What is the best poly mesh quality for 3rd party humanoid or animal for clothing?
Hello,
I am wanting to ensure there is no secondary smoothing on a humanoid mesh. What is the best density, type of shape, and level needed for clo? I know gaming is low poly, but apparel needs accuracy
-
Well that's not exactly true.
Human shape is by definition varied within a limited scope of general 'sizing' schemes. And that is what drives fitting of commercial clothing. So to say apparel needs accuracy is not so much about the human form but the context within garment sizes and end market demographic for human shape. That is really your starting point. And from that demographic of human (anthropometric size - eg: 5th - 95th percentile and then slicing and dicing that down into market specific scope) come the fit range of garment graded sizes at general tolerance (often in ISO/US/UK/AUZ/JPN 4cm size increments) and at basic sloper (simple tailoring formers for the torso and limbs that drive the garment patterns. So in terms of what variable human form (that is 'finite') you get to wear a specific sized garment (with a general size swing of 4cm) it's not so much about accuracy for any incoming human form.
It's what block sloper tailoring form you used relative to your market demographic of human scope (body-shape). For example if you look at many of the top 10% of brands that sell online > they target a basic human form that follows chest/bust > waist > hip profiles that might vary across brands - hence brand loyalty for fit by many consumers. As their basic bust / waist / hip size falls loosely into that companies tailored sloper for all basic patterns created across their organization. So in that instance for commercialisation and sales the most crucial asset is targeting the correct end user tailoring form and then specifying the range of sizes and break points (tolerances) so that the most of their patrons will fit their garments. What body shape walks into the fitting booth is then down to how well they scoped out their end market of anthropmetric variance. So having a avatar that is accurate is not really about a scatter gun spread approach to 'accuracy' or even a tight focus on polycount for an avatar. As you can fit a very low poly avatar that represents a 4cm swing quite well and still be well within tolerance of any given garment size. And that does not include additional elements like spandex, elastic, or stretch fabrics.
If you purely want to know what avatar resolution to use for testing any CLO3D garment that is simple > the same as the CLO3D mesh density is fine. BUT what you really want is to understand how you test a garment tailored block at the fundamental level of market demographic. And that is an entirely more complex question. And it is perhaps the most relevant thing you will ever need with regard to what range of avatars to test with. Because it is deciding on what avatars, with shape profiles ( silhouettes ) that you really need to focus in on. Not poly count.
A low poly count tailoring form is often suffice to get general 'fit' for patterns that can be subdivided to smooth the surface topology, however posture, size range variance across critical dimension (bust/waist/hip/torso length) is about creating a database of avatars (fit models) that are manually tested or auto-fitted to check your basic tailoring block (assembled) and avatar former are appropriate for your likely range of end customers that will try on your garments. So with CLO3D testing your garment graded size on a target (digital market) is when you can use a good human model. It needed be to high poly as that will only slow the process down with little benefit to end results - testing fit. So to do this well, you should optimize all your digital assets appropriately for end use. And not get to hung up on needing a 1 million poly human avatar to test fit a CLO3D/MD garment. That would be really in-appropriate as your tolerance is actually quite large for garment fit. What is best to focus on is how many digital humans do I need (along with their morphs to size variance) to confidently test the tailored fit of my garment size scheme. That's more valuable and also more important than any one high poly avatar.
You may also import scanned humans into CLO3D and fit them to a CLO3D avatar and save that as a morph default. You may also adjust any CLO3D avatar to a new general size and posture using the avatar measurement editor, and save that market avatar profile (as a human fit model). That is how to use CLO3D avatars sensibly. They have sufficient surface poly count and resolution to smartly get a good test of any range of garments. Where it might fall down is how the end user (marketing fashion specialist) mis-interprets what digital fit means in context to how they should structure their programme of digital assets and tailoring/former work. And this is often where fashion digital tools get a bad rap for inexperienced users whom have no fashion market 'fitting' experience. It cannot solve the issue on how to break down your companies end market (user fit profile) - that is something you need to structure and then yes - CLO3D has all the tools to complete that work including photoreal 'house fit models' accurate to the millimeter (eg: scan). You don't need overly complex human scans or high polycount avatars - what you need is assets that are structured against 'fitting' in the context of your end market that will buy your clothes. And in doing so they will form the strongest and most crucial commercial aspect of how you confidently design digital garments now and into the future. It's such an important part of the process to understand that sadly many default to using 'Alvanon' as some quick fix to the hassle of researching and doing this work yourself - and it's worth learning how all this works commercially because the fit profile you place into your tailoring forms and then the digital avatars you test on,can be your best resource to deploying digital sample making with confidence - because you know how each asset works and is constructed and toleranced.
0 -
Thank you for the detailed message as much of this will help a newbie. I simply wanted specifics for avatar as there are some interoperability aspects of 3d you may be unaware of that do have some implications to anthropomorphic accuracy from end to end interchanges between apparel and gaming that I am not permitted to speak to from a NDA perspective. This should be a published facet anyhow.
My question really doesn't have to do with fit as that is something different than the poly/ mesh of a avatar. For example, on program may require a triangulated poly, while another program can accept all sorts. Again, there is more to this question, but at the moment I can't share. I will say that the mixamo animation is poor for use if a person wanted to use a digital avatar for fit, depending on its mesh structure. Mixamo dumbs down the bone rig and we also have some other factors I can't speak to per NDA. Just something to be mindful of. I find the MDD animations tend to also be the most accurate for use in what CLO accepts. Just maybe a hint to look for what breaks there. I still would allow a user to have mixamo, but tell clients there are use cases for mixamo that do effect fit accuracy. Bone accuracy must be the same per customized rig as it is done out of purpose. This is a big bug in the system as well.
A poly mesh really has little to do with how a person or collective group of people grow and change as that is more of the math, physics, and science of shape and fit. To me that is anthropomorphic and has no relation to the mesh. Demographics again has nothing to do with the specific question asked, but fully understand that as well, If i had a question on demographics I would have looped that into the question.
The mesh does however get inaccurate based on other factors of 3d development if done wrong, or if a blind spot is not considered.
0 -
Some research links for you: I trust that these might be more appropriate technologies for you to look at.
https://bodytalk.is.tue.mpg.de/
https://www.russian3dscanner.com/
https://bodytalk.is.tue.mpg.de/
Try the links out above, you might find more appropriate technology and research groundwork has already been done, commercialized and in use. Good luck. :-)
0
评论