Open Source Customizable Digital Human ~ Vitruvian Project

[hopefullyidont1 (Hope)]

So I think we can all agree that Digital Humans are hard to get right and they often fail.

Customizable solutions like Daz and Makehuman have existed since the early 2000s but they have often played catch up and have unrealistic standards of what is possible in design.

Metahuman has promised high quality digital humans however it is still in a closed beta and is for Unreal Engine only, nothing that can compete has existed yet…

Lets agree on some basics on what is possible to deliver in a customizable solution without needing to rely on additional work from users.

• A Broad range of sliders called morphs that represent Perceptible traits that can be modeled, (ex. Bicep Size, Jaw Width)
• A Basic range of Skin Variation, (Wrinkles/Aging, Linear Melanin Scale, Skin Bumpiness)
• A Fully functional rig (Blendshapes, IK/FK)
• Basic forms of Hair (Eyelashes, Eyebrows)

What is NOT Possible in a Customizable solution:

• A Full range of Human traits, perceptible or not (the Human Genome is vast and even if we account for only perceptible traits we will be working for decades)
• Fully Customizable skin, Skin is a multi layered translucent material faked with subsurface scattering.
• Fully featured library of hair, Curve data especially hair can get very expensive and balloons file sizes
• Cloth library, Same as hair but paired with full texture sets and vertex paints, even more expensive file size wise.

Q. Well, if there’s so many restrictions on what is possible in a customizable solution, what’s the point?!
A. The point is: You don’t have to do the generic hard work that comes in making a character, only just what makes the character unique.

Q. What is ‘Vitruvian Project’
A. The Vitruvian project aims to create a digital human solution that is Creative Commons Zero, effectively Public Domain, that way we can make a no strings attached solution to Character creation Woes.

Q. This Mesh looks familiar?.. Is it all original?
A. No, The Mesh was created by Olaf Delgado-Friedrichs (ODF) on Renderosity in 2011
However I had gotten his permission to relicense under the condition that the mesh is sufficiently different from the source material. Which considering how much has changed, I think is appropriate.

Q. Why is it in the Work In Progress page?
A. The Project rides on Community Support, So I need people to give advice

Q. There are visible seams on the textures, is that normal?
A. Yes, Unfortunately it is, I can’t say I found a solution that doesn’t involve simply converting it to UDIMs and then smoothing the color textures.

Q. Why ACES? Isn’t this supposed to be Blender Centric?
A. ACES was chosen because it’s a Color Managment standard that is most agreed upon, ACES is quite messy in blender so I use ACES-Helper for using textures, Unreal Engine uses ACES, Unity uses ACES, It just makes sense to go with an Open Source standard.

Q. Some parameters are broken, Is this normal?
A. Yes, Some Parameters have unintentional effects when paired with others, please report any unintentional behavior.

Features:

• FACS Rig (100+ Blendshapes)
• 100+ Morphs
• Rudimentary Tension Maps
• 8K textures
• Texturing.XYZ VFace Support
• 3D Scan Store Face Texture Support
• Metahuman Texture Support
• Digital Emily Texture Support
• Antonia Texture Support
• Ultra Dense Texel UVs
• Volume Preserving Rig

Requirements:

• Rigify Add-on Enabled (The rig is based on Rigify)
• Blender 3.4 + (Could be as early as 3.0 but needs testing)
• Screen Space Refractions enabled (EEVEE requires this for the iris to be visible.)

Technical Recommendations

• 8GB or more VRAM (Downscaling textures is possible)
• ACEScg (Filmic will wash out textures, and ACES is standard in Game Engines/Maya)

Downloads Include: Packed Blend Files with all (4K) textures, One without textures and a folder including textures
Download:
https://drive.google.com/drive/folders/1ZjGy5TeTX-wgMIQvpUsRTK-LV22miXUb?usp=share_link

Discord:

Приєднуйтесь до сервера Discord: Blender Character Project!

