Hi all, I am trying to find an animation workflow for my needs using Marvelous. For that I use the function " import obj to garment" (without tracing the pattern). It seems like when you import a 3d mesh as cloth, the simulation is not totally performing as with drawn pattern. To illustrate that, I have made a simple pants in Marvelous with the pattern method. No fancy details, just a default pants. In neutral pose, I export my simple pant in obj (triangle mesh, thin and welded). I import back that mesh with "obj to garment" making sure to delete the other before. Set the same physical properties. You can see that it somewhat performs, But the folds are less realistic and are not horizontal like on a real pants and the one coming from marvelous drawn pattern. Would you know what is causing this issue ? Does it have a workaround to put thé dame behavior like if it was a drawn pattern ?
What's happening is that you are importing a welded mesh model that now has no warp/weft grainline that is associated to singular pattern pieces. In MD fabric weave can be set to the direction on the fabric using the gizmo in the 2D pattern editor. Therefore the fabric for the entire welded item is on the bias and likely the creasing will be less realistic as the algorithm is trying to best fit a warp/weft alignment across an entire 3D model, which is never a good thing for trousers as the crotch area is complex. For a simple teeshirt this may be okay but for realistic creasing and draping best to keep the grainline of the fabrics weave direction exactly the same as a real garment.
You should also check your new models mesh particle distance, as the default is 20mm you should set this to perhaps to a lower value (like 5mm) and see if that helps, along with making sure the default fabric is not set to a fabric that is simply simulating to a different fabric as per your original MD 2D pattern garment assembly. So maybe also set this fabric to the same as your original simulated model.
Thank you Angel for your super explanations. Now I understand better why it does that.
The solution would be to convert my garment to 2d pattern when I import the mesh, that way Marvelous would know the weave direction.
It seems that for the moment, when using this function, the seams are not fully covering the edge of each UV/pattern border and needs tweaking. I don't know if it will be enhanced in the next versions but that would be cool.
Or even better : when importing a mesh, Marvelous could "look" at the UV and write a warp/weft map or something internally without having to seam in 2d view.
For now I will change my workflow and stay in Marvelous entirely from a to z.
Tobe honest it's easier to stay in MD for the garment and create it there as it's a fast process, you can speed that up if you create standard garment start points and you can automatically grade that standard item to almost any character size using their other tools. So a digital wardrobe can be re-purposed. Then you can decide to work with retopography of garment mesh also in MD that weld the UV vertex points to a base mesh garment spun out from your high quality simulation > so that MD flexibility really gives you quite a lot of productivity power within MD that you would not get outside in other apps.
I have 8+ years on MD, and as it has evolved the power for it's use really comes from leveraging the workflow from a core library of custom assets you create over time, as everything can be reused between characters. So if you look at it like a growing wardrobe asset that you build over time, you get the benefit of speed and process flow reduction the more you invest in pattern to garment construction assemblies. And you can keep that pretty flexible so design changes can be made fast. If you externally sculpt then you end up with a massive workflow that is non-destructive that you need to repeat for each element and each garment. So although it's not immediately apparent to the nivice or newby user the real power and cost reduction comes from ensuring your core workflow is consistent and repeatable into your next pipeline tool but your garment asset is flexible. When you simply do one garment at a time in isolation that benefit to overall VFX workflow is not so apparent. So it helps to open out your thinking on how you can leverage aspects of MD to lower cost longer term in digital garment builds.
Hi all,
I am trying to find an animation workflow for my needs using Marvelous. For that I use the function " import obj to garment" (without tracing the pattern).
It seems like when you import a 3d mesh as cloth, the simulation is not totally performing as with drawn pattern.
To illustrate that, I have made a simple pants in Marvelous with the pattern method. No fancy details, just a default pants.
In neutral pose, I export my simple pant in obj (triangle mesh, thin and welded). I import back that mesh with "obj to garment" making sure to delete the other before. Set the same physical properties. You can see that it somewhat performs,
But the folds are less realistic and are not horizontal like on a real pants and the one coming from marvelous drawn pattern.
Would you know what is causing this issue ? Does it have a workaround to put thé dame behavior like if it was a drawn pattern ?
https://ibb.co/ypMyzbx = pants made in marvelous
https://ibb.co/x6Vxmzy = Mesh imported back in obj to garment
What's happening is that you are importing a welded mesh model that now has no warp/weft grainline that is associated to singular pattern pieces. In MD fabric weave can be set to the direction on the fabric using the gizmo in the 2D pattern editor. Therefore the fabric for the entire welded item is on the bias and likely the creasing will be less realistic as the algorithm is trying to best fit a warp/weft alignment across an entire 3D model, which is never a good thing for trousers as the crotch area is complex. For a simple teeshirt this may be okay but for realistic creasing and draping best to keep the grainline of the fabrics weave direction exactly the same as a real garment.
You should also check your new models mesh particle distance, as the default is 20mm you should set this to perhaps to a lower value (like 5mm) and see if that helps, along with making sure the default fabric is not set to a fabric that is simply simulating to a different fabric as per your original MD 2D pattern garment assembly. So maybe also set this fabric to the same as your original simulated model.
Hope that helps.
Thank you Angel for your super explanations. Now I understand better why it does that.
The solution would be to convert my garment to 2d pattern when I import the mesh, that way Marvelous would know the weave direction.
It seems that for the moment, when using this function, the seams are not fully covering the edge of each UV/pattern border and needs tweaking. I don't know if it will be enhanced in the next versions but that would be cool.
Or even better : when importing a mesh, Marvelous could "look" at the UV and write a warp/weft map or something internally without having to seam in 2d view.
For now I will change my workflow and stay in Marvelous entirely from a to z.
Tobe honest it's easier to stay in MD for the garment and create it there as it's a fast process, you can speed that up if you create standard garment start points and you can automatically grade that standard item to almost any character size using their other tools. So a digital wardrobe can be re-purposed. Then you can decide to work with retopography of garment mesh also in MD that weld the UV vertex points to a base mesh garment spun out from your high quality simulation > so that MD flexibility really gives you quite a lot of productivity power within MD that you would not get outside in other apps.
I have 8+ years on MD, and as it has evolved the power for it's use really comes from leveraging the workflow from a core library of custom assets you create over time, as everything can be reused between characters. So if you look at it like a growing wardrobe asset that you build over time, you get the benefit of speed and process flow reduction the more you invest in pattern to garment construction assemblies. And you can keep that pretty flexible so design changes can be made fast. If you externally sculpt then you end up with a massive workflow that is non-destructive that you need to repeat for each element and each garment. So although it's not immediately apparent to the nivice or newby user the real power and cost reduction comes from ensuring your core workflow is consistent and repeatable into your next pipeline tool but your garment asset is flexible. When you simply do one garment at a time in isolation that benefit to overall VFX workflow is not so apparent. So it helps to open out your thinking on how you can leverage aspects of MD to lower cost longer term in digital garment builds.
请先登录再写评论。