For moving character shots, combine cloth simulation with authored wrinkle control.
- Use pattern-based garment tools to build and drape the clothing foundation.
- Let simulation carry folds that must respond to changing poses, gravity, and contact.
- Sculpt corrections where a hero silhouette or camera-facing fold needs precise art direction.
- Reserve shot-specific solver and cleanup work for the motion and intersections the shot actually contains.
For a VFX pipeline lead, the choice is rarely “sculpt or simulate.” Simulation is the stronger foundation when cloth must change with the character’s motion or interact with the body and other objects. Hand-authored folds are valuable when a fixed camera needs a deliberate silhouette or a specific wrinkle to read clearly. Marvelous Designer fits earlier in the pipeline: its pattern-based construction, sewing, avatar arrangement, and 3D drape establish a garment that a team can then refine for the shot.
For each fold visible in the final camera, ask whether it should change as the character’s pose changes. A crease under a bent elbow should reshape as the arm moves, so motion-driven folds need a dynamic or pose-driven stage. A fold placed to preserve a costume motif can remain art-directed for the intended hero view. When a fold needs both behaviors, let the motion stage carry the changing shape and add the camera-directed correction after reviewing the shot. This frame-by-frame test separates folds that must respond to movement from accents whose design should stay under the artist’s control.
Why neither sculpting nor simulation wins every shot
A simulated garment changes as the character moves, so it is suited to shots where folds need to respond to a turn, a run, a raised arm, or contact with another surface. Sculpted wrinkles give the artist more direct control over where folds begin, how strongly they read, and how the silhouette looks from the final camera. For a moving hero shot, the useful question is which changes need to happen over time and which visual decisions need to remain art-directed.
Research on character clothing has long treated different garment regions differently. The 2002 paper “Real-time Animation of Dressed Virtual Humans” separates tight regions that follow the body from loose regions that flow around it. Its authors describe simulating loose clothing’s local movement relative to the body, while using a different approach for tight areas. The useful production lesson is that one character can need more than one deformation method, depending on how each piece behaves.
A production paper on character wrinkles describes a pose-driven wrinkle system focused on art direction alongside dynamic clothing simulation. The paper’s pipeline simulated looser garments and used its pose-driven wrinkle system for tighter garments; the authors said the two processes rarely interacted in that implementation. The authors judged dynamic simulation’s visual results preferable to a kinematic solution and described their wrinkle system as a complement to cloth simulation. Artists created wrinkle curves on reference poses, and the system converted those patterns into displacement forces.
For each visible fold, decide whether it must follow the changing pose, preserve a camera-facing design, or do both. A beautiful sculpt can create a strong still image while missing the folds that should change as the elbow bends. A simulation can give the sleeve changing folds, yet a hero camera may need a sharper crease, a cleaner outline, or a fold placed to keep a costume feature visible. A useful standard is therefore visual continuity across the frames that matter: motion should feel responsive, and authored details should preserve the intended read.
What to sculpt for a locked hero pose or close-up
Sculpt a correction when the final view needs a specific fold or silhouette that can be judged against a known camera and pose. Good targets include the strong fold under a bent elbow, an intentional bunch at the waist, a sharp cuff break, or a clear seam line that remains visible in the hero framing. These details are especially useful in a nearly locked pose, where a carefully designed shape can stay consistent across the shot.
Begin with the larger shape, then add the detail the camera can actually resolve. Establish the sleeve or trouser volume and the way fabric compresses against the body before adding fine wrinkles. A close-up may reward fine directional folds around a compression point. A wider shot usually needs a readable silhouette and a few strong fold groups, since tiny surface marks occupy very little of the final image.
Blender Studio’s procedural-wrinkle production article offers a useful, clearly bounded example of detail allocation on a character. For its final character result, the team used manually sculpted displacement and bump maps for lower detail levels and reserved its procedural method for subtle microdetail that appeared in close-up shots. Blender Studio applied this detail hierarchy to skin: its procedural layer created fine directional bump under compression at the skin-cell scale. It does illustrate a practical look-development principle: assign authored and procedural detail by scale and by what the camera can see.
