Photogrammetry for Shiny/Transparent Objects
ฝัง
- เผยแพร่เมื่อ 18 ก.ย. 2024
- This video shows how using a simple coating such as talcum powder, spay chalk, or, in the sample shown, 3D scanning spray, can make it possible to capture photogrammetry models of shiny and/or transparent objects (that would be impossible to scan otherwise).
This video is NOT sponsored, but if you're interested in learning more about the evaporating scanning spray used, here's a link to the site I bought it from:
visionminer.co...
Short and informative! Thank you!
How the model can be useful afterwards since the texture (albedo) is distorted? From model to model can take some time to hand texture the model and I am not sure if such process worths the effort and cost of spray since the texturing must be done by hand I suppose? Any workaround on that aspect? Thanks!
Nice... Thanks for the video. but it's for transparent objects. But maybe you can suggest something for metal things (in my case swords)? I want not just to make 3d scan of the shape, but also to get real textures, but spray will cover them
I need that mousepad...
what camera are you using?
I think I was using an old Canon 1D Mark IV for this one.
Really interesting! Intrigued by that spray product. So cool that it evaporates. I’m a bit shocked that the results were so bad - I can’t figure out why exactly. I know you said we could put the result into maya etc but honestly it would be faster to take isometric photos (front back sides top bottom) and model from that. :) Looks like you’re using “reality capture” software - maybe it’s just not good in this particular use case? I wonder if their competitors would do a better job.
Btw with sunglasses I’ve had better results than these even with no cross pol and using an iPhone app 😂 I really want to figure out why reality capture is failing so hard at this task! 🤔
@@virtualfilmer Nice. Maybe I'll have to give my iphone a try sometime.
I ran these same images through Agisoft Metashape as well and the results were less effective than RC (a number of gaps in the top bar and temple pieces and generally warbble-y all over). RC and Metashape were the only tools I tried on this set, though.
Thing is, I've been doing photogrammetry since like 2004, long before there was any hope of getting usable geo from the solve, so have this 15-year-plus habit of always doing re-topo over results that's hard to shake. Once I've got a result like this, with everything I need to to "trace" for a clean topology model, I'm good to go and any further effort for refining the result starts to bleed off the efficiency gains.
I really like photogrammetry as an assist on objects like this that are curved in many ways 3 dimensionally. I'm sure there are loads of great 3D modeling artists that could knock these out from isometric shots... but I'm not that guy .... I'm just addicted to tracing a result like this with some NURBS and knocking out the geo from there.
Plus, these will need to be rigged with joints (non-Nanite geo) to operate the temple pieces, so this was going to get re-topo'ed no matter the result anyway. :-)
but for the shiny bin bag it might be ok.
$40 for a single can? I'll pass.