in the software should be simple enough to implement, I think. I would suggest keeping the slider work as it does now, but right next to it theres a box already where we can type in a value. This value is currently tied to the slider so we cannot set a lower or higher value over what we can already set by the slider.
Instead, let us type in whatever value we want to and have the pointcloud meshed accordingly. (within reasonable limits of course, I would say adjustment between 0.05 and 5mm point distance would be more than enough and would cover from mini scans up to range and miraco scans. very high resolution to really coarse).
This way novice users can still use the slider to get a pointer on what this value should be, a sort of default quality adjustment range. However, those who are more experienced, will get more adjustability by typeing in their own value.

Hi @lajoosh that’s not how it works , each scanner you connect loading it’s own specific profile and settings according to it’s a curacy .
From practice people always slide things to maximum thinking it will deliver them the best results , well it is not this way , so to avoid troubles while processing each profile is based on the scanner capacity.

It don’t only depends on the scanner but also depending of the size of the scan.

So the idea to have all settings available for all is rather not the best choice .

For example I am scanning a coin with 0.02mm accuracy , I fuse it at 0.2mm and mesh it with a perfect grid settings of 0.2mm at 4.3 mesh level and the results are perfection to the last fine details …

Now if you slide the slider to maximum while fusing and meshing , all you would get a ring of noises without any details at all … and that’s the magic …

The one click option is actually calculating very smart the proper settings regarding accuracy, resolution, meshing and the size of the scanner volume , and deliver most accurate results .

Having additional meshing level to match the Grid setting is a must have , this issue was addressed with Range a time ago when I tested it so you have level 7 now .

Maybe it would be good to create a series of scans with multiple scanners to find their best meshing settings according to pointcloud , do the dev.team adjust proper each profile .

2 Likes

Except with the Range it reverts to 6 maximum after you merge. It’s really easy to max out the slider merging multiple Range scans.

Yes but after merging the pointcloud is not more under Range profile settings and it is handled as any other imported pointcloud for the reason the level drops to 6 .
As I mentioned early , it is not always about the maximum settings … you need to pay attention at what settings you fuse as that is your indicator for the right level for meshing …
In most cases very big scans scanned using Range that are merged have not enough meshing level , like for example the size of 2m / 6ft.

I suggested it while testing Range Beta for that reason simple scan meshing changed to level 7

2 Likes

I get it, I do. I scan a lot of large things with the Range in 18fps/Large Object mode and the suggesting settings always max the point distance at 2.0mm and mesh level somewhere > 6. It actually does fairly well scanning the large surfaces, but I get uneven point cloud density because I’m always at the max (remember I’m merging these) and low resolution meshes because I’m always at the max.

Seems an arbitrary distinction made to keep people from making absurdly large meshes for no reason, at the expense of people who actually want to make absurdly large meshes. People are always going to use software and devices in ways the creator didn’t intend, but that’s not always a bad thing.

1 Like

Yes that is true … you can create a mesh that will need more than 32GB to process while doing it on a 16GB RAM machine the end of it is a crash while processing so that’s the reason for .

In your case I understand and that is obvious but in many other cases it just don’t works this way , as technically it is wrong approach to build a mesh .

The meshing after merging scans done using Range needs higher meshing level , I stated it already .

Anyway write under suggestion or under last version of RS5 and let your voice be heard.

Don’t forget it is a Showcase thread of @lajoosh showing off his scans , let’s not kill it with arguments.

Apologies, not arguing. Just trying to support the OPs assertion of expanding the range of the slider(s). :slight_smile:

1 Like

Go for it [Windows & Mac] Revo Scan V5.3.4

As I said, I would leave the slider as it is, for the exact reasons PUTV mentioned, let the devs suggest good settings, it is helpful for everyone. I just think that if theres a reason you need to deviate from the suggested value, there should be a way make that possible.
I can think of two good ways: Let us type in higher and lower values over what the slider sets, or expand the slider like this:
image

Where the dark red area is the overly low quality grid size and the bright red is the grid qulity is higher than normal. Inbetween there is the dev suggested range. There could even be a checkmark switch right here to enable or disable this feature.
The higher quality should let us do mesh the data at the same point distance as the collected pointcloud allows, the lowest quality could be any arbitrary number, let say 50mm. generate your low poly face model instantly! :slight_smile: the point is, let us do stupid things if we want that! Professional softwares lets you do this and so you can gain a deeper understanding on how things work. If I do something stupid, the processing might take ages, or if I do something even more stupid, the software crashes, but it is on me, I will know better next time.
Coming back to quality setting, this was months ago but this happened with me using the range and making large scans… first pic: data meshed in revoscan software at maximum grid guality. second pic: the same pointcloud meshed in geomagic with my experimental settings. you can clearly see the difference…


again, this was months ago, june if I remember correct, I’m not sure how the range is doing with the current software version though.

1 Like

When merging object scanned with Range , you going to lose some points in process since the meshing level is too low . The software don’t take care after merging if it is Range or Mini , it cares only about the volume and based on that it set the meshing level for imported objects , or merged so this have to be adjusted .

Level 5-6-7 really don’t matters , what really matters is how much points you got after fusing , and how much left after meshing …
Because adding artificial points will not make details magically more visible if they was not there in first place .

Pointcloud is your true volume and true resolution , any future processing will only downgrade the accuracy and volume .
If you have a cube with 8 points , meshing it 10 times at higher polygons level will still give you only a cube .

Regarding the slider , it do not works this way for everything and each scanned model need a different settings according to the fusing settings and the amount of the points and volume .

That why you have there a Grid value for a reference . What is very important for 3D modelers .

If I have a scan of 100K points , I want a mesh with 100K vertices not 200K not 400K as I want it to be as closer to the original scan without artifacts and artificial noises what would alter the original accuracy .

@lajoosh Revopoint should add additional meshing level for merged objects especially if it was scanned with Range or Miraco in Far mode , other way we losing precious points and accuracy while meshing merged scans , that is already clear . I discovered this issue while testing Beta Ranger and it was adjusted for the meshing , but not after merging of objects and that still need to be adjusted .

1 Like

Just a quick render of the end result. The E28 differential unit was mated to a BMW E46 rear crossmember / subframe for a custom drivetrain into a rally car, necessary sheetmetal parts are in production now.
The differential unit was scanned with POP3 in feature alignment mode.


10 Likes

Just another quick reverse engineering task, using the pop3 scanner.
This one is to re-model and 3D print a part of a headlight which is unavailable as new OEM part. The original part is burnt and deteriorated.
The surface is PVD chormed, so a thick coat of scanning spray was used. Scan data processed in revoscan 5.







4 Likes

Great job !

Another smaller item scan, this time done with the mini 2. Excellent!
The part is a hydraulic clutch release bearing from Tilton engineering.
The part was used in a passenger car drivetrain assembly design.
Strictly speaking it would not have been necessary to scan the part, as there are good enough drawings of it on their website in case of a general situation. However the bearing was retrofitted to an OEM clutch assembly instead of their standard clutches, so in this case it was better to check fitment, clearences and working stroke length in CAD in order to correctly design the mating adapter parts. In general, a 3D scanned part gives much more confidence than a cad model that has been drawn back based on mechanical drawings which may contain incomplete data.






8 Likes