[hrijul_dey]

very good models very good lighting. just a better advertisement required

[hopefullyidont1 (Hope)]

What do you propose?

[hrijul_dey]

make films or video content with mocap

[StoryHierarchy (Ebenmosi)]

Easier said than done, considering this is a one person project.

[Jvry (Tremors of the Toads)]

This may be a dumb question, but shouldn’t colour management be decided by the creator of the overall composition, not by the individual pieces used? As such should not texture details hold their own independent of any particular colour management method used? Locking into the current vogue “standard” rather than letting people work with what suits their own use cases (which may not include game engines or big studios) seems a little odd to me, but I’m still very much a beginner in practically everything, so I’m sure I’m missing something.

[hopefullyidont1 (Hope)]

@Jvry

This may be a dumb question, but shouldn’t colour management be decided by the creator of the overall composition, not by the individual pieces used? As such should not texture details hold their own independent of any particular colour management method used? Locking into the current vogue “standard” rather than letting people work with what suits their own use cases (which may not include game engines or big studios) seems a little odd to me, but I’m still very much a beginner in practically everything, so I’m sure I’m missing something.

I think you misunderstand, Color management is not color grading nor is it set in stone, It’s more the meaning assigned to colors. How red is red, how yellow is yellow, Someone can easily use Filmic or AGX if they wanted to, They would just end up with different looking textures than a standardized ACES solution. Which is why I chose it, because it’s the most consistent.

[hrijul_dey]

nope it doesn’t have to be a one person project.
the community can jump into the effort.

[hopefullyidont1 (Hope)]

I mean nothing is stopping you from making a demo. Almost everything is there for a full animation.

[Nanoglyph (Lee)]

You’re off to a good start! And since you asked for advice:

Overall, the textures look pretty good, but they need to be seamless. I don’t have a lot of experience texturing in Blender, but don’t think you need UDIMs for seamless texturing. Soft Clone stamp tool over the areas seem to work fine even though the UVs are overlapped as long as they’re different materials, you just need to watch out for the fact that some of the areas the colors don’t match up particularly closely (example, on the light skin the neck is a darker tone than back). Select the UV island to move, use G+X+1 and you can move it to the proper location for the second UDIM, G+X+2 for the third, and so on.

Alternatively, you could temporarily make a second UV set that is UDIM based, and then fix your textures your preferred method. Additionally, I think the eyelashes are too short even for male figures, and the eyebrows are decently groomed, but not quite shaped correctly. Their thickness is too uniform, they should be thicker towards the nose, and thinner towards the ear:

I find shorter morph names are more user friendly in Blender - otherwise you have to sacrifice more screen space to widen the side bar so you can actually read the shapekey names. Removing _Essentialist from the morph prefix would help. I get the theory behind it, but I don’t think it’s a necessary or helpful differentiation, simply going with Gender_Female, Gender_Male, Race_X, etc. is sufficient. Technical anatomical accuracy for morph names is not as artist friendly as simpler names (Example: Chin_AnteriorMandibleWidth is way less human readable than Chin_Width. Character creation will be faster if people can tell what a morph does by looking at the name, without having to stop and think or look up technical nomenclature or just dial it in to see what it does).

Re-ordering the shapekeys so the expressions are at the bottom of the list and the actual character morphs are closer to the top would be much more useable as well. Having to scroll down past all the expression morphs to the useful face morphs is frustrating. I’d also re-order the morphs to be grouped more logically as well (ex. all chin morphs together, all partial face morphs in a separate section from all partial body morphs). I haven’t messed with the face morphs much because currently they’re disorganized and a lot of them have names that make it unclear what exactly they do, but I think you need more nose and mouth morphs. I realize this is still a WIP though.

I don’t find the race morphs particularly convincing, some of them look pretty odd - the European morph especially. The male morph however was a much needed addition to the Antonia Polygon mesh to make it more useful as a general character creation tool. Having a more generic starting point for the female mesh is also useful as well.

