the 1mm magnets are strong , not getting off from metal that easy , and they get close enough to each other to be usable but not too close what is good .
@b0ssk tested out and the magnet are not visible to the IF , no matter markers on or not
the red dots are the magnets with reflective markers
the green dots are magnets only
the yellow dots are POP2 “original” paper markers
the trick to avoid bumps are keeping the gain settings low , higher gain settings will create the bumps from the black paint surrounding the markers as it does when scanning black text.
1 Like
My next test with markers is placing it not on the flat plate but below the object , in my case I stick the paper markers to the POP2 small turntable and put it on my rotating big turntable, the result is excellent , no mandala effect
usually when you place small objects on the plate the markers are barely picked up , this way you don’t have to stick it to your object and they are always visible to the sensors.
Still Marker mode
1 Like
Awsome, I’ll keep that in mind when my markers show up! Have to wait till Wednesday, in the meantime I’ll just play around and get comfortable with the unit! I ended up pulled off the cold shoe place too and using a spare Arca Swiss Plate and mount, gives me easier connection options with all my little camera bits, adapter and tripods I’ve got laying around.
1 Like
Yes I am using my ball head mount too and cam arm , much better , can also attach my POP2 to my Nikon when capturing pictures for manual texturing of objects at the same angles while scanning .
So you’ll generate a point cloud and mesh with the Pop, then generate textures after from your nikon captures? How does that work? Do you bring the two different data sets into something like RC? Or does the Revo software support that? This aspect of stuff is totally new to me!
No, I am projecting the photos using Zbrush & Photoshop after everything is captured and ready for post production , mostly stuff for my work .
The same way Revo Scan is doing when you press the Texturing button , just manually and with precision and of course much better quality , but it take some time to produce so I reserve this workflow only for my job productions .
Also using polarized filter to get rid of specular reflections , especially when producing human textures or for objects I scanned outside , I don’t have that issue when scanning in my light box since there is not direct light coming from behind .
As I talk to @kaczorefx in his thread about improving scanning here is little preview reducing the IR gain manually using ND filters over the IR projector that I came up with yesterday , specifically if you scan small white object and you are very close and still have red indicator in your Depth Cam preview , the red you see will still produce some noises , so to eliminate that I used my ND Gel filters over the IR Projector to reduce the gain manually and it needs no less than ND 32 and up
results was great , the gain was down , not need denoiser in Feature mode , object at 15 cm … meshed at 6, zero Denoiser , pitch point at 0.1mm , clean and sharp
and here without the ND filter
My turntable was at 60 sec per 360 degree as the slower rotation do produce even less noises but your object can’t be over scanned , just one 360 degree turn .
Now my next goal is to make the IR pattern even more sharper on distance scans …as you know at 60 cm it starting to lose completely the fine details .
Redneck stuff but hey it works , I don’t want to open my POP2 because I would lose the warranty , but once it is over … lots of things gonna happen 
4 Likes
I am mildly surprised to learn that these work.
Other than the white circle being 20% larger than the Revopoint markers, how would you compare them?
This really no matters what the size of the black or white circle really is , 2mm 10mm or 20 mm
The Revopoint markers white dots diameter is 4mm , the one on the turntable plate is 5mm , the software recognise everything as a marker point even a 1mm , put some bread crumbs on the black plate and use marker mode , you will see tons of marker points showing up on your screen . The software do not calculate the object in space using the marker mode , it only use it for internal visual tracking ( not the object tracking ) in one scanning session , because when you scan one side of the object using marker mode you can’t flipt it on other side and continue scanning , there is no object tracking feature anymore , for that reason the marker mode can be only used for partial Scanning of the objects that need to be exported each time and merged in Revo Studio or CC , why it is like that ? because the software does not record where the markers are in the space when flipping the object , it is just visual tracking in real time when the markers are visible by the sensors.
however 1mm round white sugar sprinkles works best. but please do not do that , as it can easy move , and will interrupt with your tracking
3 Likes
But seriously , if you are on a budget and you want extra tracking spots on your black plate or black turntable , you can get yourself 1000 white sticker dots , it will only works on black surface or any other that is not visible by POP2 sensors like black silicone or rubber or shiny metals
What is the attenuation value of the ND filter?
