That what I am asking for , for over a year now to allow us to save full projects via mobile devices … that would be cool and you would gain 100% quality . We see what the future bring to us .
3 Likes
That’s true @shadows44 , also scanning via WiFi is much slower , if I good remember around 12fpm to maximum 18fps via USB .
How slower the frames how more issues with tracking , the more expensive portable scanners uses around 80fps .
However I did not experienced huge lag difference between tablet/PC win11 or my PC workstation while scanning , on my android phone I did .
Right now I skipped using Android phones and tablets and use win11 tablet/PC via USB or wifi scanning direct to my laptop and stick … and that is my personal best setup .
1 Like
Yes but when you are outside may b at 2 mt high it’s very difficult to scan with a laptop…
I use tablet PC win11 so easy to hold in one hand or under the scanner attached like a phone
So, how much detail do we really lose when using a smartphone?
Let’s find out !
I scanned a bread and some walnuts (I feared the bread might not have enough features, and I figured the walnuts would improve tracking).
Revopoint RANGE on a tripod, subject on a turntable. The setup did not move between scans.
Capture settings:
- high accuracy
- alignment: features
- scan object: general
- texture: color
I scanned a partial turn each time, starting and ending at (roughly) the same point.
Reconstruction done in Revo Scan / in the Android App, highest quality settings, no denoising/smooting.
Top: smartphone (Galaxy S10, USB mode), bottom: PC/USB 3.0.
Top: smartphone (Galaxy S10, USB mode), bottom: PC/USB 3.0.
For my final review I’ll probably showcase a more detailed view of these 3D models.
Is there a difference? Indeed.
Is it that bad? It really depends on want you intend to use the 3D model for.
Sidenote, though: If I’m not mistaken, the current version of the Android app does not fill holes, even when the option is checked. Might be a bug.
Here are a few screenshots of the textured model (PC version):
3 Likes
Hi @shadows44
Now you should Batch processing the scanned project at maximum quality , you going to have even better results as in Revo Scan default meshing .
Maybe even by 30% better .
For me Android app is good if I want to make some quick model for sharing online , more for the textures model because the mesh have very low accuracy basically not much usable in my work pipeline . Unless I use the fused files then it still can be recovered to workable mesh with RGD data .
1 Like
Let’s reprocess using Revo Studio, then! 
Same Android scan as before, for reference :
Model reprocessed in Revo Studio, overlapping points removed (overlap detection: medium), “Basic” conversion mode, HD Mesh Level: HD.
And PC/USB scan reprocessed in Revo Studio, removed overlapping points removed (overlap detection: medium), “Basic” conversion mode, HD Mesh Level: HD.
Overall, reprocessing isn’t magic, what wasn’t captured still isn’t there, however you definitely get sharper details. Which is great!
Now do that PC version via Batch processing … at maximum settings no denoiser , you going to see the whole capacity of Range , Batch processing use 30% more points from the point cloud than Revo Studio or Revo Scan by default . Best accuracy and details if your object was scanned with Range at 30cm distance .
1 Like
That is not exactly the truth when we speak about Range , Revo Scan version 4.3.1 is not yet fully optimal version for use with Range , Revo Scan 5 will be .
Do the Batch processing , you going to see extra details not just sharpness .
here an example
3 Likes
Let’s re-reprocess 
using Revo Scan / Batch mode, then !
There are more details, but also quite a lot of noise.
I tried to apply some smoothing (level: medium)
It might still be a bit better than the Revo Studio version (below, for reference), although smoothing the mesh probably crushed some details.
Is there any announcement yet as to when Revo Scan 5.0 will be available/what features might be included, beyond improved processing quality?
2 Likes
No , not yet
At this moment not much informations can be shared since it is still in production and beta trials did not started yet .
The bigger object scans profit mostly from batch processing and depends of how much frames was captured so I am talking about 1200 and above , the batch processing use different meshing algorithms than Revo Scan default or the simple Revo Studio meshing .
