cs.CVOct 1, 2026

EditHero: A Benchmark for Long-Horizon Part-Level 3D Editing and Vibe Modeling

Authors: Ruihan Yu, Yu-Ju Tsai, Muyao Niu, Runyi Li, Lian Fu, Hanqing Liu, Zheng-Hui Huang, Yonghao Yu, 

Organizations: Alaya Lab · The University of Tokyo · Institute of Science Tokyo · University of California, Merced

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

3D editing methods are usually tested on a single edit, yet an asset is built through a long sequence of revisions, each of which must implement the requested change while leaving everything else unchanged. We introduce EditHero, to our knowledge the first benchmark for long-horizon, part-level 3D editing, with natural-language instructions and target images for both geometry and texture. A deterministic assembly engine produces the exact target after every edit, and every sequence is reviewed by hand. We use EditHero to compare 2 opposite approaches to 3D editing. Non-agentic methods operate top down, regenerating the object from a learned 3D representation and inferring what to keep. In contrast, LLM/VLM agents operate bottom up, editing through code that inspects the mesh and rewrites only the parts required by instructions. The non-agentic methods often miss the requested change and disturb regions that should stay fixed. Most LLMs follow instructions more closely, and all of them preserve the unedited parts better, but each of their edits takes minutes. We will release the engine and the edit sequences to support research on reliable iterative 3D editing.

