Scanned a tow bar to create a custom fairing for it. Scan itself was done inside a garage with some direct sunlight coming in through the windows.

Scanning

Scanning was done at 0,75mm target resolution and no scan spray was used, Trackit SR was used wireless. Being wireless I feel a good bit more comfortable having it stand in a crowded space, no danger of somebody tripping over the cables and either ruining my usb ports, the scanner or the scan.
I placed to scanner strategically to ensure the whole tow bar was within the tracking field of view and started scanning in cross lines and auto exposure. Whole scan took roughly 7min with a few times checking for coverage on my tablet (remote desktop to my workstation). Scan process itself was positively unspectacular, just worked (when choosing higher resolutions it hangs sometimes a few seconds. Issue is actively worked on from what I know).

Post-Processing

Processing was done by simply fusing at 0,75mm (software even recommended 0,5mm but I didn't need the higher resolution, guess I scanned a bit too long). Then isolation, overlay detection and meshing at 0,75mm. Quicksurface was used for reducing the scan size and aligning the scan. Also removed some marker spheres I placed on the tow bar initially because I was skeptical it could be scanned from one position, didn't need them but also didn't bother removing them initially.

Results

Results look quite good, captured everything I needed. Only thing that bothers me is that I touched one of the cables while scanning so now it is doubled but not needed anyway in this detail...

Sketchfab

Sketchfab is like printables for 3d scans with a nice integrated viewer in browser and you can also download the scan, just look at the scan yourself.
Reddit sadly blocks the short links to Sketchfab, you have to search for the title instead: "Tow Bar - Revopoint Trackit SR"

PC Specs

Since a lot of people ask for it:

• AMD Ryzen 7700X
• 64GB DDR5 RAM
• RTX 5070Ti Desktop
• A few TB of NVME storage with PCIe Gen4 interface

My 3D Printable Accessories for the Trackit SR

Like the power bank holder or hanger for the hand: https://www.printables.com/@PrintedForFun/models

Had to scan this historic sled for digitization and put the Trackit SR to good use on the task. Initially the sled looks quite easy to scan but it actually is quite tricky: The wood is too narrow and feature derived to be either scanned with feature tracking or markers on the wood itself. But due to the substantial height of the sled simply putting markers on the floor doesn't work. Distance would be too great and therefore scan quality poor. In the end a tracker based system is needed.

Scanning

Scanning was done at 1mm target resolution and no scan spray was used, Trackit SR was used wireless. Created two scans in total: one with the sled upside and a second one downside. Due to the scanner being freshly calibrated tracking quality was indicated as excellent (when used in thermally much different conditions tracking quality will decrease). Luckily the whole calibration process only takes 20min so it can be done relatively fast.
Speed wise I can say: wireless scanning with the tracker reaches roughly 55FPS, in marker tracking mode even 75FPS wireless. Each scan took roughly 5min. Currently there are still hickups whilst scanning where the software hangs for a few seconds before continuing (more likely to happen with higher resolution). Keep in mind: this is the very first software release for the Trackit SR and overall it works already very well. I confirmed with Revopoint that they are working on the issue and it should be fixed in hopefully the next update.
Another short thought: wireless is really awesome and sometimes even leads me to forget the reduced tracker volume vs the original Trackit. Especially in multi position scanning.

Post-Processing

Processing work was minimal by first fusing at 1m0.75m (scanned at 1mm but software recommended 0.75mm) and then aligning the partial scans to one whole scan with comming point picking due to relatively minimal overlap between the two scans. Even then I couldn't see overlap artifacts so this speaks quite good about accuracy (as well as my testing does but this is for another post/video, still not sure. What I can say: Independent from tracking distance the accuracy error for a 100mm ball place is roughly 0,02mm).
Meshing was done at recommended settings and alignment to the coordinate system with Quicksurface.

Results

Best have a look at Sketchfab, result looks quite good and It captured what I needed. For me this scan still was to 75% getting to know the Trackit SR and how to use it (more like understanding the groove of the scanner).

Sketchfab

Sketchfab is like printables for 3d scans with a nice integrated viewer in browser and you can also download the scan, just look at the scan yourself.
Reddit sadly blocks the short links to Sketchfab, you have to search for the title instead: "Historic Sled - Revopoint Trackit SR"

PC Specs

Since a lot of people ask for it:

• AMD Ryzen 7700X
• 64GB DDR5 RAM
• RTX 5070Ti Desktop
• A few TB of NVME storage with PCIe Gen4 interface

My 3D Printable Accessories for the Trackit SR

Like the power bank holder or hanger for the hand: https://www.printables.com/@PrintedForFun/models

Originated form the same church I scanned the columns in.

