← All ideas
  • Published 18 Apr 2022

Why Blender would need mesh wrapping features?

Hello Comunity !

I'm starting this topic to share my view regarding all the advantages that mesh wrapping can bring to modeling workflow and hopefully start discussions to build a worklow for blender. As a senior character artist in the video game industry, I have seen this technique mostly used for Character Production, but it may be usefull for other fields, and from my previous experiences it as drastically improved my production time. It is now an essential part of Character production in Video game, VFX, and animation industry.

Small Disclamer here : Wrapping is now essential part of Character production, but Blender would first need a more robust manual retopology tools and workflow before it, which is already in developpement

What is Mesh Wrapping ?

Mesh Wrapping allow you to Wrap/Conform/Morph a mesh onto another mesh with a different shape, polycount and topology. You will most likely want to wrap lower polycount, topology controlled mesh to a higher polycount sculpt or scan. This wrapping process only affect Vertex position which is a key aspect, and you will see the benefits later in this thread The leader and only serious wrapping software in the industry is actually R3DS Wrap3D . Because there is not much alternative out there, I think it is important to understand their workflow and why it becomes so popular to better integrate this workflow into Blende's environement and UX. I will also go into details of the R3DS workflow later in this thread

Why Wrapping Meshes ?

Manual Retopology is a long and tedious process, but the only way the have 100% control over the topology, flow, and dentity of a mesh. Even if a lot of automatic retopology tools exists, which also allow you to paint the denity you want, or to trace line to orient edge loops to guides the topology flow, you will not always get the exact result you want. Wrapping workflow starts to shine when you have to produce multiple Characters : you can spend time on the first retopo making sure it is perfect for your need, make UVs, make it approved by Riggers, Animators, TDs and be sure anyone that will have to work with it is happy. And then Wrapping this approved topo onto all of your Characters which will save everyone's time and energy. As long as the characters have the same morphology, this technique will work perfectly, but you will not be able to wrap a Human topo onto a octopus scupt obviously.

So the main benefit is time saving which allows you to get better quality thanks to more iteration you can spend on one single topology

The second main Benefit is topology sharing which is most transparent, but allows many interesting things if it is well used later in production

All mesh shares the same UVs

-- you can use one models texturesets to another one which can help you create variation easily -- you can use one textureset as a base to texture anoter model, and save time easily

All Mesh shares the same Vertex IDs

-- you can morph two models to get a third variation -- you can create a new variation by morphing part of one model into another model ( morph the nose of one head into another head) -- you can easily Transfer High Frequency Details from one mesh to another -- Many tools can be designed as Vertex ID based --- we could easily transfer vertex groups, bevel weight, mark sharp etc.. based on vertex ID --- we could easily create rig based on Vertex ID. For exemple a facial rig generally needs a rigourous bone placement on specific vertices ( around the mouth, eyes, nose, forehead etc ... ) The rig could be created and skinned once and then transfered on all the other meshes, snapping bones on the same vertex IDs.

The list could go on and on. Basically if you know that all meshes are structurally identical it brings a lot of freedom and interoperability afterward.

The R3DS Workflow

It is a node based software ( a python version also exists ), and the main wrapping node allow you to have a split view of both reference and target mesh.

  • You need to click on landmarks one by one on both mesh in the same order.
  • If you placed landmark 1 on the tip of the nose on the reference mesh, landmark 1 should be place on the tip of the nose also in the target mesh. All these landmarks pair will guide the algorythm for wrapping
  • Click the compute button that will iteratively deform, wrap and relax the reference mesh to the target mesh's shape

Here is a small video to illustrate it.

This is the absolute basic, and it is the element that is missing in Blender at the moment. But R3DS comes with many other nodes/tool that make it convenient to use. Some of these tools are partially or already here in blender.

