Sorry @c_v2 but it is no more available to download . The latest Revo Scan software has fixed already the point cloud alignment issues.
Hi PopUpTheVolume,
First of all, my appreciation for your very interesting posts.
Are polyhedras with markers printed by you or have you bought them? And if yes, where?
Thanks in advance for your answer
1 Like
can you post the whole process of scanning a person because the way i see it this is impossible with pop2, it just loses tracking ALL THE TIME even for just going around a head
Found these 3D printable scanning Pyramid Markers
2 Likes
Hi @PUTV and thank you for all of the energy you put into sharing your wisdom.
If you can spare the time, I have a few questions:
My current setup:
I have a 50³cm³ photo Box with a 30cm black turntable (tried to find things similar to what you described) which on the slowest speed settings will take 60 seconds for one revolution.
The Photo Box has bright LED’s mounted at the top and a white piece of cloth that can be attached below the LED to act as diffusor.
The backdrop and turntable inside the box are black, the other ‘walls’ are a silvery reflective diffusive (i.e. non-smooth) surface (which i guess is typical for these boxxes).
Previous and current projects
I have scanned non-organic items (tools, household-items) for CAD-modeling and 3D printing before, using scan-spray, since the texture is irellevant for this application.
I am now trying to scan Organic things (fruits, vegetables) with texture, primarily in order to improve my skill with the device.
Problems / Questions:
- Lighting
Despite the Diffusor, the lighting is ver uneven - the resulting RGB values clearly show very strong variation in brightness across the geometry. the data from top-parts the the surface is close to orthogonal to the light coming from the led lightsource is a lot brighter than surfaces that might be almost facing downwards, towards the black turntable, which are often only hit very little with indirect light.
I tried putting in an additional light-blocking sheet between the diffusor-cloth and the LED in an attempt to decrease the amount of light coming from the top, but that barely did anything at all for me.
Could it make sense to integrate LED-strips at the bottom/sides of the Photo-box?
How do you get even lighting with your setup?
Or do you somehow instead fix everything digitally?
- I find it very hard to scan in one go:
When I pause and rotate things around, RevoScan will often be unable to recognize/match the new orientation properly. This hat worked better (although I still needed several attempts for any scan then) with the previous less organic objects on which I had applied the spray.
The only solution I have right now is the CloudCompare workflow you have so wonderfully described in your Youtube tutorials.
Do you still use CloudCompare in your workflow, or do you these days more often just use RevoScan?
I have seen you perform complete scans where you just move the camera while scanning in one go to get all angles - I feel like I’m really having trouble pulling that off, I don’t know if my hands are steady enough. Also If I scan something with a more spherical shape like many of the objects I am trying right now (Orange, Onions, potatos, apples, …) it is virtually impossible to get the down-facing surface without reorienting the object at least once.
- I often have trouble with the accuracy.
In case you still do multi-part scans that you merge in CC later, how many frames do you capture?
With the turntable at 60s/rev clearly its more than 300 frames - I usually end up somewhere in the range of 520-650 frames.
But sometimes the “seam” for lack of a better term - I mean the part where the points from the last frames overlap with the points from the first frame - will be inacurate and force me to repeat the whole scan.
I will make some video or screenshots if that helps.
Thanks again so much for your expertise so far, without them I would hardly know how to improve the results I get.
1 Like
Thanks Davide , the best would be dodecagon but could not find it in black silicone so more of a mix of dodecagon with icosahedron , got from Amazon .
I put on my own reflective markers I bought also online on Amazon .
@Miamijerry you just could print black dodecagon at 20mm diameter , would be the best , pyramids taking too many space , you want to have as many faces visible from any angles , that is the point of using dodecagon .
1 Like
@c_v2 I will thinking about , I made many scans already so if this was impossible you would heard about that from me already .
Maybe I do some PDF Tutorial when I have little free time on a hand
2 Likes
Hi Julian , thanks !
yes the top LEDs did not worked well so what I did is placed it in front top and bottom , I also purchased some additional light strips to stick them on top for the little extra , make sure you have a cold white light color for the straps .
You can use white electric tape and tape down the LEDs so they produce more diffuse light , this way you will have bottom and top equal in lighting and the extra LED you put on top and sides that will give you additional backlighting for the edges but do not put too much in the back or you will get Fresnel effect and we do not want that . The additional LED strips are just for little extras so not full power , best way to have control over it manually to adjust it to your needs . All you should care is the 180 degree FOV so only frontal and sides, the big LEDs you got with your light box should go in front .
When you scan with color textures or data , don’t put the object directly on a black surface , use something to lift it up like little transparent plastic you cut off from a water bottle
With organic models the trouble starting , many of them needs marker mode , that why I made for myself the new markers on magnets and dodecagon , so you can move freely from top to bottom .
But that don’t works with any object , so sometimes you just need to scan the angles and merge , especially when you are using marker mode , not recommend to change the position as matter mode do not care about the object features , it only track the markers .