Scanning

Scanning was done with the Sermoon P1 in large IR mode with geometry tracking. Feature tracking was luckily no problem for this part due to the abundant detail. Scan took roughly 10min due to me trying to capture all crevices. Initially even tried small IR mode but at some point the performance dropped quite sharply (simply too large part for small mode). The high ambient temperature of over 40C put quite a stress on the cooling system so after multiple scans frame rate began to drop slightly. So it settled from initial 20FPS to ~15FPS.

Post-Processing

For post processing I first transferred the scan data from the P1 to my workstation. I then fused the scan at 0,5mm with 25% noise removal to retain detail (default noise removal tends to be a bit high) and meshed it with default settings. Quicksurface was used to align the scans, delete unnecessary data (like parts of the ceiling it scanned also) and reduce triangle count under 4M.

Result

Best have a look at Sketchfab, quite content with the result. Ideally I would have used small IR mode but this would have either required PC scanning or breaking it up in multiple scans.

Sketchfab

Sketchfab is like printables for 3d scans with a nice integrated viewer in browser and you can also download the scan, just look at the scan yourself.
Reddit sadly blocks the short links to Sketchfab, you have to search for the title instead: "Wooden Altar Carving- Creality Sermoon P1"

Another car seat scanned for a custom upholstery project.

Scanning

Scanning was done in standalone mode at 0.5mm resolution and markerless laser mode. No special preparation was done, just straight up scanned at the parking lot. Tracking worked very fine due to the stitching on the seat. Whole scan was 2 to 3 min.

Post-Processing

Default processing at semi-watertight, this includes point cloud generation on my PC and then meshing. Stitching and seams came out pretty good and this is the main part that interests me.

Sketchfab

Sketchfab is like printables for 3d scans with a nice integrated viewer in browser and you can also download the scan, just look at the scan yourself.
Reddit sadly blocks the short links to Sketchfab, you have to search for the title instead: "Mercedes GLC Seat - Einscan Rigil"

PC Specs

Since a lot of people ask for it:

• AMD Ryzen 7950X
• 128GBGB DDR5 RAM
• RTX 5070Ti Desktop
• A few TB of NVME storage with PCIe Gen4 interface

My Marker Geometries

Quite a lot of people regularly ask about the geometries I use for easier tracking, here they are: https://www.printables.com/model/1543571-marker-geometries-for-3d-scanning-including-marker

These are the accompaning photos to the quicksurface tutorial about scanning and printing the original part and repaired part

A few rough outlines:

• scanned with Einscan Rigil in laser mode (no scan spray used)
• whole reverse engineering process was done within Quicksurface Pro (even the CAD part so Pro was needed)
  • Main RE steps:
    1. Import & Alignment
    2. Creation of base bodies by sketching & extruding
    3. Creating cutouts by rotating sketches
    4. Thread creation & final fillets
• After finishing the design I simply exported as .step and printed it from foodsafe PETG on my Creality K2 Part is now in use for roughly 4 months and there are no signs of damage or abrasion till now.

Scanned the two column tops for a client to replicate in a smaller form factor. The columns are from a modern-ish church near lake garda.

Scanning

Scanning was done with the Sermoon P1 in large IR mode with geometry tracking for both columns. Due to the vast amount of features tracking was no problem at all. Both scans took per piece roughly 3min. The high ambient temperature of over 40C put quite a stress on the cooling system so after multiple scans frame rate began to drop slightly. So it settled from initial 20FPS to ~15FPS.

Post-Processing

For post processing I first transferred the scan data from the P1 to my workstation. I then fused the scan at 1mm with 25% noise removal to retain detail (default noise removal tends to be a bit high) and meshed it with default settings. Quicksurface was used to align the scans, delete unnecessary data (like parts of the ceiling it scanned also) and reduce triangle count under 4M (still within 0,03mm tolerance from the original mesh).

Result

Best have a look at Sketchfab, quite content with the result. Maybe should have let myself take a bit more time to also scan the top of the columns. Captured detail is fine, rough stone texture is quite hard to nicely scan.