So for the best results it is the best option at this moment .
I can see the difference in your last processing , especially now comparing it to android app mesh .
3 Likes
Another human scan: sitting on the turntable is a good way for the subject to remain still!
Probably my best human scan so far.
(scan done using Revo Scan on Windows)
I tried to recover more details using batch processing (custom settings: point pitch set at 0.1mm, mesh level 6, denoise 1, fill holes on).
It works (this is quite obvious when looking at the stitches, pockets, shoes and even the face) but then there are a few issues on the mesh. @PUTV any suggestions on how to avoid these? I know they could be cleaned up, but it would be even better to avoid them altogether. Maybe by tweaking the batch processing settings, or by slowing down the turntable?
And here’s the texture: far from perfect but it might be good enough if you want to use the scan as a background character in an architectural render, for example. Or as a reference while creating a new texture using Substance tools.
6 Likes
Very good @shadows44
What a difference in quality !
The small artifacts are something that you have to life with .
It doesn’t happen always but sometimes it do , mostly with overlapped frames .
I hope RS5 will have a fix for that as you know 4.3.1 is not yet optimal so small errors are to be expected .
I try to scan full body as fast I can , Infrared scanner at 18 fps don’t like to hang too much in the same places , it accumulate too many frames that can result in the artifacts .
CC could fix things in this case , hopefully RS5 will .
Good job 
I like it so don’t get too hard on yourself , you going to master it in no time .
2 Likes
Thanks for the feedback!
I will therefore keep the same turntable settings for my next scans: at or near top rotation speed. I assume this helps to avoid accumulating too many frames in the same area.
Overall, at this stage my takeaway on the RANGE would be something like this:
very nice tech, impressive geometry for the price, some frustrating issues (tracking/alignment issues are quite frequent, textures are not very good), retouching the models in post is a solution but hopefully software improvements will help with the drawbacks.
If you’re aware of the drawbacks and are fine with them, go for it. If you plan to scan humans, the turntable is quite helpful.
Sidenote: I noticed that Revo Scan does store individual frames in a folder ( C:\Users\YourUserName\RevoScanData\project\YourProjectName\cache ), so re-aligning misaligned frames or removing specific frames is probably doable. Again, we’ll have to wait and see what future software updates will bring.
1 Like
The files are point cells frames , it can’t be removed randomly and not after the cells are fused into point cloud , it is not that the whole cell frame got bad , just a fragment where it overlapped another cells points by things like a simple shaking .
But that is of course the software that need to be addressed and future versions of it .
Regarding the drawbacks, it is random based on many factors and not really happening each time in the same situation .
Regarding textures , the textures are too small to provide as much information for close ups like humans , it would need to be at least 64K to be zoom able quality , with 4K you don’t get as much pixels for this size of model . 8K would be better and minimum requirement for that size of model . Normally I use 4 to 8K only for the head … to have a decent quality .
3 Likes
Thanks for the info about cells frames!
Indeed, tracking/alignment issues are a bit random. That isn’t a deal breaker of course, as long as people know some postprocessing or trial/error will often be required.
Yes, 4k texture is low for a full human. That being said, I feel the end result could be improved even with 4k textures: at the moment they feel oversharpened, with hard transitions between individual pixels.
Here is a close-up picture of the textured model I shared earlier:
It seems that .img files in the cache folder of each project are jpeg files captured during the scan (1280x800 pixels), by the way.
Here’s a cropped .img file from the previous scan.
I assume the texture is created using these pictures. If so, blending or improved blending + ghost removal might help a bit.
The files are used for masking the object out for the texture mapping , each frame have own files when scanning with colors .
It is too small format for that size of object , scanning smaller objects works better , it is actually impossible to get as much pixels with this size , each point is your pixel ,
I have very nice results on a smaller objects where. 4K was OK
3D render
8 Likes
From what you said might I assume that if I scan the head with the scanner closer to the head to have a bigger image of the head it would improve the pixel quality?
1 Like