I do almost always until I have a very easy object to scan then can run it all in one session .\You see when you scan vegetables or any kind of materials , you will need to use marker mode , make sure the object is as much as possible visible in the depth camera preview , you can use a nail or a metal pin , and pin the vegetables on it , so you actually scan almost everything at once , and if you scan only 2 sides separately you will need to use color mode and later use the color for references as your marker points for merge , some objects are soo smooth it is impossible to find a marker point , but color per vertex data can helps you , better than textures as textures shifting on each object one way or another , color per vertex data is more accurate .
yes I do scan around 600+ frames with 60 s/rev and 300 for 30s/rev
if you see strange things after fusing of your point cloud you can’t go forward and need to be redone , the mesh can be only as good as your point cloud , CC can fix it sometimes so you can try before merging .
I was scanning the other day my DSLR , around 11 sides , then when I started to merge it, I found one important side with double element that actually ruined everything and have to re scan this side again , CC was not able to fix it , so pay attention , do not fix holes in your partial scans . For serious scans I always do many sides and export my Fused partial point clouds only , the whole process can take sometimes 2 hours from scanning to final mesh and if texturing then even longer .
I hope it answered some of the questions , and thanks for the support Julian ! if you like to see some specific topic for a next video, let me know .
And if you run in troubles next time , drop in with some screenshots , it would be easier to figure out what is going on .
2 Likes
Omg thank you so much - it’s so simple yet makes so much sense ^^ I want to try that immidiately (don’t have additional led strips for sides yet) - the only issue is: the LEDs integrated in the box are stapled to the top of the box - I feel like I would want to not remove them if possible. I will instead just rotate the box, only issue then is that the velcro mount for the backdrop is not in the right spot anymore, but that I think can be fixxed.
Thanks so much again for the very elaborate post and all the input.
1 Like
My top LEDs came with magnets so you can attach them everywhere on the metal frame , in your case you need to get some LED strips and put them in front and sides for main light source , I got the one with flexible connectors so they bend nicely on return .
https://www.amazon.com/gp/product/B08TBBZY65/ref=ppx_yo_dt_b_search_asin_title?ie=UTF8&th=1
you can put 2 on left middle, 2 on top -2 below front and 2 on middle right
something like that
and the top you have already will be the ambient source for the top and back sides , then you regulate them for good balance for no shadows, the white electrical tape over the LED strips produce nice soft box light without sharp reflections or shadows.
I hope this helps to get your an idea.
Putting the LED strips the original way on top and sides that I mentioned before do not need the extra connectors as the one that came with was long enough .
3 Likes
Small tip for today ; When you are about to get yourself AESUB 3D scanner spray , go for the Orange , because it is much finer than the Blue one , especially for small scans , it creates much flatter surface so less details are lost and actually will last much longer, and do not smudge that easy .
5 Likes
OK, this is the closest I found;
5 Likes
@Miamijerry that’s perfect Gerald ! now if you print it in black filament it would be perfect , you will get 3 markers from any angle visible . Before you print make sure each face fits 10mm marker , or you can use simple 6mm white sticker dots , but reflective works so much better.
Then you can print couple in white matte filament and you have some support for small scans in Feature mode
2 Likes
Wouldn’t it be useful to deboss some designs into each face so that the scanner could work in Feature mode?
Why? if you have 10 of them already , they do not need additional details , they are the details
@JeffLindstrom look a perfect capture and I did 2 angles in one session moving from side captured fused, then from top captured fused .
Feature Mode with White
No Mandala effect 
I am going to tests it with the gray one I have left and cut them in half for use with even smaller objects and see how it works . The magic here are the faces on a different angle
And without
2 Likes
I had some spare Chromalabel white dots stickers that fits exactly in the empty spaces so now 4-5 markers are visible on an angle . I prefer the reflective as they works better on an angle but could not get smaller size than 10mm/6mm but anyway it works
There are always at least 2 markers visible at the same time to keep tracking so no blind spots
in my case 4 right now , and since it change constantly the pattern is always unique in space .
So I put now 4 only for test and it worked as long 3 of them are visible by the angle of POP2
The point of this is to capture as less as possible data for faster processing and less cleaning after.
Also reminder , you can’t flip the object to capture another angle when using Marker mode and hoping it align together , but you can change the angle of the POP2 to capture different side while your model and markers are still in the same space and not moved from it’s locations.
Marker modes are for scanning individual sides to merge them together later , so not design to capture the whole model in one session , unless you actually moving the POP2 around capturing the object but in most cases it would impossible to get everything as one side always facing the bottom on the turntable .
3 Likes
With individual markings on each face, you don’t need 10 at once. That takes up a lot of space and markings on the facets will allow closer scans, right?
Regarding Feature mode
Faces with extra additional features do not matters in this case since it has already many faces it don’t need anything extra .
And regarding Marker mode
As I posted above , less is needed and I scanned 19mm objects with it and objects without features .
It all also depends what angle you use and how close you are to the object means narrow view at 15cm , I tested it between 15cm to 20cm +
When you scan from above you need less , the most needed was when scanned from front at 90 degree , so the best angle is at 45 degree as usual .