However, both the neutral base and body/gender morphs have artifacts, and I’m not sure they really work that well together. They may need further adjustment.

(And, uh, I hate to ask but you’re absolutely 100% sure you have permission to relicense the CC-BY Antonia Polygon by ODF/Olaf Delgado-Friedrichs as CC0?)

[hopefullyidont1 (Hope)]

Honestly I might take the time to convert everything to UDIMs, I can fix eyebrows by weight painting density.

I was in an email correspondence so I know that I can relicense it, But that he wasn’t necessarily a fan of the idea, but if it was different enough he would be okay with it.
I’ll look at renaming morphs, I agree it may be alienating for users
Do you have any examples of what morphs you like to see?

[joseph (J)]

@hopefullyidont1

I was in an email correspondence so I know that I can relicense it, But that he wasn’t necessarily a fan of the idea, but if it was different enough he would be okay with

Actually… only the publisher (creator) can relicense a CC asset, and it’s not a case-by-case thing, the publisher can’t give a CC0 license for your case and a CC-BY for everyone else. It sounds like the publisher isn’t aware of this, but it is a requirement of using CC licenses

[Nanoglyph (Lee)]

Good to hear. I tend to be cautious about licensing.

As for face morphs, length/width/height/angle/depth/etc as appropriate for the different parts of the face. Being able to adjust lip thickness for the top and bottom lips specifically would be helpful, as well as being able to adjust the curves of the lips, and peak height/width and bow (pointiness of the bottom of the middle of the top lip). For the nose, I’d include width morphs for the nose tip, nostril, and bridge, such as angle/length/depth/whatever as appropriate. You don’t need to replicate 100% of human variety with your morphs of course, and too many morphs can be overwhelming as well, but noses benefit from a bit of granularity I think.

[hopefullyidont1 (Hope)]

Yes, I don’t think he was aware of this but I have no idea how to resolve something like that, because flat out asking to relicense the original which he worked on for years is not respectful than asking for permission on a derivative, IANAL but it seems more an issue of what the author has approved than necessarily the license they chose. It’s messy but I believe what I’m currently doing is appropriate. Worst case scenario it’s CC-BY which I’m okay with.
Though that means only the mesh itself, textures and shaders being independent.

[hopefullyidont1 (Hope)]

I’ll look at it, Those types of morphs aren’t too hard to implement.

[hopefullyidont1 (Hope)]

Anyways I’ve updated it, it fixes some concerns mostly some names and ordering but some stuff like UDIMs will take more time.

[Nanoglyph (Lee)]

@joseph

Actually… only the publisher (creator) can relicense a CC asset, and it’s not a case-by-case thing, the publisher can’t give a CC0 license for your case and a CC-BY for everyone else.

From the Creative Commons FAQ, I think the creator CAN grant someone the right to use the license under separate terms from the main license:

Can I enter into separate or supplemental agreements with users of my work?

Yes. CC licenses are nonexclusive. Licensors always have the option of entering into separate arrangements for the sharing of their material in addition to applying a CC license. However, those different arrangements are not “CC” or “Creative Commons” licenses.

This is an odd edge case, but from that page, since the creator can offer separate or supplemental licensing terms, provided the supplemental licensing doesn’t conflict with the original licensing except in cases where more permissions are being granted or license conditions are waived, I think the creator CAN privately give someone permission to release a separate derivative version of their work as a separate license.

(Saying “you can release your version as CC-0” is arguably different than saying “this is CC-0 only for you.” Though I’ll be the first to agree this example is in enough of a gray area that sticking with CC-BY may be safer).

[Nanoglyph (Lee)]

It looks like the rig bones may be aligned with the female shape more than the middle shape. I think the male shape would bend a bit better if the rig had wider shoulders (Aligning it with the neutral base is probably safer than aligning it with the male though).

[hopefullyidont1 (Hope)]

I can look into it.

[hopefullyidont1 (Hope)]

There was actually something I planned for Reom that I could start working on but it can’t be mocapped,(Besides Facial Expressions.)