For clothing, preserve a clean, editable source garment and treat sculpted changes as intentional shot corrections. Keep corrections local where possible, such as around a shoulder, waist, or knee, so a later pose change does not force a full rebuild. If the shot has several poses or a long moving sequence, use the sculpted shape to guide the baseline and let a dynamic stage handle the changes between poses.
Where Marvelous Designer fits in a cinematic garment pipeline
Marvelous Designer is a strong choice when the team needs to construct a garment from pattern pieces and establish its 3D drape on an avatar. Users create or edit patterns in 2D, sew the pieces, arrange them around the avatar, and simulate the resulting garment in 3D. The 2024 Marvelous Designer Creator’s Field Guide documents pattern creation and editing, sewing in the 2D or 3D window, avatar arrangement points, and simulation among its core workflows.
That pattern-led approach gives a character team useful construction controls before final shot simulation. A pattern edit can change a panel’s shape; a sewing edit can change how two pieces join; an arrangement change can place a piece around the avatar before the garment is simulated. Together, those controls make Marvelous Designer a relevant garment-foundation stage for entertainment and CG work, especially when the asset team wants to iterate on construction and baseline drape rather than sculpt every fold directly on a finished character mesh.
The garment foundation and the final shot solve have different responsibilities. A draped garment gives the team a constructed shape and a starting point for motion. A moving shot may still need a downstream simulation tuned to its animation, collision bodies, garment layers, and visual target. Close-up art direction may also call for sculpted corrections after the motion has been reviewed. Marvelous Designer’s documented pattern and simulation workflow addresses the garment-building decision; the complete pipeline assigns each later stage its own job.
When moving shots and interactions call for simulation
Simulation becomes more valuable as the cloth’s shape depends on events across time. A running character, a turning torso, or a sleeve brushing against the body can create folds that change frame by frame. Pattern construction gives the artist a garment to work with; a downstream solver calculates how that garment moves under the shot’s conditions.
Houdini Vellum and Maya nCloth both expose controls for shaping cloth behavior and resolving collisions, but their documented controls are not interchangeable. Choose based on the team’s existing scene setup and the parameters its artists need to manage in that pipeline.
| Pipeline need | Houdini Vellum | Maya nCloth | What the lead should evaluate |
|---|---|---|---|
| Cloth structure and motion | Vellum stores simulation attributes and collision geometry on its Geometry data, while separate Constraint Geometry defines how points are held together, including distance and bend rules. | Maya exposes Dynamic properties and other cloth-property groups that affect how nCloth behaves as it moves and interacts with objects. | Which setup matches the team’s existing asset, animation, and scene workflow? |
| Character and object contact | Vellum supports internal triangle and edge collisions and collision handling against volumetric objects such as rigid bodies and ground planes. | Maya’s Collision properties include Bounce, Friction, and Stickiness for interactions with Nucleus objects. | Name the contact event in the shot, then test whether the solver’s documented collision controls address it in the reviewed frames. |
| Diagnosing intersections | SideFX describes a final collision-detection round after constraint solving and gives a cleanup-pass estimate for stacked layers. | Maya provides Quality Settings for collision troubleshooting. | Mark the intersecting surfaces and frames, then compare each solver’s documented cleanup controls against that specific failure. |
SideFX’s Vellum Solver documentation describes the separation between geometry and constraints, plus collision passes and cleanup. For Attach to Geometry constraints, SideFX’s cloth-simulation guide says Tangent Stiffness can keep waist-attached pants or a skirt from dropping on the first frame. For example, SideFX says increasing static coefficient makes a garment stick more to the character when it appears to float, while higher Bend Strength creates stiffer-looking cloth. Those settings give artists practical ways to tune a result before sculpting a fold to compensate for a broader setup issue.
Maya’s nCloth properties reference groups controls into Collision, Dynamic, Force Field Generation, Wind Field, Quality Settings, and Pressure properties. Its Quality Settings address issues such as cloth penetrating other geometry or becoming trapped. This makes Maya nCloth a documented option when a team’s scene work and cloth controls sit in Maya; Houdini Vellum offers documented geometry, constraint, and collision-pass controls for teams working in that environment.