  • A node that help match scale and orientation of reference and target mesh
  • A way to disable a selection of polygon to be treated during the Wrapping process
  • A relax node, to reduce stetching after wrapping
  • A way to save and load points landmarks to avoid select them for each and every mesh you need to wrap.
  • a fix symetry node
  • a set of node to Relax, Project, move etc.
  • You can save an entire nodegraph and batch process many meshes through the same nodegraph for a consistent result

Blender Workflow Ideas

This is the most complicated part of this thread for me because the goal is not to "recreate" R3DS into Blender. It is more about find the workflow that would fit the best into blender environment. I'm also not sur to know the limitations of the software at the moment. But here is still some insights

Should it be Nodal ?

I would YES, it's so flexible, it can be artist friendly and there is already a powerfull nodal system in Blender

Should it be an independent Nodal editor ?

I feel like it could be implemented into Geometry node by his nature ( Mesh as input --> modified mesh as output ). But I'm not sure if Geometry node is able to have an iterative node like the wrapping node ( simulation ?). I feel like many nodes already in geometry node could be usefull durring the wrapping process, but most of them are a bit too low level to be usefull in this specific workflow. That's why I'm thinking about a dedicated node editor would be better

How to select Point Pairs ?

This one part that would need a time and iteration to find a porper way of doing it, and there is no similar workflow already implemented in Blender as far as I know. And because the selection can be long and tedious, it should be as fluid and seamless as it could be. A dedicated Workspace would help setting view up easily.

Thinking outside of the box

Instead of having a two step process:

  1. Placing points
  2. compute the wrap

The process could be more interactive, interacting with the mesh a bit more dirrectly and seeing the result of the wraped mesh in the viewport.

But it could be harder to replicate as a batch operation, if you have to wrap many meshes in a row.

Conclusion

Thank you if you have reached to this line :slight_smile: I have unfortuantely no solid design to propose but I wanted to put the light on this part of the character production pipeline that is not discussed much and would make blender a more versatile piece of software. I would love to read your thoughts and continue the discussion.

Tilapiatsu

91.7%

Development Status: Open

Upvote 11 Downvote 1
tilapiatsu
18 Apr 2022

6 Comments

koko_ze

I've seen a few Blender addons that emulate R3DS in a way with cloth physics. But having a native support for this inside of the retopology mode would be even better

tilapiatsu Author

Indeed there is the Soft Wrap addon that I've bought already.

It is a really well made addon, but it miss many thing to be pieline ready, like the ability to batch process multiple Wraps without humain interactions. And from what I saw blender wouldn't be too far from this point. Even more since the Blender Studio annouced that they want to put the emphasis on realism with their last project Heist and Wrapping is used a lot for retopo Scan data.

1
thinsoldier

Is there an open source project that does this kind of wrapping? If the blender devs have to recreate the main functionality from scratch it's not going to happen this decade.

tilapiatsu Author

Sorry I haven't heard about an open source alternative to Wrap3D. Thats why I was thinking blender could be the one :). I know Blender developers can be pretty busy, but I think it is really about making one toping being a higher priority and finding the good person to develop it. When I see how fast Sculpting in Blender have evolved the past years, I still keep the faith ^^

thinsoldier

ehh... i wouldn't use sculpting as a good example. Pablo Dobarro was like a miracle. He didn't improve many of the biggest problem areas with sculpt but he did add some cool things that were very specific to his personal workflow. Your typical zbrush user who makes dinosaurs and photorealistic humans every day didn't really get anything out of the sculpt development done over the last 2 years.

tilapiatsu Author

I think will not agree here. But it is not a big deal :) Before 2.8 Sculpting in blender was really complicated, barely usable. and after two years of developement you have something that can be used, and many people are using it as their main sculpting program now. This is a huge step, and even if not everything is perfect, and it still missing major features like sculpt Layer for exemple, It definitely going in the good direction from my standpoint.

And it is also hard to compare for me a Company like Pixologix/Maxon who sells thouthands of paid licences to almost all Video game, VFX, animation companies, with a entire team dedicated to "sculpting features" and Blender, open source software with more limited budget that have to split their ressources between many feature of a DCC.

It is not the point of this thread, so to refocus the discussion I would say that as a comunity we should try to help blender developers to focus on the most wanted features, and this is what I'm trying to do here by talking about mesh Wrapping. But sometime it is tough because everyones need are differents so choices have to be made.