Sketchfab

Sketchfab is like printables for 3d scans with a nice integrated viewer in browser and you can also download the scan, just look at the scan yourself.
Reddit sadly blocks the short links to Sketchfab, you have to search for the title instead: "Column Stoneworks - Creality Sermoon P1 IR"

PC Specs

Since a lot of people ask for it:

• AMD Ryzen 7950X
• 128GBGB DDR5 RAM
• RTX 5070Ti Desktop
• A few TB of NVME storage with PCIe Gen4 interface

Scanning

Scan was performed in VCSEL mode over WiFI. As preparation I simply placed the dog plush in a well lit area (very important since this basically dictates the whole outcome of the texture and splats, ideally you would have a light box or similar). Texture camera exposure was set to manual to keep uniform lighting and fill light was disabled. I then tool roughly 30 pictures with my phone from every angle (exposure was locked to get realistic lighting).
After scanning where I ensured to capture the dog from all angels I performed default processing up until the mesh was created. From there I continued with gaussian splatting (high setting).

Splatting

Splatting is very simple: Just mesh the data (ensure you enabled texture capture while scanning), then click on "3DGS", point to a folder with the additional pictures (optional, so it still works as previously) and wait until the splatting is finished. First takes a minute or two to prepare and then pounds the GPU for training the splat (currently only on either Nvidia GPUs or Apple Silicon Macs). In my case the dog took roughly 15min of training. With the additional pictures it takes a bit longer than without.

Results

Putting the two splats side by side the difference is quite clear: with additional pictures much more detail is reconstructed in the fur and even the wood grain. The whole scene looks also more naturally exposed. Only downside I noticed is that the fur isn't fully opaque, guess this could be solved in post processing or with taking more pictures to start with.

Viewing the Gaussian Spalt

I uploaded the splat to supersplat if anyone wants to view it themselves. Simply open the link or search for the title: "Dog Plush - Revopoint POP4 with additional Pictures"

Just after testing the gaussian splat function for the POP4 an update was released that allows adding additional pictures to the splatting. The pictures can be taken with any camera (I used my Samsung S25 Ultra) and you just provide a path to the folder with the pictures. Aside from the the other main two changes are: GPU support for FullField fusion (only takes a few seconds now) and finally a transform tool to reorientate scans to the coordinate system. Still very manual and not 321 alignment but a step in the right direction.

Scanning

Scan was performed in VCSEL mode over WiFI. As preparation I simply placed the dog plush in a well lit area (very important since this basically dictates the whole outcome of the texture and splats, ideally you would have a light box or similar). Texture camera exposure was set to manual to keep uniform lighting and fill light was disabled. I then tool roughly 30 pictures with my phone from every angle (exposure was locked to get realistic lighting).
After scanning where I ensured to capture the dog from all angels I performed default processing up until the mesh was created. From there I continued with gaussian splatting (high setting).

Splatting

Splatting is very simple: Just mesh the data (ensure you enabled texture capture while scanning), then click on "3DGS", point to a folder with the additional pictures (optional, so it still works as previously) and wait until the splatting is finished. First takes a minute or two to prepare and then pounds the GPU for training the splat (currently only on either Nvidia GPUs or Apple Silicon Macs). In my case the dog took roughly 15min of training. With the additional pictures it takes a bit longer than without.

Results

Putting the two splats side by side the difference is quite clear: with additional pictures much more detail is reconstructed in the fur and even the wood grain. The whole scene looks also more naturally exposed. Only downside I noticed is that the fur isn't fully opaque, guess this could be solved in post processing or with taking more pictures to start with.

Viewing the Gaussian Spalt

I uploaded the splat to supersplat if anyone wants to view it themselves. Simply open the link or search for the title:
"Dog Plush - Revopoint POP4 with additional Pictures"

PC Specs

Since a lot of people ask for it:

• AMD Ryzen 7950X (16C/32T)
• 128GB DDR5 RAM
• RTX 5070Ti Desktop
• Fast NVME storage

Finally wanted to test the gaussian splat feature of the POP4. My main domain is reverse engineering and mechanical construction so it is not my native field of experience. But I have been playing around with splats a bit more in recent time for VR tours of Lidar scanned sites so I am not a complete stranger. Normally splatting requires third-party software you feed a base point-cloud and images, software most of the times isn't free to use (either payed one time or as a subscription). RevoScan 6 is free to use so I was curious how usable it would be.