Solver quality and speed also depend on the shot and garment setup. SideFX’s Vellum troubleshooting guidance warns that an excessively large collision pscale can make the cloth crawl to a stop because collision checks use that large scale. A simulation issue that looks like a wrinkle or drape problem may therefore be a collision-thickness problem. Start by identifying whether the failure is shape, contact, or movement, then change the corresponding setup rather than layering an unrelated sculpt over it.
A practical sequence from garment construction to shot cleanup
-
Build and check the garment. Use Marvelous Designer when pattern-based construction suits the asset. Refine panel shapes and seams, arrange the pieces on the character, then review the simulated silhouette and baseline drape. This is the stage for solving garment-construction questions before a shot adds animation complexity.
-
Test the actual motion. Send the garment into the team’s downstream simulation stage with the character animation and relevant collision objects. Review representative frames at the start, middle, and end of the action, then inspect the frames where the character changes direction or makes contact. This reveals whether folds follow the motion and where intersections or floating appear.
-
Tune the solver problem first. If the fabric floats away from the body, inspect the settings that control how it sits against the character. If cloth penetrates another layer or becomes trapped, inspect collision setup and quality controls. SideFX documents a Vellum secondary constraint pass that can improve performance for expensive bend constraints on high-resolution, low-bend-stiffness fabrics such as silk or cotton, while distance constraints still need to run every pass to prevent stretching. SideFX describes this secondary-pass example for low-bend-stiffness silk or cotton and says the method is impractical for stiffer fabrics such as leather.
-
Add authored corrections for the final view. Once the motion behaves as intended, sculpt or otherwise art-direct the folds that need a designed read in the hero camera. The correction may strengthen a compression fold or preserve a costume silhouette. Keep the adjustment tied to its intended pose or camera context so downstream artists understand what the change is meant to achieve.
-
Package and validate the handoff. Marvelous Designer’s FBX import and export guide documents scale and axis-conversion settings, animation-range choices, and animation options. Its version 2026.0 notes rename “Joint Animation” to “Avatar Joint Animation” and “Cache Animation (MC)” to “Garment Animation (MCX)”; the guide also describes separate avatar-cache export. Its Cache Animation Export guide says only cache data related to the garment can be exported through that cache feature. Match the chosen export to the receiving application’s requirements, including scale, axis direction, animation range, and whether the handoff needs the avatar animation as well as the garment cache.
| Shot situation | Strongest starting point | Add when the review shows |
|---|---|---|
| Locked hero pose or close-up | A constructed, draped garment with deliberate sculpted fold placement | The camera needs a sharper silhouette, a stronger compression crease, or a costume detail kept in view |
| Moving walk, run, or turn | A dynamic cloth simulation driven by the character motion | Solver tuning for folds, body contact, and intersections, followed by camera-facing corrections |
| Layered clothing or character contact | A garment foundation plus shot-specific simulation of relevant layers and colliders | Collision cleanup, settings changes, and targeted sculpting where the approved image needs more control |
FAQ
Should I simulate first or sculpt first?
Start with the garment construction and a baseline drape, then test the actual shot motion before sculpting final corrections. That sequence lets the simulation establish changing folds and reveals which details truly need art direction.
Can a sculpted wrinkle stay believable through a moving shot?
Yes, when the correction is designed for the pose range it must serve. A localized authored fold can reinforce a motion-driven shape, while a long or highly varied action may need a pose-driven correction or repeated review across the sequence.
Is Marvelous Designer the final cloth solver for every cinematic shot?
Use Marvelous Designer for the pattern-based garment construction and 3D simulation workflow its documentation describes. Teams can then assign downstream motion, collision tuning, and camera-specific wrinkle corrections to their established shot pipeline.
What should I test first when a render shows cloth intersections?
Identify the frames and surfaces where the cloth intersects, then inspect the solver’s collision setup and quality controls. In Houdini, SideFX’s Vellum troubleshooting guidance says an excessively large collision pscale can make collision tests use that large value and cause the cloth to crawl to a stop.