DazToHue vs DTL
JQP
Posts: 511
Can anyone speak to the nitty-gritty comparison between these two projects?
DazToHue
https://mrpdean.gumroad.com/l/rpusr
DTL
https://virtuality.space/dtl/index.html
One point in DTH's favor, as far as I'm concerned, is that it works with Houdini. I have the free version of Houdini installed, and there is no free version of Maya (student versions don't count IMO) that I'm aware of. (I'm assuming DTH works with Houdini Apprentice, but I'm not certain). But I'm interested to know how the two stack up against each other otherwise. (Edit: DTH page says Houdini Indie or higher is required, so there goes DTH's one advantage out the window)
Daz 3D is part of 
Connect
DAZ Productions, Inc.
7533 S Center View Ct #4664
West Jordan, UT 84084
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
I'm also a little confused about why a pipeline for getting Daz characters from DS to Houdini requires the Daz to Maya plugin. I'm still on Houdini Indie, so I can't speak to DazToHou.
I'm pretty sure DazToHue works with Houdini Indie. Just not apprentice.
Per your confusion, I just wish there was a pipeline to Blender or Unreal Engine direct.
I meant Apprentice, I haven't sprung for Indie yet.
Hi,
I'm the creator of DazToHue so take my views with a grain of salt. I'll try to be objective as I can noting that I've never used DTL, however I am on the DTL discord server and do follow it's development closely.
I'd say the biggest difference between the DTH and DTL is ease of use vs complete control and flexibility.
DTL is certainly easier to use in that it can convert a figure with just a few clicks from what I understand, however you don't get much control during the process. DTH is a little more complicated to use however you get complete control of all aspects of the converted figure. For example, uv layouts, material slot counts and names, polygon counts, joint names etc.
DTL currently only converts G8 figures, although I believe that G9 will soon be supported. Whereas DTH can convert any Daz figure type, humanoid or otherwise.
Both can generate corrective morphs optimised for linear skinning in Unreal Engine although they use different methods for achieving this.
Both support clothing.
Both support geografts.
Speed used to be a huge difference. DTH has always been able to convert sub-d 1 figures with clothing in a few minutes (typically under 5 minutes). DTL could take several hours to convert the same figure from what I've been told however according to dev notes, a recent update to DTL has seen a 50% reduction in conversion times.
DTH allows for poly-reduction or subdivision at the time of conversion and you get complete control over where to spend your poly budget on the figure. E.g. more polygons on the face etc. DTH can generate LOD's for your figure. As far as I know DTL doesn't provide any poly-reduction functionality.
DTH can transfer animation from Daz to UE whereas DTL can't.
DTH was designed for those who largely know what's involved in converting Daz figures for use in Unreal, with optimisation in mind. It simply takes what used to be very tedious manual processes such as merging uv's and material slots, renaming joints, correcting the eye geometry to work with UE's eye shader etc, and either fully automates those processes or significantly speeds them up without ever taking control away from the user. It's more a set of utilities than a one-click solution.
DTL I would say was designed to be a simple way of getting G8 figures into Unreal with nice looking corrective morphs.
The reason DTH uses the Daz Studio side of the DazToMaya bridge plugin is because it's an already existing way of getting a nice subdivided FBX figure out of Daz Studio, along with the fig.dtu file which contains useful information such as the rest orientations for bones etc. Sure, I could probably clone the DazToMaya bridge repo and rename it to DazToHoudini but there doesn't seem much point at this stage. The FBX file that the DazToMaya plugin exports is already perfectly well suited for use in Houdini whereas the FBX files that the DazToUnreal or DazToBlender plugins export, aren't.
DTH does require Houdini Indie or higher. Unfortunately there isn't anything I can do about that as file exporting is locked in Houdini Apprentice sadly.
So, I finally understand why the DTM bridge is needed there. Many thanks for the clarification !
Would DTH work with different bridges? I use the Daz to C4D bridge.
TBH I hadn't tested FBX files from other Daz bridges as you can have multiple bridges installed at the same time so why not just use the Maya bridge plugin?
