Hello,
I am looking for some input on how to scan this large car hood, specifically in the middle and what kind of tracking marker objects to use.
I first attempted marker dot tracking. I put dots all over, but the scanner was not picking them up past a distance of about 300-400 mm from the object so i gave up quickly on this method because of how large the hood is and needing to use the scanner at its full distance. General or Body mode seemed about the same. Is this distance limitation normal or did i do something wrong?
So i proceeded with feature tracking and had great results overall but started to struggle with the plain large center of the hood. When the scanner no longer sees a headlight or those louvered vents, it obviously needs something else to locate on so i tried crumpled paper and these small wood cubes. The smaller cubes are 13mm and the larger are 19mm. The scanner did well with them when closer but as i started to scan the center of the hood it would lose tracking (not entirely but would overlap incorrectly) and after about 15 tries, the attached snips are my best result, but it still shows doubles of them in the middle area.
Are cubes not a good tracking aid, and is even the 19mm too small?
Additionally, the doubling effect of the cubes seems to be different depending on the cloud fusion distance. Please see the images for 0.3, 1, and 2mm. The worst is at 0.3mm. Why does the fusion distance matter in this case?
Hi, thanks for your detailed sharing, hope the following suggestions could help.
To scan the large car hood, you can use the marker tracking as well as feature tracking.
(1) When using marker tracking mode, please distribute the markers randomly at intervals of 2–4 cm, and ensure at least 5 markers are visible in each frame during scanning.
(2) When using Feature Tracking Mode, the goal is to provide the scanner with visually distinct, non-repetitive features that help it maintain tracking. You should use objects that have:
Irregular shapes, distinct textures or patterns, different sizes and heights, surfaces that are not too reflective or transparent.
Those small cubes with total geometric similarity may lose tracking. Reference objects at least 30–50 mm in size with the shapes above are recommended.
A smaller cloud fusion distance captures more surface variation, but also more noise, especially if tracking is lost slightly.
And for smaller distances, the software tries to preserve every tiny detail, including noise and misalignments between frames, and then poor quality is displayed.
An appropriate distance for the best balance of smoothness and accuracy would be better.
Thank you for the reply. That makes sense about the totally similar cubes causing possible issues. I will try some objects that are variable in size and shape.
For the marker dot stickers, should the range1 scanner be able to detect them at 1 meter distance ?
Range 1 should have markers placed at not bigger than 8 cm distance from each other in an unique and not symmetrical combination .
The distance between the scanned object and the scanner should be at maximum 80 cm.
If you use Cubes the markets are not needed since you can use feature mode only .
The cubes will works great as long they are placed in none symmetrical combination on top of the surface
If you use just markers , use marker mode only and cubes are not needed .
When using marker mode the features of the object are ignored.
From someone who is also regularly scanning featureless objects with the range 1, I can say generally you are doing pretty good, you have got the right idea using objects as markers.
If I were you I would try to perhaps make a few different scans of the car hood in sections and then use the software to merge the scans together.
Doing it that way gets me the scans that I need. Just make sure you have a decent amount of overlap on your scans while merging them.
Hopefully this helps
Edit
I would also add that I generally get my most accurate scans with the range 1 by keeping the scans as brief as possible and avoiding scanning over the same area more than once. Then with multiple scans merge them together with the software.
Keeping the scans brief and not scanning over the same area multiple times can help to not get the doubling of the blocks, if you are getting doubling of the blocks it could mean that you are losing accuracy.
Just to be sure, have you verified that the scanning distance has been maximized?
I connected my original Range to Revo Scan V5.5.3. It defaulted to 300-800mm for scanning distance but gave me the option of sliding up to 1000mm. I did and more details showed up in the depth and main windows (a couple seconds later).
I think depending on what settings you select it can give upto either 1100mm or 1300mm
But I found that bigger field of view doesn’t always equate to better tracking (even for large objects) For me with my range 1 the sweet spot is around 1000mm scanning distance.
And for some reason for me sometimes selecting “body” scan mode gives me the best tracking even though I am not scanning a body.
The “Body” is just for a reference of an objects not specifically made just for scanning a human body. But an objects that are larger and have slightly less features for tracking.
