Wrap retoplogy tool
polycounter lvl 15
http://www.russian3dscanner.com/
This tool pretty much enables you to wrap a good topology mesh onto a sculpt with some very impressive results
you can input reference points and the tools will solve one mesh to the other
Give it a look and a try, theres a demo
So are any of you using this already?
[vv]101497043[/vv]
This tool pretty much enables you to wrap a good topology mesh onto a sculpt with some very impressive results
you can input reference points and the tools will solve one mesh to the other
Give it a look and a try, theres a demo
So are any of you using this already?
[vv]101497043[/vv]
Replies
http://i.imgur.com/OILDONL.jpg
results are decent, their brush tool is very useful though.
lol i didnt even know that thing was still out there :P
*Pinned and Shared*
3dsmax has the conform brush which can be used in a similar way.
http://www.youtube.com/watch?v=Hzus80m6kJk?t=2m28s
But it doesn't have the point to point tool that they showed.
maybe for something like clothes this would help? but I dunno I am just failing to see where it properly fits into a workflow.
There are a bunch of studios that use the same topology for every character's face, so using a tool like this to deform a base head into shape for new characters would be exceptionally useful. There are certainly other ways to get a very similar result, but this tool is designed to make the process a lot faster.
Have a look at this video about the Thief face pipeline: https://www.youtube.com/watch?v=rY52p12bnIo
That's just a single example of where a tool like this would instantly save time and money for a large production. Other uses would of course include things like clothing, but you could also do things like deforming a base mesh for digital extras, creating morph targets, and adjusting the proportions of an existing model to fit new design goals. Honestly, the uses for a tool like this are almost limitless.
So a workflow could be as simple as lowpoly, Highpoly, Wrap the lowpoly to high, bake and boom you are done no need for re topology?
seems very interesting from that perspective.
This wrap topology tools is really just the first piece of a much more robust tool chain. It is essentially a mapping that allows for deformations to be accurately moved from one model to another. Even if they are wildly different in structure. It also shares a lot of foundational math with some of the more usuable example based deformation technology.
This is REALLY good stuff. The closest thing to this kind of technology that has been in a usable state in a commercial package is Houdini's blendpose CHOP.
A wrap tool like the one showed here is hugely beneficial for this kind of approach.
Pretty much. That's how one would tackle any big project with a lot of avatar options, especially if the style is anything close to realistic. It has the added benefit of unified UVs, which can be used (up to a certain extend...) for UV-dependent overlays too. Building that one solid base head is one of preproduction duties of the character art team.
What level of detail are we talking about here? Cinematic models? I feel like this sort of thing could be useful in a lot of situations but I'm curious to see at what point it would break down.
I`m testing this atm with some stylized faces where the topology of the basemesh differs a lot from the sculpt, mostly finding problems at areas where my lowpoly mesh is pinched and the sculpt is broader like a half closed eyelid vs an open one.
wow at this level of detail... just use your lowpoly as your basemesh, this snapping to surface is a insane waste of time.
Another option would be using skinwrap - have a generic base, slightly opened mouth, sightly closed eyes, and the according lowpoly. skinwrap the lowpoly onto the highpoly and morph the highpoly to the new state, voila lowpoly done.
Jacque, man, you can get rid of the snapping for free with what max delivers since years.
with the wrap tool in pace i can imagine using dynamesh or similar approaches but in a production using a basemesh already to create these vast variations of heads, there are more straight forqard approaches.
We did have a few cases where we wanted to fit a costume or armor from a male character to a female character, like Shepard -> Femshep in ME3. In those cases the poly count was in the hundred thousands range and these approaches just didn't work there any more.
I would agree that the way it was presented in that video was indeed quite time-waisting. Maybe that's just because he only showed the most simple tool to the camera.
In practice, there are tools like Topogun making the process very smooth ; and even in Max, one could use soft selection and a relax brush to manipulate large chunks of verts at a time. At the end of the day I think it is an approach that *seem* to take some time at first, because it looks like an extra step ; but in practice I find it to be very valuable.
This technology is very exciting, the topology stuff is great but this is just the tip of the iceberg. The resulting mesh that comes from fitting the topology to a template can be used for computing an affine transformation that allows for the accurate transfer of mesh deformations. This obviously means blendshapes can be mapped from a master library to any character, with any topology, of any density, with any vertex order in a matter of minutes (how long it takes to place the markers to accurately transfer the deformation).
There are less obvious applications but just as exciting, and potentially time saving, but they depend on how the matrix decomposition has been implemented in the core technology.
Also this technology actually scales very well, it has no problems with large vertex counts. I wouldn't recommend trying it with R3DS until they get their solution optimized though.
It projects from your camera onto the surface whereas the other conform brushes project based off the normal of the LP model, I've found in many situations its more precise to use extend to just put the vert where you want it with one click, rather than with a larger brush, that could cause verts to project onto a backface, or go off in a wrong direction and result in crazy edgeflow that you need to correct one by one anyways.
for example ears, if you have an ear in a different place and of different size, but you want to keep the specific edgeflow along the surface.
I've moved from Topogun to the graphite tools and find the workflow MUCH faster, mostly because of the ability to use max's modeling tools in tandem with the retopo tools. and the Extend tool is one I use quite often.
in the comments Jacque said they needed to have specific edgeloops match between the characters, for example the loop for the nasolabial fold needed to be the same on all characters.
But I digress, I havent tried this Wrap tool and looks like it could be just lighyears faster than my current workflow. deffinately will be giving this a shot.
Kind of back to square one of the thought how is this useful lol.
For some situation the conform method in Max is just fine and the method used by R3DS does not provide an advantage.
Are the tools you have used available for the public to purchase or are they all in-house tools? I would love to see more advanced versions of this type of software for sure. It's potential uses are very exciting to me!
I`m sure there are inhouse tools that are better at it with an entire team of TDs and programmers behind it but that doesnt help anyone that isn`t working at that specific company unless they`re willing to share their findings/tools at GDC or something simular.
No, the tools are unfortunately not publicly available, the research however is:
http://people.csail.mit.edu/sumner/research/deftransfer/Sumner2004DTF.pdf
Technology that's 10 years old(the previous work that deformation transfer is based on goes back 13 years!) works extremely well, saves mountains of time, and there is still no out of the box commercial solution available.
However Houdini has a CHOP node called 'blendpose' that provides a lot of the core functionality; multi-dimensional scattered interpolation and matrix decomposition. Wrapping that node into a digital asset to get the desired results is not a trivial task, but it's better than having to implement everything from scratch.
The tools I used were written by one person who saw some presentations at Siggraph that peaked his interest and imagination. Implementation isn't the Herculean effort that research and discovery is. That isn't to say it wasn't hard work, it was, but it was effort well spent.
So even if production ready tools aren't being shared, the information to make them is. Personally I find that more valuable than a black box tool that you have no idea whats actually going on inside.