OK, right now Revo Studio does not support textures.
Not only does it NOT create a texture file during meshing like RevoScan does.
But it actually removes the UV channel from all meshes!
If you import the fused_mesh_textured.obj or ply into RevoStudio and then export it back to obj - the resulting obj model is stripped from teh UV data - a model like that cannot be textured later.
Please fix this, because for now RevoStudio is not only useless but actually dangerous since users get a lower quality model out of it than the model they got from scanning
OK, right now Revo Studio does not support textures.
For real ? I would be not so sure about that
you need to learn a lot Filip !
It is best to working on fused point clouds only in *.ply mode , it include RGB data ( if scanned with color) that will be not destroyed , after you finish your edit you can bring it back to Revo Scan and generate new Textures .
The UVs are generated once you click the Texture button in Revo Scan and not early .
It is impossible in any program to edit a geometry and keep UVs , if you change the verticles your UVs are gone … natural process
The “make Textures/UVs” function is still missing in Revo Studio and I hope they add it soon .
“It is impossible in any program to edit a geometry and keep UVs” - what?
I do this every day all day. Every single program that lets you edit geometry, preserves the UV channel.
Meshlab works properly with the ply generated by RevoScan, does not destroy the UVs, and works with texture.
The texture generated by RevoScan cannot be recreated just from the point cloud and vertex color data - it is much more detailed then the vertex density. If I had to guess I’d say that in the last texturing step, RevoScan is projecting stored RGB images onto the mesh filtering it by mesh normals. This cannot be done from the ply files alone - you need to use the Revoscan csprj source data.
This shouldn’t be a problem for RevoStudio, but since the data is in proprietary format, no other professional software can do it
I still have a lot to learn, you’re right, but after 20 years of professional work in CG I think I know a thing or two
That said, the right scans are from Pop2? The skin pores are sub mm features. they shouldn’t be possible with a point pitch of 0.1mm.
How did you achieve that?
And by being dangerous I meant exacly that - you export a scan from RevoScan to an obj file with a texture - it looks amazing. You go through a couple of simple edits in RevoStudio - export it again to an obj file, open it in any other application and now the scan looks like crap - the mesh hasn’t changed, but the texture now is just from the vertex color data, and not from the scanned texture - the difference is huge
How do you do that? When I reload an object file back into Revo Scan, I cannot get it to do anything unless I start a scan and stop it almost immediately; this can introduce errors.
I was curious about that, too, but it occurs to me that it could also, maybe, be noise.
@kaczorefx did I hit your ego ? lol you can have 20 years experiences in CG I don’t doubt about that, but 2 months maybe with POP2 …
and BTW all my textures are recreated from color per vertex data as they are the best … but since you have 20 years experiences you should know that .
Yes you are right that the csprj visual studio project files include all the data + img files to create new textures and UVs as it use photogrammetry to texture the object separating the object from everything else with a mask using all frames . I cracked it down already .
Since you lost your UVs by editing the geometry vertices you got yourself an answer …
and yes the scans as many others I made are from POP2 , the more detailed are meshed in Revo Studio
Well there is some very tiny level of noise but the skin surface is not even as you noticed and have a different features across the whole face and neck, I posted the close up before showing the deeper details .
Revo Scan meshing level is 75% lower than Revo Studio can do , it looks good , don’t it ?
well that is all what matters . The Revo Scan mesh is 2 millions poly and the Revo Studio is 9 million poly it chance the details a lot .
Not realy an ego thing - just wanted to get the discussion on the right track
What do you mean by " textures are recreated from color per vertex data as they are the best" - the RGB based textures have way more detail. The vertex color texture is limited by the mesh resolution - the rgb texture is not.
Yes you’re right that is just the beginning with Pop2, but Pop2 is not the first scanner on the market, far from it. We already have established workflows for scanning assets, and no matter how good Pop2 may be, if it doesn’t adhere to existing standards it will remain a toy for hobbyist.
Also you recreated the textures from vertex colors, fine, using what? RevoStudio? Nope, doesn’t support textures. You said you can import edited mesh back to RevoScan - mind sharing how? All I see is that you can only import a csprj file into RevoScan
According to Revopoint true claim the point cloud data can reach 0.02mm when meshing using Revo Studio . 0.1mm to 0.5mm when meshing in Revo Scan.