To clarify, all DTH needs is the .fbx and .dtu files from the bridge. It doesn't need the app specific side of the bridges.
Anyway, I did a quick test of .fbx and .dtu files generated by other bridges and here is what I've found:
Definitely works
Seems to work but not tested thoroughly
Definitely doesn't work
I'm the creator of DTL, so as with MrPDean's response, please take what I say with a pinch of salt. I've not used DTH but I do follow progress and am very impressed with the system's capabilities.
DTL Conversion
My background is character animation, so DTL is primarily focused on creating a character that can be dropped into Unreal and will work out of the box with existing animation systems, demos and control rigs etc. For example, a figure produced by the release version of DTL (not yet released) can be dropped into the standard 3rd person UE demo project with almost zero configuration (see here: https://youtu.be/BPQgRld3I7A?si=c3hRBkFHzLhLNyXX).
During the conversion process, DTL completely rebuilds the character's skeleton, matching the joint orientations, rotation orders and names to exactly match that of Unreal's Metahuman / MannyQuinn figures. DTL also automatically merges the figure's meshes (including geografts) and produces a set of morphs that work on the merged mesh.
Both the current beta version of DTL and the release version also produce a complete set of game engine compatible JCMs that cover the complete range of human motion. The DTL demo projects contain a pre-configured Animation Blueprint Post Process that uses UE's posedriver technology to drive the JCMs at runtime so that realistic body shape and muscle deformation are maintained even when the character is performing extreme motions like acrobatics.
Associated Software
DTL uses Maya to perform the conversion. There is an Indie version of Maya available for $300 a year and a trial version is also available. Please see here for links: https://virtuality.space/dtl/dtl-prerequistes.html#software
After conversion, a DTL converted figure is ready for further editing in Maya / Blender / Houdini etc if, for example, extra morphs are to be made or UV atlasses are to be produced.
The reason DTL uses the DazToUnreal plugin as part of the process is that the plugin contains an implementation of Pixar's subdivision algorithm for exporting fbx files with higher subdivision levels (i.e. higher polygon counts). As well as producing an fbx file, the DazToUnreal plugin also produces a .dtu file that contains useful information that DTL uses during the conversion process. DazToUnreal also automatically creates a set of materials in Unreal Engine that can be applied to the DTL converted figure. Kudos is owed to David Vodhanel for his amazing work on the DazToUnreal plugin.
It important to note that the figures DTL produces can be used with any game engine that can import clean, well formed, game ready character fbx files - despite the use of the DazToUnreal plugin during their creation process, DTL converted figures are not limited to use in Unreal Engine.
The Future of DTL
The release version of DTL (now planned for early 2024) will include the following features in addtion to 100% Unreal Engine skeleton compatibility:
~ Morph driven joint transforms that enable the morphing of one character to another in realtime, give full functionality to morphs like 'mouth open' and ensure rigid follower meshes like buttons and zips stay in place when morphs are applied at runtime.
~ G9 (and possibly G3) support
The 100% UE compatible skeleton functionality is already finished. The morph driven joint transforms and G9 support are in a working state and just need refining.
The automatic creation of a custom assets for UE's Machine Learning system for muscle deformation is also planned for a subsequent release.
The Future of Character Animation
Further down the line, I plan to make DTL capable of creating assets for neural network based locomotion animation systems. Neural network based animation systems stand to replace state trees and blendspaces with a neural network and produce far, far more compelling results than traditional animation techniques. These systems are also incredibly performant at runtime.
One of the world's leading researchers in the field (Daniel Holden) started working for Epic earlier this year. I've been following Daniel's work (and others in the field) since 2017 when he published his seminal paper on the subject.
All of the work I have done so far on DTL so far has been with the intention of being ready for when a neural network based animation system become commercially available - without doubt, they are the future of character animation: See here for the current state of the art (as produced by Sebastian Starke, another leading researcher in the field): https://youtu.be/YhH4PYEkVnY?si=9kffOsUgvdGmDx36
I feel there is a very high chance we will finally see these systems being rolled out in a user friendly, commercial form during 2024.
Good to know, many thanks!