1/2? 1/4?
ND32 -5 Stops -1.5 Optical Density
I used just simple ND Gel filter attached over the IR projector, you can cut them exactly to match the size.
The best between ND32 and ND48
You have also glass and metallic-coated ND filters that attenuate differently , by just absorption or reflection and absorption , but since we do not take pictures here and just reducing the strength of IR light, this really don’t apply here.
Any cheap ND Gel filter can do it , starting at ND32
I talk yesterday ro Revopoin programmers showing them the results , they going to evaluate it and adjust the proper gain settings if possible via the software , so we may not need custom altering .
Quick Showcase for today
I did used my ND32 filter over the IR sensor, the skull is very reflective at the distance so spraying it do not helps either. But slower turntable at 60 sec per rotation and ND filter over the IR projector did the trick. No fusing between changing positions , all in one session .
I am planning a very special video for Friday the 13th , you DON’T want to miss it !
5 Likes
Got some pretty cheap and accessible filters arriving today. If they seem to be any good, I’ll post here.
1 Like
Hello, what is called ND gel filter? do you have an example? where do you find these density filters? Personally I have several ND filters but in glass for my Canon camera…but I don’t know what a gel filter is?
1 Like
That is the same as the glass filter you have for your camera , but made from gel like plastic, so you can cut it to your needs , you can find ND Gel filters on Amazon , if they don’t have ND32 you can get ND16 and use 2 on top of each other . I use it sometimes with my 8mm lenses that can’t accept glass filters on top, and the gel filter can be slided from the other side as in this case it is the only option for this kind of lenses . ND gel filter are very cheap
But the programmers will fix the gain issue soon anyway
1 Like
Cut it to fit on the top exactly you don’t even need to attached it with anything as long it lay flat on top of the IR projector . It will reduce the intensity of the IR and make scanning small white objects at 15cm distance much better , also it improved the tracking somehow on the small objects .
Well, what I ended up buying were some cheap IR filter lenses for a Chinese-made drone.
I know nothing about lenses and filters, so you might be thinking “Oh no, that’s not what you want.” but they actually work. Not done much in the way of a “scientific” test, I just had a quick play to see if they’d do anything.
With the ND32, I have to raise the software gain to 10 just to pick up anything. The ND16 is the only other one I tried so far… and it’s very promising. With that one taped over the IR emitter, I could get the Pop2 a fair bit closer to the object without the red blowout or too few points being detected. Gain was set to 6, so had a bit of wiggle room either side.
Did a very quick scan of an old, small WilyKit figurine which I’d previously attempted with the Pop2 and got a much better scan this time. No surface prep etc. just plonked it down on the turntable to see what would happen.
The only problem being that the filters are a lot smaller than I expected - they cover the IR emitter but are a massive PITA to tape in place without getting fingerprints all over them.
Edit: I should add that I always scan in almost darkness (just the laptop screen) which may have affected the performance of the filters in relation to the gain. I don’t know.
@Karl-D yes exactly , you should go for the gel filter because they do not block other stuff for the reason you need to set the gain so high , but thankfully it works for you .
The reason I asked for the gel filter is because not only it reduce the IR, it also diffuse it especially if you use 2 ND16 on top each other , I read some technical papers the other day about new techniques where the filter was used to diffuse the IR light in 3D scanning resulting in better scans .
I found out that the pattern matters well if it is slightly diffused , it diffuse the reflections on the shiny surfaces reducing the noises , so not only my ND gel filter reduced the gain , it also diffused the IR pattern very slightly because it was a gel , and not metallic coated.
It works only for scanning small objects so not recommended for full body scans or big scans that are far away without red gain visible .
When you meshing the difference may be not there , but when you use point cloud data as me, the difference matters a lot .
here some examples of the differences , you see the noise path on the cloud point after fusing without filter? that would case some distortions in the mesh when meshed at higher resolution than 6 in Revo Studio or other software , remember Revo Scan meshing quality at max is only 25%
1 Like