Scanning

Scan was performed in VCSEL mode over WiFI. As preparation I simply placed the dog plush in a well lit area (very important since this basically dictates the whole outcome of the texture and splats, ideally you would have a light box or similar). Texture camera exposure was set to manual to keep uniform lighting and fill light was disabled.
After scanning where I ensured to capture the dog from all angels I performed default processing up until the mesh was created. From there I continued on three tracks: coloring with texture images, vertex coloring and gaussian splatting (high setting).

Splatting

Short PSA: With initial beta version of RevoScan 6 I sometimes encountered crashing of RevoScan while splatting but luckily this issue is fixed in the latest version. Haven't had a crash since the update.
Splatting is very simple: Just mesh the data (ensure you enabled texture capture while scanning), then click on "3DGS" and wait until the splatting is finished. First takes a minute or two to prepare and then pounds the GPU for training the splat (currently only on either Nvidia GPUs or Apple Silicon Macs). In my case the dog took roughly 10min of training.
Only thing I really wish for is an integrated alignment function in RevoScan.

Results

I was most interested in comparing texture quality to the gaussian splat. Texturing is possible in two ways within RevoScan 6: image or vertex. I performed both and additionally generated the splat in high settings.
Looking at the comparison images the difference between image and texture color is quite apparent, texture simply looks better. Comparing the splat and texture image shows that the splat quality is quite high, I even tend to prefer the quality over the image texture due to the inherent "softness" a splat can have and the nice behavior across different viewing angles. Even the wood texture was captured quite nice in the splat.

Viewing the Gaussian Spalt

I uploaded the splat to supersplat if anyone wants to view it themselves. Simply open the link or search for the title: "Dog Plush - Revopoint POP4"

PC Specs

Since a lot of people ask for it:

• AMD Ryzen 7950X (16C/32T)
• 128GB DDR5 RAM
• RTX 5070Ti Desktop
• Fast NVME storage

This bronze statue is a bit smaller than life size and is free-hanging on a white, feature-less wall against bright, sunny sky. Which are quite hart conditions for scanning. I scanned the whole statue in large mode and separately scanned the head section in small IR mode.

Scanning

Scanning was done with the Sermoon P1 in large IR mode with geometry tracking. The P1 was used standalone outdoors in bright sunlight, this caused initially some problems with auto exposure due to the very bright sky but with manual exposure corrected to the statue itself it worked very good. In the camera preview the main scan object should be correctly exposed (even if it only is a small part of the picture against the bright sky). Small mode was scanned in the same manner with the only difference in mode selection.
With initial firmware versions the P1 only could do 5-7fps in IR mode standalone, with the latest version it reaches a stable 19-20FPS. With the increased speed the scan only took roughly 6min without any tracking problems (was actually a bit surprised about this).

Post-Processing

For post processing I first transferred the scan data from the P1 to my workstation. You can either do it by cable where a nice import menu opens or over WiFi where you select the project on the scanner, join the hotspot of the scanner with your PC and send it to the PC (option I chose, simply less cables and quite fast with up to 120MB/s from testing). I then fused the large scan at 1.5mm with 25% noise removal to retain detail (default noise removal tends to be a bit high) and meshed it with default settings. Small mode was processed in the same manner with the difference of using 0,5mm point distance for fusion. Maybe bit more noise removal would have been beneficial for small mode but the noise is still tolerable.

Result

Best have a look at Sketchfab. Looking at the direct comparison it is quite obvious that the small IR mode simply provides more detail in the final result. Due to the closer working distance small mode requires I had some tracking issues on the head, wheres large mode had no problems with tracking at all. Ultimately even in small mode the scan worked and didn't result in garbage data.

Sketchfab

Sketchfab is like printables for 3d scans with a nice integrated viewer in browser and you can also download the scan, just look at the scan yourself.
Reddit sadly blocks the short links to Sketchfab, you have to search for the title instead:

Large Scan: "Statue Monk - Creality Sermoon P1 IR large"
Small Scan: "Statue Monk - Creality Sermoon P1 IR small"

PC Specs

Since a lot of people ask for it:

• AMD Ryzen 7950X
• 128GBGB DDR5 RAM
• RTX 5070Ti Desktop
• A few TB of NVME storage with PCIe Gen4 interface

Printables Link: https://www.printables.com/model/1736976-revopoint-trackit-cold-shoe-mount-for-smartphone-h

