Information-Regularized Constrained Inversion for Stable Avatar Editing from Sparse Supervision

• URL: http://arxiv.org/abs/2604.02883v1
• Date: Fri, 03 Apr 2026 08:46:54 GMT
• Title: Information-Regularized Constrained Inversion for Stable Avatar Editing from Sparse Supervision
• Abstract summary: We propose a conditioning-guided edited reconstruction framework that performs editing as a constrained inversion in a structured avatar latent space.<n>The resulting method operates on small subspace matrices and can be implemented efficiently.
• Score: 32.22231480259466
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Editing animatable human avatars typically relies on sparse supervision, often a few edited keyframes, yet naively fitting a reconstructed avatar to these edits frequently causes identity leakage and pose-dependent temporal flicker. We argue that these failures are best understood as an ill-conditioned inversion: the available edited constraints do not sufficiently determine the latent directions responsible for the intended edit. We propose a conditioning-guided edited reconstruction framework that performs editing as a constrained inversion in a structured avatar latent space, restricting updates to a low-dimensional, part-specific edit subspace to prevent unintended identity changes. Crucially, we design the editing constraints during inversion by optimizing a conditioning objective derived from a local linearization of the full decoding-and-rendering pipeline, yielding an edit-subspace information matrix whose spectrum predicts stability and drives frame reweighting / keyframe activation. The resulting method operates on small subspace matrices and can be implemented efficiently (e.g., via Hessian-vector products), and improves stability under limited edited supervision.

Related papers

• ReDesign: Recovering Editable Design Structures from Images via Agentic Decomposition [60.25038184390257]
  ReDesign is an agentic framework that grows an editable layer hierarchy by selecting and composing specialized tools across modalities.<n>To evaluate editability at scale, we introduce the Figma Edit Replay Benchmark consisting of 909 raw Figma files and 14,796 controlled edit instructions that replay edits on reconstructed outputs.
  arXiv  Detail & Related papers  (2026-07-28T10:53:33Z)
• Wavelet-Guided Semantic Signal Compensation for Inversion-Free Image Editing [12.036164124605248]
  Text-guided image editing aims to modify visual content according to a target prompt while preserving the background.<n>Recent inversion-free image editing frameworks such as FlowEdit have demonstrated strong editing capability without requiring inversion.<n>We propose an inversion-free, frequency-aware semantic compensation strategy that strengthens the effective signal in the early stage of generation.
  arXiv  Detail & Related papers  (2026-07-02T16:50:26Z)
• Conditioning Matters: Stabilizing Inversion and Attention in Diffusion Image Editing [27.02750589908665]
  Inversion-based image editing offers flexible and training-free control but still struggles with inversion accuracy and the trade-off between editing fidelity and background preservation.<n>We show that the precision of textual conditioning influences inversion stability by modulating the geometry of the diffusion velocity field.<n>We propose SimEdit, a conditioning-aware framework with two complementary components: conditioning refinement and cross-branch attention control.
  arXiv  Detail & Related papers  (2026-06-12T05:13:01Z)
• DSCA: Dynamic Subspace Concept Alignment for Lifelong VLM Editing [1.6830191160943109]
  Lifelong editing is a challenging task, prone to disrupting previously learned concepts.<n>Current methods algorithmically control edits via optimization rather than structurally separating knowledge.<n>We introduce Dynamic Subspace Concept Alignment (DSCA) which by design mitigates this limitation.<n>Our method achieves 98 percent single edit success, remains over 95 percent after 1000 sequential edits, hallucination by 3 to 5 percent, and the best backward transfer (BWT) scores on continual instruction tuning benchmarks.
  arXiv  Detail & Related papers  (2026-04-09T08:25:54Z)
• CrispEdit: Low-Curvature Projections for Scalable Non-Destructive LLM Editing [25.410131838577907]
  CrispEdit is a principled second-order editing algorithm that treats capability preservation as an explicit constraint.<n>We show that CrispEdit achieves high edit success while keeping capability degradation below 1% on average across datasets.
  arXiv  Detail & Related papers  (2026-02-17T18:58:04Z)
• Beyond Hard Writes and Rigid Preservation: Soft Recursive Least-Squares for Lifelong LLM Editing [16.768713543557443]
  Model editing updates a pre-trained with new facts or rules without re-training, while preserving unrelated behavior.<n>We propose RLSEdit, a least-squares editor for long sequential editing.<n>Experiments on multiple model families demonstrate stable scaling to 10K edits, outperforming strong baselines in both edit success and holistic stability.
  arXiv  Detail & Related papers  (2026-01-22T06:11:44Z)
• Spectral Characterization and Mitigation of Sequential Knowledge Editing Collapse [44.49646322759214]
  We show that a model's general abilities are closely associated with dominant singular directions of pretrained weight matrices.<n>We propose REVIVE, a plug-and-play framework that stabilizes sequential editing by explicitly preserving the dominant singular subspace.
  arXiv  Detail & Related papers  (2026-01-16T07:18:14Z)
• FlowDC: Flow-Based Decoupling-Decay for Complex Image Editing [52.54102743380658]
  We propose FlowDC, which decouples the complex editing into multiple sub-editing effects and superposes them in parallel during the editing process.<n>FlowDC shows superior results compared with existing methods.
  arXiv  Detail & Related papers  (2025-12-12T09:08:39Z)
• Kontinuous Kontext: Continuous Strength Control for Instruction-based Image Editing [76.44219733285898]
  Kontinuous Kontext is an instruction-driven editing model that provides a new dimension of control over edit strength.<n>A lightweight projector network maps the input scalar and the edit instruction to coefficients in the model's modulation space.<n>For training our model, we synthesize a diverse dataset of image-edit-instruction-strength quadruplets using existing generative models.
  arXiv  Detail & Related papers  (2025-10-09T17:51:03Z)
• Follow-Your-Shape: Shape-Aware Image Editing via Trajectory-Guided Region Control [52.87568958372421]
  Follow-Your-Shape is a training-free and mask-free framework that supports precise and controllable editing of object shapes.<n>We compute a Trajectory Divergence Map (TDM) by comparing token-wise velocity differences between the inversion and denoising paths.<n>Our method achieves superior editability and visual fidelity, particularly in tasks requiring large-scale shape replacement.
  arXiv  Detail & Related papers  (2025-08-11T16:10:00Z)
• Tuning-Free Image Editing with Fidelity and Editability via Unified Latent Diffusion Model [60.82962950960996]
  We introduce UnifyEdit, a tuning-free method that performs diffusion latent optimization.<n>We develop two attention-based constraints: a self-attention (SA) preservation constraint for structural fidelity, and a cross-attention (CA) alignment constraint to enhance text alignment.<n>Our approach achieves a robust balance between structure preservation and text alignment across various editing tasks, outperforming other state-of-the-art methods.
  arXiv  Detail & Related papers  (2025-04-08T01:02:50Z)
• Lost in Edits? A $λ$-Compass for AIGC Provenance [119.95562081325552]
  We propose a novel latent-space attribution method that robustly identifies and differentiates authentic outputs from manipulated ones.<n>LambdaTracer is effective across diverse iterative editing processes, whether automated through text-guided editing tools such as InstructPix2Pix or performed manually with editing software such as Adobe Photoshop.
  arXiv  Detail & Related papers  (2025-02-05T06:24:25Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.