Close



Results 1 to 7 of 7

Thread: How to Align

  1. #1
    Student
    Join Date
    Jul 2015
    Location
    Calif.
    Posts
    26

    How to Align

    Hi guys,
    How do you us the align when 2 scans do not line up????? I have done the points. but what do you do after that. would really help to get a manual for all of this. I have gotten a coupe of good scans so far.
    This program really needs a back button and a cancel button.
    Thanks for any help.

  2. #2
    Student
    Join Date
    Jul 2015
    Location
    Calif.
    Posts
    26
    Has anyone gotten this to work?

  3. #3
    Engineer-in-Training Hugues's Avatar
    Join Date
    May 2014
    Location
    Switzerland
    Posts
    219
    You have to select 3 points, after that the system will immediately try to align.

    But for me so far the proposed alignment is way off, never worked. I contacted Shining but no news so far.

  4. #4
    Student
    Join Date
    Jul 2015
    Location
    Chestermere, Alberta, Canada
    Posts
    34
    Add Nerv on Facebook
    The manual alignment will only work well if you haven't scanned too much data, and if the points you select are really close to being the same points on the part. The software will try to match the clouds as long as you get them close enough.

    Here is a hint in regards to scanning larger objects: Pan your scanner rather than rotate it around the part if you are trying to get aligned shots of a side of your part. This way the overlapping area will be from the same angle and more of the data will be the same. Once you have the entire side scanned this way, then turn the scanner to take the side from a new angle. This greatly reduces the chances of misalignment.

    In most cases if the scanned data isn't automatically aligning, it is better to take another scan with more overlap than to try to align the one you took. The EinScan-S software is pretty good at finding where your scanned data should go, though there are times I still resort to alignment manually.

    Overall, the key to getting good scans is to try not to get too much overlapping data as a whole. I know that sounds counter-intuitive, because alignment requires overlapping data, but the more scans you have, the more you will have errors stacking up, not too mention as you get a bunch of data you will start to risk crashing. I have built reference surfaces covered in black with only a few distinct registration features that are light colored so when I scan a large part it can find those features and lock in alignment so I don't have to worry about having massive amounts of scan data to have accurate alignment. I'm scanning parts well beyond the size and complexity that Shining 3D intended (They have told me as much as few times) so I have had to become creative. This is one of the main reasons I think higher quality scanners use registration markers.

    I am working with them to improve the software to tackle larger and more complex objects. They are working hard toward making this software better for everyone so have faith that buying this scanner is a good investment, it is getting the software improved as we go along.

    Dave @ Nerv

  5. #5
    Engineer-in-Training Hugues's Avatar
    Join Date
    May 2014
    Location
    Switzerland
    Posts
    219
    Quote Originally Posted by Nerv View Post
    ... I have built reference surfaces covered in black with only a few distinct registration features that are light colored so when I scan a large part it can find those features and lock in alignment ....

    Dave @ Nerv
    Thanks for the tips Dave, much appreciated.

    Can you share a picture or drawing of these reference surfaces ? Which material do you use ? How thick ? How do you stick them ?

  6. #6
    Student
    Join Date
    Jul 2015
    Location
    Chestermere, Alberta, Canada
    Posts
    34
    Add Nerv on Facebook
    For now I use 9/16" diameter precision bearings (pinball size, can be purchased cheap from Lee Valley) that I sandblast and primer flat grey. Then, I get small 5mm diameter neodymium magnets from Walmart, as they are the ones found in the little fridge magnets they sell. I just crush the plastic in a vise to get the magnets out. This is the cheapest way to get a whole bunch of magnets I've found.

    Precision spheres are great because as you may have found, you can't scan bolt holes. If you place a sphere in the hole, not only will you get a nice clean scan, but with that sphere you can find the exact center of that hole every time. The more accurate the sphere, the more accurate the hole.

    In this case, precision bearings are about as accurate as you are going to get, and they work great for this scanner. I made a whole bunch and they really show when the scanner is accurate and when it isn't. As it get far in the distance or close to the sides you can tell by the spheres if the data is right. Luckily with other software I've been playing with it can automatically tell you how much the points are deviating and you can start to make a more informed decision about what you want to keep in your scans. Long term I think it would be nice if we could get the Einscan software to detect this for you so the scans are highly accurate everywhere.

    We'll see. As for the reference base, I would flip over a mirror or take a sheet of glass and paint it flat black. Then maybe paint some flat grey stripes on it or some sort of pattern, you want the pattern to be non repeating and to have probably less than 10% coverage. The more grey you have, you more points will be generated each scan that hits the surface. Since the surface is flat and you have the spheres for positional reference, you don't need much grey at all on the flat plate.

    I hope this helps everyone.

    Dave @ Nerv

  7. #7
    Student
    Join Date
    Jul 2015
    Location
    Chestermere, Alberta, Canada
    Posts
    34
    Add Nerv on Facebook
    After many more hours scanning I have found a really good alignment tool is a small block of wood, wrapped mostly and randomly in black vinyl electrical tape. This keeps the captured scanned data small from the block, but the grains of the wood are random and the scanner does pick them up so the alignment is virtually exact each time. Another benefit of the wood is the grain is a very subtle surface deviation that can be picked up with a very small included angle from the surface of the block. This way when you take a scan from the opposite side of the block, it will still pickup the same grains it did before. By contrast, spheres and any other objects that are more pronounced (unless the software is designed to derive an alignment center-point from the sphere - like high end metrology scanning software programs do) will not align if you take another scan from 180 degrees of the previous scan... there just are no points that overlap. The near flat textured wood surface is an excellent alternative, at least until the software has sphere alignment in it.

    Regards,

    Dave @ Nerv

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •