How to Proper Meshing Your PointCloud

Can you please elaborate a bit more on this? How does this guidance based on the fusing setting works?

It is very simple

Image you scanned a cube , a volume of a cube needs only 8 points to hold the perfect volume , now some users may thinking meshing it at 3 millions points will magically make the scan better … but all they do is just induce artificial points that really don’t support the volume aka the main pointcloud cage and are truly not nesesery. The mesh can’t be better than the point cloud already is .

There is actually not much to explain , the software provide you exactly with the specific Grid information so you know how far you should go with the meshing settings remembering of course your point cloud settings to guide you .

The Grid guide settings is very important , if you mesh the pointcloud below it’s original Pointcloud level , the accuracy will change dramatically and you lose greatly on your volume , having it exactly will produce 100% correct accuracy as the original Pointcloud , having it at higher setting will only induce artificial points between the main cage points and add nothing to the volume but noises and bumps and in some case even lost of details you looking for in first place .

When I am on my PC later I will upload for you some examples on this subject showing you what is happening if too higher settings are used , the lost of details and bad surface .

It will be probably much clearer for you to see what is happening , after all we talk about 3D modeling here not an English class what is BTW one of my 5 languages I speak so sometimes it can get difficult to explain in words the exact vision of the situation . For the same reason why most people that write 3D softwares can’t create any 3D models themselves …

One more point here that the Revo Scan software is aware of every step you go through processing , the algorithms calculate in background the proper fusing settings based on your RAW scan as long you clean out the raw data to some degrees before fusing , it also calculate quickly the proper meshing settings based on the fusing and pitch point distance between points , so very smart … no rocket science here , anybody can do that …

But as I said if you know exactly what you doing , you will improve your results .

Same with Cloud Compare software that precalculate the minimal-maxinal meshing level that your pointcloud will profit from without to thinking to much about … or if you don’t know exactly the true pitch point settings while fusing of thr original pointcloud.

Seems nothing but so important , if not the most important knowledge while processing your scans .

I will be back at you shortly

2 Likes

Thank you so much! I believe i understood your explanation perfectly! :slight_smile: And yes if you can share those videos i believe it will be even clearer. Is this the reason why you said the current “one step does it all” option is not working properly? From your explanation it seems that it is possible for “a” software (at least CC) to estimated precisely what settings to use…right? Cant revopoint software do the same atm?
I thought the increased mesh grid would work differently. Example, you scan a donut. The circle of it because of lack of details in the capture (poor scanner maybe) resulted in a pointcloud that looks like a hexagon (just to be easy to understand). And by increasing the mesh grid you would add points to that solid surfaces that would ultimately round up the surface, resulting in something closer to a circle….I now see thats not how it works! The correctness of the surface being captured is more related to scanner hardware capability such as precision and accuracy!

It’s working very good , the reason I don’t like to use it sometimes is :

  1. It calculate the settings based on Raw scan data , if your Raw scan data have more than just your scanned object , fir example you scanned a part of a floor, or some background by accident , the calculations will be based on the total volume what can decrease the overall accuracy … if you scan was cleaned under RAW settings and parts removed it may produce better results .

  2. It do not close holes , so you have to go back again and fix it wasting time .

Yes it do that why you have Grid setting under Meshing … it tells you what is the best meshing setting for your current cloud point … in case you imported pointcloud and do not remember the fusing settings .

Sadly that is not what is happening, the meshing function in Revo Scan do not include smoothing , if it did you would lose the original volume and the accuracy , higher level meshing than suggested Grid will not smooth out volume . It will add artificial points … and that is all .

Exactly, it only add the geometry to the points and only based on the points , it do not add more details but it can remove details if set too low reducing the level of points what is something you don’t want to do .

The Volume accuracy is already build into the hardware of the scanner , it can’t be altered and make things better .
Fused ( organized) pointcloud is the best representation of the scanned volume , any processing after will only decrease the accuracy , that include simplifying the pointcloud so please never use this function if you really don’t need , I see so many videos on YT where new users simplified their pointcloud and lost the accuracy but did not realized it thinking that the Scanner’s capabilities was at the finest … well it was user error …

Changing the fusing setting ( pitch point ) aka point distance do affect the final resolution of the final mesh , but it is not an accuracy …
Resolution is the detail of the surface ,not detail of the volume , however if you scanned with 0.02mm accuracy and fused in Advanced mode at 0.2mm it will not change the accuracy , Advanced mode use special algorithms that works completely different than the Standard mode , what I don’t like to use to be honest on organic models with curves , too much to cleaning after and messy with too much data that in 60% is just junk that need to be cleaned …

Advanced mode have great algorithms, and organize the points in a very clean way eliminating the 60% of all the junk data and points from frames that are no more usable in the process .

As always …how cleaner and organized is your point cloud, so will be your mesh .

Thanks again! Learning a lot! This is what voldemort scanner does that i do like. I clean the pointcloud, remove what is clearly offseted and is noise and then fuse it. Something our miraco and revoscan could do also!
Also that one click option…ive shared my thoughts with support before and i believe they are working into sole sort of templates for it. One click A, one click B, etc, where each set of parameters is ment for each type of subject to be scanned…For ne it would be ideal. The software would only benefit from being user friendly for both dummies (that would be me) and pros (that would be you)!

I asked also for the options to separate the process as sometimes you don’t need everything . Hopefully it will be added soon .

Well you not so newbie anymore and definitely not a dummie , you never old enough to learn new tools and that goes for both of us .

Thanks. You are very helfull as always!

Hello ! You mentioned earlier that you would be providing some example. Have you post theme somewhere else, or they are not available yet ?

I would love to see your theory in images.

Thank you in advance
George

Hi George

Sorry I did not , it was a very busy time we had .

BTW it’s not a theory , it is how the software works and how 3D modeling works .

We are in holidays , next week I will post more stuff around about , stay tuned

Thank you for your reply ! You got my attention with this. Curious to see the results of the mesh whit lower quality settings.

It is not a lower quality settings , it is according to the Grid and points , the proper way .

Caint wait to see this !