Wash your face well , apply anti glare NYX Studio Perfect photo primer to get rid of any reflections from skin and oil, oily surfaces diffuse the infrared light pattern , the photo primer works like a polarizer filter , feature mode at High Accuracy . Fuse and export *.ply , import to Revo Studio and clean out the point cloud from all overlapped points , when clean, mesh it at the highest option what is 7 , after meshing you will get raw mesh with some little level of noises , it depends how well the face was scanned to begin with , after that use the Smooth option and choice the denoiser, use it as many times you like until you archive the desired level .
When you scan anything , make sure you do not overlap the same regions over and over again as it will be a beast to clean it out and tones of noises will show up after meshing that actually will cover all the fine surface that gonna be lost .
Regarding color per vertex , I am using Zbrush to project color per vertex data to a remeshed mesh with my own UVs .
Another user in this forum follow my workflow with a great results as well , the problem with the Revo Scan Textures is that the textures are not always generated perfectly , if you overlap the scan it will create horrible results not to mention if the light is not even . Not my favorite
here are textures from the color per vertex data and small workflow with retopology in Zbrush
I don’t know man.
I’m scanning in a light box and I’m getting individual hairs on my fingers on the texture - the small light ones.
Even with point cloud interpolation in MeshLab - 23mil polys - the hair just shows up as darker splotches in the vertex color data - that’s why I’m so hell bent on keeping the RevoScan texture all the way to retopo
Sure the scan overlap sometimes messes things up, but that’s why it would be great if we could do the retexturing process in RevoStudio after cleanup.
That’s what you do with other scanners.
Sure they are way more expensive, but I’m pretty sure the hardware bares the most of the cost. When we’re down to software, I’d expect the same features in RevoStudio
The Revo Scan textures are fussy , there is not much of sharpness to it , but the color per vertex data generate exactly the same quality … I use Light box too , and I tried everything , sometimes I am lucky , and sometimes it just a mess , until Revo Studio don’t provide texturing I am continue to using color per vertex to texture in Zbrush .
look bellow . not much difference
ply color per vertex
Well, an apple doesn’t really have small details
I’m going through the project data structure - it’s actually quite simple.
I think there might be a way to reverse engineer the process and write my own mapper for the mesh.
I’ll just need to find the internal coordinate system through trail and error, as I think they are using one for all scans - that’s why some of the exported models are quite away from 0,0,0
I found today the mask source for the texture projection , I used the data in Meshroom and recreated exactly the same model from it . What make me thinking , is it really just a fancy photogrammetry style texturing after all ?
you need convert the .img files from the cache to jpg
The original file under cache , it only appear when scanning with color data , no wonder the textures are so fuzzy at that resolution .
Jeff already asked you about this but I didn’t find your answer. Maybe I’ve overread it. Would you please explain how you were able to achieve this?
I’m not able to do further steps with the model in Revo Scan anymore after importing it back from Studio.
And that is correct , since normally you can’t , but if you scanned with color , you will have the fuse files in the project that includes the color per vertex data , so you have to work with point cloud , you can replace the fuse ( point cloud after editing ) and use the batch processing to generate new obj and color textures since all data is still available in the cache folder under your project .
Just don’t generate any obj or textures in Revo Scan before edit , only fuse the scan data , import it to Revo Studio, edit and save it back to the same location and run batch processing to mesh it and create textures .
The importing back function in Revo Scan is only to continue scanning where you left last time , because Revo Scan allows you to do that even if you already meshed the scan , you can always return and add more scan data , it will load back the saved by the software fused point cloud and you can continue . This function is not for importing objects from other software , only for the already scanned data in your project folders.
Ah okay, thanks for the detailed steps but just for clarifying:
- I do a scan with color
- I fuse the point cloud
- I import the fuse.ply to Studio
- Only make changes on the model while in Point mode. I shouldn’t mesh the model
- Export/Save and overwrite the changes to the existing fuse.ply. Do not save to a new file!
- In Revo Scan I then see the changes made in Studio immediately and then I’m able to do the further steps up to creating a texture.
For this process, the scan doesn’t have to be active in Revo Scan. I also can continue working on the file if I open the project later on without losing any of the data.
Is this correct?
That is correct , if you generate textures once , you can’t create again , unless you remove the mesh files from the project with the textures then maybe you can .
The only data you lose is the data you altered in Revo Studio . Make sure you save the original fuse.ply files before overwriting it just in case .