Better Resolution in a scan

The Ein Scan website says that resolution down to 0.1mm are possible. Well .. I donno about that. I believe that Nerv reported scans picking up embossing 0.2mm in height.

So. Trying to scan this figure. It is approximately 4 in. tall. I sprayed it with a white primer and used a dark wash to try and see if that would help. I tried five scans. One with the figure on the turntable .. and four more with a block under the model so it tilted forward .. with all 4-axises raised and scanned (titled forward, backward, left and right). This was all done in hi-res.

Attachment 7210

Here's a close-up of the front of the figure to show what is being scanned.
Attachment 7211

and here is the result. Really soft.


Attachment 7212

sooo .. suggestions to increase the detail? This was scanned with the overhead lights off so room was fairly dark .. just some light from windows ... not too much of that.
I

eTraxx,

i would try this without the turntable, as it looks like the auto mesh is ruining the quality. What you have to remember the auto align works best when it can relate to the previous shots, and as I've discovered with objects that have similar features, the software gets confused and fudges the surfaces and the most noticeable effect is the resolution looks poor. If you do a Free-Scan, the resolution will improve, but it'll take longer.

My 2¢
Mike

you should be able to get better resolution,

one thing you did not mention, i think, is the settings you use at the begining, where it says, describe your object, at the bottom you can choose the level of details of the final mesh, did you choose the highest setting ?

Also, i presume you don't "simplify" before saving, right ?

Mike. Ok. Trying that now. Couple of questions then ..

1) I have the model on a lazy-susan (for cakes), centered. After each scan and having hit "Next" I am given a choice to "Continue" or "Align". The pics of the little bear shows that if the object is being scanned while rotating you select the "Continue" which envokes the auto-alignment. If the objects scanned is offset then you "Align" manually ..

Are you then manually aligning each scan to the previous? I did glue a small block of wood down on the base so that would be easier.

2) I moved the model back and forth until the cross-hair was sharp and crisp. Is this cross-hair just to center the scan or is there a mininum distance from the scanner head to the object (cross-hair focused?)

3) I used a flat-white primer over the model. Is there any concensus as to what scans better? How about surface texture .. would a slight texturing of the surface help with the scan?

eTraxx,

Just press the Continue and let the program auto align. What you'll notice is that the images you see in the L & R camera views what to have some area of reference from the last scan position. Don't be scared to use a combo of scanner moves and turntable moves to create a continuous scan. I've found if I want good surface texture you need to maintain a sharp focus, which is keeping my scanner/projector about 430mm from the surface I am scanning. Ihave used manual align, but to be honest if the self align doesn't work well enough, I Delete the model and move the object or scanner a little less from the previous position, and rescan by pressing Continue.

Hope this makes sense, but Nerv & Hugues have more experience than me.

Mike

It could also be that the dark wash you added creates holes or problems for the scan, and the software fills the holes with a mesh to make it a closed surface ?
as far as I know lighter or darker colors can create different scan results.

if you scan a white surface with a patch of black, you will see that the black patch will stick out or be lower then the surface itself.
I wonder of a certain grey is better for scanning, then plain white ?

A structured light scanner looks at the deformation of the projected patterns, not at dark and light spots :)

Quote:

---

Originally Posted by This

I wonder of a certain grey is better for scanning, then plain white ?

---

I have used a matt red for some test objects on the turntable, and now also fill holes with Plasticene or masking tape, but these are for bigger objects.
I am purely concerned with reverse engineering, so reducing the depth of the holes is beneficial for a one pass scan on the turntable, but I have less issues when using the Free Scan mode.
I normally do a test scan to see how the object is meshing, and then change the strategy to overcome poor areas.

Interesting comment about white surface & black areas affecting the scan interpolation, as not tested this, but I'll watch for this now. :)

Mike

This,

just noticed this, it's a Magic marker 93.xx scrawl in black pen. Notice how it's raised in the scan of this piston crown. What you might not realise the alignment has got confused and rotated the image 180º, and that's why the different sized valve cutouts are so poorly resolved.

Attachment 7243

Mike

Hi Mike :

The result above you posted was scanned in the free scan model , right ? if its that so , please don't , the free scan model is just for the objects which have many features , when you want to scan a flat or less features object, I recommend you chose the turntable scan model to do it .

Eric

Hi again Eric,

No I only use the turntable at the moment to do back to back tests. I have a collection of test objects, and it's when you do more than one pass, the amalgamation of the previous scans causes the problem.

Mike

Mike

So you did multi-scan in the turntable model , like free scan model , if the object doesn't have enough features , the auto-align will be hard to succeed .

Eric

Yes, because the shape of the object, and the camera angle using the provided stand, does not allow it to see enough of the object, and there are areas where scan information is missing.
If you do a second pass, by rotating the object to give the camera a better view, you have to be careful, not to move it too much, as it can only reference with respect to what it saw previously using its scanned data.
I thought it would have realised, the valve cutouts were a different shape, and maybe picked up on the "93.xx mm writing" on the dome, and aligned this, but I think looks for larger feature variations and aligns with this??

Mike