I get quick response for my mesh with 18M faces and 9M vertices, colors and textures, a 2GB .OBJ file.
Do you have enough RAM? I have 128 MB and never see all of it in use so there is never paging to disk while working on the mesh. Look at the performance monitor while working on your mesh. Is more than 16GB in use?
When I work on my 100M faces mesh then Rhino is much slower. Also the first time I select the 18M faces mesh I have to wait 15 sec for it to be highlighted but after that it is fast.
I do have many cores on my machine (>8) but I do not think that makes as much difference as having enough RAM. I also use a M.2 SSD with PCie 3.0 X4 interface with a BW of 3GB/sec read and 2GB/sec write but if you have enough RAM Rhino will not page to this “disk” so it should not make much difference once your mesh is loaded into Rhino.
Rhino can provide good performance when working with large meshes (>10M faces). So if it is not the RAM it could be other issues like the display setting on which I can offer no advice.
Back to using meshes and the speed thing is still a pain, I see I posted on it.
I have a mesh, its an obj imported from Agisoft Metashape.
Open double precision polygon mesh: 2141079 vertices, 3882715 polygons
obj file is 337,884Kb
it saves as an .3dm file 187,811kb and takes a good 29 secs or so to save.
attached my performance tab during a rotate then during two saves.
As I am about to take this pc and install win10 64bit pro onto a new 250Gb SSD (replacing the existing 250Gb SSD C drive), same hardware, I am hoping v7 is better, as this is hopeless.
Turn off file compression (V7) and it will save faster.
Saving a 350 Mb point cloud file here takes about 5 seconds (Win 8.1/SSD, computer from 2014)
Saving a 1.8Gb 30 million poly mesh file (uncompressed .3dm) took about 20 seconds here (same 2014 computer)
Dynamic display performance (on-screen) will be largely dependent on your graphics card, but V7 is significantly better than V5. The above mesh file is surprisingly snappy even on my old 980Ti - OK, it’s just a simple terrain mesh with nothing fancy.
Use only wireframe and shaded display modes. Go to Options>View and tweak display to trade speed for quality: turn off antialiasing and turn off shading in the shaded mode.
Hi,
my poly is 3.88 million and obj was 0.33Gb and thats 30 secs and yours for 30million 1.8Gb 20 secs.
Every time it does an auto save I have to wait 30 secs.
I can but hope that v7 is faster as my hardware spec is ok ? its PC ram at play.
what of my performance chart, as asked for by Terry, does that point to any problems ? is available ram ok ?
should I turn off all other progs, such as photoshop ?
I have both running as both I use when working in Rhino so turning it off is against my needs.
I need to be working with the hi res textures, as shaded has a far less useful texture to work to, and wireframe is just a complete mass of dense black triangles, cant work there at all.
12 secs to do project with a straight line projected to the mesh.
Hi,
I am not editing meshes though, but using them as reference to use all the tools Rhino has, and I am familiar with, to create Rhino models, surfaces etc, so its all about how Rhino handles these meshes, as 30 secs every time it does an autosave, 20 secs for a rotate etc, is impractical.
So looking at my spec and comparing that and the performance .jpg should it be better ?
I have a load of geometry objects to create from meshes coming my way, its the workflow I have been working towards, and currently V5 and my hardware spec is VERY SLOW.
I still wonder if Rhino is in fact capable of working on meshes as a reference to make rhino objects.
What Andrew may be getting as is you should decimate the mesh to a smaller # of triangles that will still preserve the detail that you need.
However, many of the tools that decimate meshes may also affect the texture, which as I understand it is critical to what you are doing.
Its likely with the the right application you could reduce to 1/2 or even 1/4 of the original number of triangles without losing significant detail in the mesh. However, at the same time you need the color texture associated with the mesh to be unaffected.
Hi,
V5
I think the mesh count on flat areas versus detail areas is the key.
No system seems to exist, correct me if I am wrong, that gives detail where needed and far less polys where they are not needed. If I reduce the mesh count its universal, and the detail suffers.
I tried it in metashape, decimate mesh, and the screw threads etc melted away, when I just wanted to have less polys on what were smooth areas, which had slight pitting from age. Even if one could mask the areas somehow telling the prog what to decimate, would be useful.
Then having done that , run the make texture command.
In Agisoft the 1.7.4 metashape is bugged and blurs areas of the mesh, Photoscan 1.2.6 does a perfect job. so I havw to save as an earlier version open in Pscan and then texture it, then bring it back into metaashape. They havent fixed that bug yet after months.
As for selective mesh count, no chance of that from them.
The mesh looks overly meshed, I dont need all that.
If I could intelligent mesh it, not sure how I would then get it back in for texturing. Something I need to ask them about I guess but answers can take many weeks, not the 2/3 days they say.
despite that, something is wrong, comparing to Helvetosaur :-
my poly is 3.88 million and obj was 0.33Gb and thats 30 secs and yours for 30million 1.8Gb 20 secs.
Every time it does an auto save I have to wait 30 secs.
is there a test mesh we can all time for saves, see how it varies.
There are applications that decimate based on the local curvature, providing fewer triangles in flat areas.
Geomagic is excellent for curvature based decimation but it comes at a cost.
Meshlab can do a reasonable job considering that it is free, but I’m not sure it will preserve your texture. It may be able to, I’ve just never tried decimating a textured object in Meshlab.
