Since 2016, I have been on a path to understand 3D Design for apparel, gaming, and meta uses. On this path I have been able to understand that there needs to be a long vision for use, to correctly map the processes needed for the brand. Because 3D in apparel can't be ambiguous in the product development stage, there are some flaws and nuances to the process, depending on the brand products. Some questions to ask are...
Do I make a product that can be effectively rendered in 3D?
Do I want to be able to create a product out of the design?
Will the company sell items in meta verse or gaming?
Will the company want items in AR/VR media?
Do I know where 3D breaks down?
What is the cost?
How much time does it take?
What files and items do I need to also create above and beyond the patterns?
Fatty Tissue & Breast
For example, the 3D design platforms, and scanning tech can have a hard time with any type of fatty tissue overlays, and landmark measurements found between them. For this type of problem, the use of 3D may be used to ideate the concept a designer wants to create, but the actual pattern may still be done in the standard 2d and drape methods. Bra design can be very similar. This is also a hard item to gain accuracy within 3D because we don't have the soft tissue humanoids fully tested and fleshed out to perform with accuracy of fit. Smaller breasts can be done in 3d, but larger breasts become more inaccurate.
Posing and Animation Used for Fit
Posing and animation can be helpful but can also be challenging based on the actual rigging of the form and the fluidity or programming of the motion capture. For a new to 3D designer, they may think this is an easy process and not understand that some aspects in 3d design should be used to market and drive consumer interest. While the actual fit form maybe different. For example, I will use a fit form for creating the clothing, but once I animate or want to move the item into a better raytracing and modeling program, this will change. It is giving me the simulation of what I want to see. For example, the armpit zone, legs and crotch may have some deformers for the animation to give space to the clothing for the movement simulation to occur. This is why I use animation purely for the esthetic and shock/ awe value. Gaming field is okay with this as the system data needs to be lean to allow for fast moving controls. Also, they are in ambiguity and not in a product development region that requires accuracy.
With posing, it does depend on the mesh accuracy, but also bone rigging. If there is an auto-rig that changes original bone rig, the posing may not be as accurate as the real life. Auto-rigs make work fast and cheaper. They also help the program know where to place items in accordance with a set up structure.
Unity has come out with an excellent mocap suit, but it's important to note that the movements will still need cleaning and may not work as well with other programs or computing systems that differ from one another. Animator designers that work in CGI film are generally the best for this accuracy.
Computer System Architecture
Does it really matter to have a good system with a strong CPU and CPU... Yes!!
The system architecture is highly important to the quality outcome of fit and fabric simulations. The other component often missed is also the mother boards & your actual memory. We need exceptional systems to allow for a quicker workflow, but also how the actual rending comes out. For example, I can render the same outfit and have an exceptional image based on a 3090 RTX GPU, and 24- 32 core CPU, and a motherboard specified for arts or gaming. My memory should be at least 4T. All these allow the system to do a better job with the physics and complexity of the design. The more layers and additions to a set of images and to a scene, requires more computational power. It's also important to know if the program is a CPU driven render vs a GPU driven render. If a company needs a team to work on many products, I advise having a company create the computers needed, so they are all unified. Also, monitors require a good Delta color qualification or pantone approved. This goes far in the assurance of similar output.
To put this in perspective, in can take up to 4 years to make a 85-minute Pixar film.
A team of about 75 animators worked on the movie over the course of roughly four years. Encanto took 5 years to create- one year alone to render frame and assemble for film production.
Frames Per Second: If each frame takes 30 seconds to render, then you will be able to render two frames per minute or need 720 minutes (12 hours) of render time.
The average frame (a movie has 24 frames per second) takes about seven hours to render, although some can take nearly 39 hours of computing time. The Pixar building houses two massive render farms, each of which contains hundreds of servers running 24 hours a day.
The future is getting better and the use of NeRF will greatly impact the speed of development, however it is still in testing for fit accuracies and apparel development. We may see a good outcome in 1.5-3 years from 2022. 2025 will be an exceptional year for computing power and 3D driven products. To learn more about NeRF, watch the video below.