Tried a lot of different options with the Trackit to live-view my scan progress without constantly staring at the PC screen. Started with RDP to my tablet, then RDP to my AR glassed and now landed at the final solution I like the most: I just use a cold-shoe smartphone holder directly at the hand piece so I don't need any special equipment and no additional stuff to carry around. To enable RDP even at offsite locations I have a small tp-link travel router (WR-3002AX) that can be powered over 5V USB, therefore I just plug in LAN and USB to my PC and connect my phone to the WiFI. Et voila RDP works.
Mounting the smartphone to the hand-piece was the hardest part. Initially had no idea how to fit it without disturbing the grip or button presses. Then I got a nice adjustable smartphone holder for birthday and finally had a solution in my hands. Only thing missing: A cold-shoe mount on the Trackit hand-piece which I had to design. To ensure proper fitments and ergonomics I created multiple scans as a basis.

Scans

I used the POP 4 to create three scans:

1. Trackit hand-piece (laser mode, dark object setting, no spray used but maybe should have used since the carbon weave became visible). Marker geometries were placed on the hand-piece's rim, scanner did also see the 10mm markers already on it.
2. Smartphone holder in a position that works (laser mode, normal setting, spray was used due to the fact that the black anodizing absorbed too much light)
3. Comfortable grip whilst pressing the top-most button (needed to ensure enough space is available to press all buttons under the holder, Hybrid HD mode with feature tracking) All scans were processed with default settings. Third scan may have also worked in VSCEL since I didn't need the higher detail of Hybrid HD

Design

For design I imported all scans in Quicksurface Pro and first aligned the 1st scan to the coordinate system and cleaned it up. Then I fit the third scan by picking common points and isolated the hand in the third scan. Based on these two scans I now could place the smartphone holder in a position that doesn't obstruct the button press and grip (nearly a perfect fit xD).
I then auto fit a surface in the crevice of the hand-piece which will be used as contact area for double-sided tape. After thickening and cutting it to the right size to nicely fit in the niche I combined it with my default cold-shoe mount I imported in Quicksurface as a STEP-file. Last steps were some fillets and chamfers. Not a too hard design but for me quite useful. Whole scan and design took roughly 2h including one revision to enlarge the contact area for the double-sided tape (includes wait time for the first print).

Printing

Printing is best done from PETG or ABS. YOu may want to roughen up the surface of the print and use IPA to wipe any grease from the contact areas before attaching the double-sided tape (I used two layers to have a bit of additional thickness). Firmly press the mount in place to ensure good contact. Then enjoy your new freedom from a PC screen with the Trackit.

Smartphone Mount used

Any other tiltable cold shoe smartphone mount should also work, just like this one. But this one is tested to ensure buttons are still reachable and has endless swivel points for adjustments. https://www.smallrig.com/de/smallrig-Metal-Phone-Holder-with-Cold-Shoe-Mount-4382.html

Just seen that it got updated and finally a MacOS version is supported.

When creating scans for my model library I scanned this lion sculpture recently. Nothing special but can be quite decorative when placed properly.

Scanning

Scanning was done in IR large mode with a target resolution of 1mm. For tracking I only used geometry tracking (hybrid with texture can be a bit challenging when scanning against the bright sky). Took roughly 4min in standalone mode to scan the statue, the rough surface provided enough features for tracking to work easily.

Post-Processing

For post processing I first transferred the scan data to my workstation using the wired connection (simply faster than wireless, process for file transfer is luckily similar for both modes). Then generated the point cloud and meshed it with default settings. Alignment the scan with internal tooling of Exscan Rigil (first quick alignment and then fine alignment) and reduced the triangle count to roughly 20%. Good thing is that even tough alignment is in a processing step after the mesh editing (reduction part of this) you can switch between them easily. Even keeps the textures correctly.

Result

Best have a look at the sketchfab model, texture was only a nice addon for me. Mainly need the 3d model. Turns out Exscan Rigil can export textured .ply-files. Finally small enough in comparison to .obj to also upload textured to sketchfabs

Sketchfab

Sketchfab is like printables for 3d scans with a nice integrated viewer in browser and you can also download the scan, just look at the scan yourself.
Reddit sadly blocks the short links to Sketchfab, you have to search for the title instead: "Lion Stone Statue - Einscan Rigil IR"

PC Specs

Since a lot of people ask for it:

• AMD Ryzen 7950X
• 128GBGB DDR5 RAM
• RTX 5070Ti Desktop
• A few TB of NVME storage with PCIe Gen4 interface