Figures & tables

Figure 1: Editing is a sequence. The figurine goes through 5 edits, and every turn has to keep what the earlier turns built. It gains sunglasses ( t1t_{1} ), loses its cap ( t2t_{2} ), turns brown-haired ( t3t_{3} ), changes its outfit ( t4t_{4} ) and finally puts on a witch hat ( t5t_{5} ). The 2 plinths show 2 ways of getting there. A traditional, non-agentic method regenerates the whole figurine at every turn, so parts nobody asked to change can drift. An agentic method rewrites only the named part through code, which keeps the rest fixed, although the new parts come out simpler. Can either meet the demands of asset production, and how can we reach vibe modeling?
Figure 2: Constructing EditHero. (a) A host object is divided into named slots, and candidate parts are drawn from a captioned library. (b) The addition shown illustrates edit selection, part retrieval, visual screening and placement checks. Each accepted edit is recorded in the chain log, and reviewers refine the drafted instructions. (c) With the part library, the log rebuilds the exact target of any turn. A method under test never sees the log. It receives the instruction and the target render of each turn. The example combines recorded turns from 4 chains on the same host; all images are renders. Colors indicate the 4 edit operations.
Figure 3: EditHero statistics. The benchmark contains 457 chains, 2755 edits, and 252 hosts. (a) Edits by chain family and operation. (b) Chain lengths, with lengths 19 and above pooled. (c) Share of each operation by turn index, with turns 10 and above pooled. (d) Host categories.
Figure 4: Along a chain. (a,b) Means on the same 135 chains with at least 7 turns, completed by all 4 non-agentic methods (103 hosts). Only defined scores enter each mean; shading shows 95% host-clustered bootstrap intervals. (a) IF relative to the method’s previous output. (b) CC 0 in regions never edited so far. (c) IoU to the target on the 147 chains of the reset diagnostic with at least 5 turns (Appendix 14 ). Solid lines: each turn starts from the model’s own output. Dashed lines: each turn starts from the ground-truth previous state.
Figure 5: Six consecutive replacements on a covered wagon (turns 1–6). The barrel becomes a wooden crate, the cover becomes a striped peaked canopy and then gray, beige and pale yellow fabric covers, and the crate becomes a science-fiction storage box. Columns are turns. In each block, the upper row is the method’s own output under self-rollout, and the lower row is its per-pixel mean absolute RGB error to the ground truth of that turn. In the ground-truth block, the lower row shows the change from the previous turn. Left: the ground truth and 3 non-agentic methods. Right: VoxHammer and 3 agents (Section 5.4 ). Same camera and lighting throughout.
BenchmarkDomainTargetTurnsRef.Own
MagicBrush [ 231 ]2D imageannotated imagesequence (few)✓✓
3DEditVerse [ 199 ]3Dpose edits / 2D lift1––
Nano3D-Edit-100k [ 216 ]3Dnative edit1––
Uni3DEdit-Bench [ 193 ]3Dsemantic-part edits1––
EDIT3D-BENCH (Steer3D) [ 126 ]3D2D lift1––
Delta3D [ 145 ]3Dpart removal1––
Table 4: EditHero compared with existing editing benchmarks. Target : how the reference state is obtained ( 2D lift : an edited image reconstructed as a 3D target; native edit : a native 3D editing method guided by an edited image; none : no target asset, scored against an edited image). Turns : 1 , one instruction per sample; goal , one target task, the agent may iterate; bundle , several edits given in one prompt; sequence , different instructions applied in order. Ref. : a reference is supplied after each instruction in a multi-turn sequence; a dash also covers single-turn or single-goal settings where this property does not apply. Own : turn kk is applied to the method’s output of turn k−1k-1 .
Figure 6: Why the metrics are defined on regions and on the model’s own state. Four turns from the evaluation, in the conditioning view. Outlines mark the edited parts before and after the turn, dashed on the model’s own states. For each turn, the left card shows how a target-similarity or ground-truth-based score can mislead, and the right card shows the corresponding IF or CC score (Eqs. ( 1 )–( 5 )). The tag on the right names the design choice the turn motivates.
Figure 7: Per-turn metrics on the 416 chains that all 4 non-agentic methods complete (228 hosts). Shaded: 95% host-clustered bootstrap intervals. The dashed line is the no-op baseline, which has IF =0=0 and CC 0=1{}^{0}=1 by definition. Its IoU compares the unchanged initial object with the current target.
Figure 8: Metrics by operation. Means over completed turns with defined scores. Retexture IF is undefined (N/A, not zero). Coverage and conditioning differ by method, so the methods are not compared on the same turns here.
Figure 9: Retexture turns from 3 chains in the conditioning view. For each method the left tile is its render and the right tile its per-pixel error to the target render (mean absolute RGB difference; color bar at the bottom right). The numbers in the top-right corner are (PSNR in dB / LPIPS) of the render against the target view. The ground truth changes only the appearance of the named part, while the errors of the 4 methods spread over the whole object.
Figure 10: Conditioning and held-out views. GT, PartFlow and VoxHammer at turn 6 of one chain, in the conditioning view and the 3 held-out views.
Figure 11: A table setting (turns 1–5): turns 1–3 and 5 change the material of the sushi tray, the soup bowl or the compartment tray, and turn 4 replaces the sushi tray and the sushi with a wooden compartment tray. Shown in the conditioning view.
Figure 12: A house (turns 1–5): 3 additions and 2 removals on the porch. Shown from a held-out view other than the conditioning view.
Figure 13: A bicycle (turns 1–5): the saddle is replaced by a wooden stool and repainted, the basket is replaced by a medical case and repainted, and the stool is replaced by a mushroom cap. Shown in the conditioning view.
Figure 14: A flying saucer (turns 1–5): the dome is replaced by a parachute canopy, the landing legs are removed, the canopy is replaced by a transparent cylinder, and a gas cylinder and a barrel are added. Shown from a held-out view other than the conditioning view.
Figure 15: A character (turns 1–5, all 4 operations): the green cap is removed, the legs are replaced by legs in black thigh-high socks, a yellow hard hat is added, the arms are retextured with red and white candy stripes, and the legs are retextured as honey-oak wood. Shown from a held-out view other than the conditioning view.
Figure 16: Add-only and remove-only chains.
Figure 17: Removal and replacement chains.
Figure 18: Replacement chains.
Figure 19: Add-and-remove chains.
Figure 20: Add-and-remove and mixed-edit chains.
Figure 21: Mixed-edit chains.
Figure 22: Chains with all four operations.
Figure 23: Retexture-only chains.
Figure 24: Retexture and geometry-edit examples.
Figure 25: Scene editing.
Figure 26: Scene and character editing.
Figure 27: Character editing.
Figure 28: Object editing.
More in this topic