Some of the mesh modeling apps have ways to decimate triangles and maintain the texture … they have to generate new UV coordinates to correspond to the decimated mesh.
Your computer system and HDD write speed vs mine
Rhino V5 vs V7
Actually, I’m not saving the file to my SSD either, as only my system and programs are on that, I am saving to a good ol rotary 1TB disk (that is also 7 years old).
Hi,
perhaps I need to try meshlab, intelligent decimation, and see if it preserves texture, else I will have to place the mesh back into Photoscan and generate texture, and pray.
I feel its about time that mesh making progs woke up and had the option to generate intelligent meshes with finer mesh in areas of detail,. Crazy having to have fine mesh everywhere just for the detailed parts.
I am also saving to a 2Tb spinny HDD 1 year old, WD green 5200 rpm.
OS is on C drive Samsung SSD 250Gb as is Rhino V5.
every so often I have to wait 30 secs as an autosave happens.
Well,
I am now with a new PC as of a year or so ago, all SSD drives, built for speed, for Rhono V8 and Video editing
64Gb DDR4
Nvidia 3070
holo test very good etc etc.
and I open a 5.6Gb obj sent me by a 160 camera bank photogrammetry shoot unit, and after 5 mins Rhino V8 is still trying to open it.
I install latest Blender (not used before) and in under 1 min its open.
something tells me Rhino , whatever version, is incapable, with meshes, what with the speed it operates at on those I do work over.50 secs doing a save and so on.
My GUT INSTINCT is Rhino was never designed from the outset to work with meshes and the engine is incapable, compared to mesh progs like Blender.
That I need to import meshes and reverse engineer them and create Rhino type objects from them, I am in for a real struggle.
and I had this PC built so as to be able to use V8 on meshes.
Rhino is primarily a NURBS surface modeler, has been from the beginning. It can do lots of things with meshes these days, but it will never compare to a program specifically designed for meshes, especially for extreme cases.
Have you tried to decimate it using Metashape … I believe you are using that?
To retain the texture, you would have to transfer it from the higher resolution model to the lower resolution one.
Windows 11 (10.0.26100 SR0.0) or greater (Physical RAM: 128GB)
.NET 9.0.5
Computer platform: DESKTOP
Standard graphics configuration on OpenGL
Primary display: NVIDIA RTX A5000 (NVidia) Memory: 24GB, Driver date: 5-12-2025 (M-D-Y). OpenGL(4.6.0 NVIDIA 573.24)
> Accelerated graphics device with 4 adapter port(s)
- Windows Main Display attached to adapter port #0
- Secondary monitor attached to adapter port #1
Secondary graphics devices.
NVIDIA Quadro K2200 (NVidia) Memory: 4GB, Driver date: 5-12-2025 (M-D-Y).
> Accelerated graphics device with 4 adapter port(s)
- There are no monitors attached to this device!
OpenGL Settings
Safe mode: Off
Use accelerated hardware modes: On
GPU Tessellation is: On
Redraw scene when viewports are exposed: On
Graphics level being used: OpenGL 4.6 (primary GPU’s maximum)
Anti-alias mode: 4x
Mip Map Filtering: Linear
Anisotropic Filtering Mode: High
Vendor Name: NVIDIA Corporation
Render version: 4.6
Shading Language: 4.60 NVIDIA
Driver Date: 5-12-2025
Driver Version: 32.0.15.7324
Maximum Texture size: 32768 x 32768
Z-Buffer depth: 24 bits
Maximum Viewport size: 32768 x 32768
Total Video Memory: 24564 MB
Rhino plugins that do not ship with Rhino
C:\Program Files\Rhino 9 WIP\Plug-ins\ConstraintsUI.rhp “Constraints UI” 9.0.25154.12305
I tend to insert heavy files instead of importing or opening. While the initial time to get the file into Rhino might not be quicker, saving will be faster and the file size is lower. It depends what you do of course.
Hi,
is INSERT better than IMPORT, better than OPEN , for a 5.5Gb obj ?
for decimation, Blender has allegedly intelligent decimation, one can protect areas needing finer detail.
If thats so I like that and have wished for it for ages.
However the basic fact is 5mins and counting versus 1 min same obj, Rhino v8 versus Blender.
Hi, V8,
just took 15 secs to go from perspective to Top view, with a mesh showing.
I was hoping V8 would be QUICKER. with obj’s etc.
I NEED RHINO TO HANDLE MESHES FAST !!!
I had all my hopes on V8 being so. I just get the feeling Rhino is incapable.
However maybe import is bad practice.
So I need to use Insert and browse to the obj instead. Is that so ?
I had used import obj and browsed to the obj, then V8 allocates the raster image to it, whilst V5 one had to browse to the image.
If i were to try insert for the exact same obj, would it locate to the same position as the existing.
It took me ages to orient it and scale it, as I have three of them that must match, being the same object, one with rulers in the result from Agisoft. matched to have the same spider shit, the same rust spot exact same place.
I can’t redo it all again.
whats the best way of getting them all to match position using insert ?
I need to test my Rhino V8 on mesh handling.
Is there a test file we can use, and time how long it takes to do a set number of tasks etc ?
(not holomark)..but a mesh made by McNeel , just one item, to turn on and off, go to top view then front view etc, project a line to, actual working commands etc.
