Just a rundown of some of the scans I’ve been doing with the POP3 Plus in testing - some you may have seen before from the original discussion started by Jane, but just replicating them here so everything is in one place.
So to kick things off…
When I do head scans with the Miraco I tend to scan the whole head in Far mode, then switch to Near mode and scan the face into the same scan file, using the Far data as the locator.
Then in processing I generate that hybrid Near/Far scan into the whole head, which slightly improves on the definition of just the Far mode on its own, but doesn’t get all that juicy Near mode detail.
To get that I go back into the frames, delete all the Far frames and then regenerate the Near frames into a new model. Then I take the previously saved Near/Far hybrid mesh plus the high detail Near mesh into my editing software and combine the two into a final, ‘proper’ hybrid model as the final deliverable.
Well, I can do that too with the POP3 Plus. Scan the head in normal 1x mode, then switch to 2x and scan the features and follow the exact same steps as I would for the Miraco scan… except actually, seeing as you’re meshing both at the same level, you can essentially just combine them inside RS5 at the Fusion stage.
It’s not quite as pronounced a difference as Far/Near on the Miraco, but if you want to wring as much detail as possible out of a head scan with the POP3 Plus? As the Mandalorians would say… “This Is The Way”
So there you go - one of my head scan secrets for you for free. You’re welcome The one drawback? Ye gods look at all those wrinkles. I kinda prefer the cosmetically smoothed look…
Couple of renders to demonstrate. Left: Hybrid 1x/2x mesh. Right: ‘Combined’ final mesh.
Just a few side-by-side renders of the POP3 (left) vs POP3Plus (right - 2x zoom mode) using a deer skull I’ve scanned before with the POP3, so bear in mind that the POP3 version was scanned with an older version of the software.
The texture is definitely a lot more even, with less contrast and hence less saturation, with less magenta in the mix (very evident on the rear antler), which is all good. There could still be a little more yellow put into the mix, but overall there’s a marked improvement here in terms of realism and faithful reproduction, however, as stated above this is probably down to the new software versions, as I’m aware that they recently tackled the high contrast issue. I’d have to do a new scan with the POP3 to see if there’s any change in the actual optics and hardware processing.
The data, even with the slight softening of the Colour Image mesh, shows a little improvement when using the 2x zoom mode, although as we can see below, this comes into its own more in Per Vertex or just exporting the untextured mesh.
There’s a definite improvement here when using 2x zoom for that extra definition. It’s almost like an anti-aliasing filter has been removed as with the Nikon 7100 and 810 when they came out, giving that crispness and edge to detail. You can see it more pronounced in this zoomed image:
It’s a subtle improvement, obviously not as marked as, say, the Miraco’s Near and Far mode, but it’s a definite boon to a previous/potential POP3 user to have this extra tool in their kit bag.
I redid the deer skull, treating it more like an object to scan this time, as opposed to a test object.
I find the 2x view has a reasonably large FOV beyond the view pane - a little like the MINI2 in that it extends to each side a little more than the view, which helps keep tracking on course.
The texture looks great - for the best texture result you really have to keep it at that “on the border between Excellent and Good” where the colour camera focus point wants to be. If you just sit in the middle of Excellent, it won’t quite be as crisp… although maybe the data might be a little better, but I can’t confirm that.
This is the Per Vertex or Mesh model with the Colour Image map overlaid, so you get the best of both worlds (as the Colour Image model is usually a tiny bit softer).
Everything is as it comes from RevoScan5 - no editing in any third-party package. If this were a final asset I’d take the texture into Photoshop and definitely tweak it to take away a tiny bit more magenta and add in some yellow, but that’s only comparing it to the original. If you just take it as it is it looks pretty great.
I need to re read this several times to get a good understanding of the near/far combination and seperation, too many long days and late nights at the mo lol.
I’ve scanned a cat flap in far mode, then continued in near mode but didn’t seperate the frames. I’m wondering if understanding your technique would give me better results.
A scan of a print of a scan! This was a full-sized clay head from The Dark Knight scanned back in 2007, which I then printed on an original ANYCUBIC Photon… and then used the result as a small test object for other scanners!
Again, details this small and intricate are asking a lot from the POP3Plus, even in 2x mode, but it does an amazing job especially in continuous mode (as you can sometimes wring out a little more detail in Single Shot mode).
To be honest it’s asking a lot from any system outside of blue light, and is definitely pushing the envelope of what’s possible with standard IR. Good job!
No post editing - this is a raw mesh straight from the scanning software.
I might now try in Single Shot mode just to see how much detail I can get…
Quick (around 4 minute) scan of a 6.5"/16cm action figure (bonus points if you can name the comic book, character and series he’s from!) to test out the colour capture ability of the POP3 Plus.
This is raw mesh data generated directly in RevoScan5. The mesh could use a little light smoothing/editing, although it’s worth pointing out that this is only because I generated the initial point cloud at 0.1mm - at 0.2mm or 0.25mm the mesh is considerably cleaner for only a tiny loss in definition, but as I was after as good a Per Vertex colour map as I could get, I was trying to maximise those vertices at the Fusion stage, so accepted a little rougher mesh.
All in all though a really nice diffuse texture map for what was a fairly shiny plastic model, with good fidelity to the original colours.
Sorry for the slightly ungainly forward-leaning stance - it was the only way to get him to balance and not fold back at the waist!
Mode: Feature - High Quality - Dark Mode
Fusion: 0.1mm
Mesh: Level 7
Render: Keyshot
Scanned with a POP3Plus - not my best paint job ever (I was practicing my novice airbrushing at the time!) and I didn’t even get around to finishing one of the eyes… sorry!
But it was a good test of the system’s colour capture ability. I used the internal illuminators and generated the map from Per Vertex colour information, which doesn’t give quite as crisp a texture as the Colour Image, but is more even and fine for something this size.
It was actually also a pretty good test of the geometry capture too, seeing as there were quite a few reflective areas and obviously no treatment was used. There’s a little noise on the silver areas, but not much.
The model was created by my industry buddy, the incredibly talented Jet Cooper (Jet3D (@jetcooper3d) • Instagram photos and videos) so I went through the bizarre and slightly pointless process of taking a digital model, printing it in resin, painting it… and then scanning it back into digital form again!
The one drawback? Ye gods look at all those wrinkles. I kinda prefer the cosmetically smoothed look…