3D Asset Editing

See all 18 papers

Explore similar work

CardsList
Feb 23, 2026

Vinedresser3D: Agentic Text-guided 3D Editing

Text-guided 3D editing aims to modify existing 3D assets using natural-language instructions. Current methods struggle to jointly understand complex prompts, automatically localize edits in 3D, and preserve unedited content. We introduce Vinedresser3D, an agentic framework for high-quality text-guided 3D editing that operates directly in the latent space of a native 3D generative model. Given a 3D asset and an editing prompt, Vinedresser3D uses a multimodal large language model to infer rich descriptions of the original asset, identify the edit region and edit type (addition, modification, deletion), and generate decomposed structural and appearance-level text guidance. The agent then selects an informative view and applies an image editing model to obtain visual guidance. Finally, an inversion-based rectified-flow inpainting pipeline with an interleaved sampling module performs editing in the 3D latent space, enforcing prompt alignment while maintaining 3D coherence and unedited regions. Experiments on diverse 3D edits demonstrate that Vinedresser3D outperforms prior baselines in both automatic metrics and human preference studies, while enabling precise, coherent, and mask-free 3D editing.

Yankuan Chi, Xiang Li, Zixuan Huang +1

The Hong Kong University of Science and Technology · University of Illinois Urbana-Champaign

Text-Guided 3D Scene Editing3D Asset Editing3D Shape Editing3D Diffusion Models
AskPDF
Aug 25, 2026

ViSculpt: Visual-Centric Agentic Geometry Editing

3D geometry editing is a critical yet labor-intensive part of the graphics pipeline, requiring artists to translate creative intent into precise operations in complex professional software. Large language models (LLMs) have shown promise for script-based 3D creation, but script generation is less suited to perception-driven editing of arbitrary existing meshes, where execution must remain visually grounded and untouched regions should be preserved. We present a visual-centric, training-free multi-agent system that edits existing 3D meshes directly in Blender by emulating the iterative workflow of human artists. Rather than generating scripts or regenerating geometry, our system operates through the Blender GUI: multimodal LLM agents observe the viewport, reason about the current mesh state, and execute localized edits through simulated user interactions. Experiments on a curated benchmark provide initial evidence that this agentic approach can follow natural language instructions, perform representative localized mesh edits, and preserve the overall identity of the input asset. Our results highlight a complementary regime for language-driven 3D editing: direct in-place modification of existing meshes within the native 3D editing workflow. We view this work as an exploratory step toward visual-centric agentic geometry editing in professional graphics software.

Bo Pang, Jiaqi Pan, Xiaocheng Zhang +3

Peking University, China

3D Shape EditingGUI AgentsMulti-Agent Systems
AskPDF
Apr 17, 2026

neuralCAD-Edit: An Expert Benchmark for Multimodal-Instructed 3D CAD Model Editing

We introduce neuralCAD-Edit, the first benchmark for editing 3D CAD models collected from expert CAD engineers. Instead of text conditioning as in prior works, we collect realistic CAD editing requests by capturing videos of professional designers, interacting directly with CAD models in CAD software, while talking, pointing and drawing. We recruited ten consenting designers to contribute to this contained study. We benchmark leading foundation models against human CAD experts carrying out edits, and find a large performance gap in both automatic metrics and human evaluations. Even the best foundation model (GPT 5.2) scores 53% lower (absolute) than CAD experts in human acceptance trials, demonstrating the challenge of neuralCAD-Edit. We hope neuralCAD-Edit will provide a solid foundation against which 3D CAD editing approaches and foundation models can be developed. Code/data: https://autodeskailab.github.io/neuralCAD-Edit

Toby Perrett, Matthew Bouchard, William McCarthy

Autodesk Research

3D Shape EditingParametric CAD ModelingMultimodal Model EvaluationMultimodal Instruction Tuning3D Asset Editing
AskPDF