SketchBloom: Training-Free Vector Sketch Outpainting with Stroke-Level Style Control

Quang-Thinh Nguyen; Duy-Khang Do; Tam V. Nguyen; Minh-Triet Tran; Trung-Nghia Le

cs.CVOct 30, 2026

SketchBloom: Training-Free Vector Sketch Outpainting with Stroke-Level Style Control

Authors: Quang-Thinh Nguyen, Duy-Khang Do, Tam V. Nguyen, Minh-Triet Tran, Trung-Nghia Le

Organizations: University of Science, Ho Chi Minh City, Vietnam · Vietnam National University, Ho Chi Minh City, Vietnam · University of Dayton, Ohio, United States

Abstract

Outpainting sparse sketches while preserving contour coherence and stroke style remains challenging. We present SketchBloom, a training-free framework that combines pretrained raster outpainting and style-transfer guidance with per-instance vector optimization. Given a sketch, a scene prompt, and an optional reference style, it produces an editable Scalable Vector Graphic (SVG) composed of Bezier strokes. Adaptive latent blending balances content contours and reference texture without model retraining. A reference-derived stroke vocabulary initializes contour- and DINO-attention-guided strokes, which are optimized separately for structure and texture before contour-emphasizing multiply fusion. As in prior optimization-based vector sketching, no task-specific network is trained; our contribution is the synthesis design specialized to sketch outpainting. On LineArtOut-1K, comprising 1,024 input and style combinations, SketchBloom improves perceptual, semantic, and structural fidelity over the evaluated vector baselines when all methods vectorize the same outpainted raster. This protocol evaluates vector synthesis rather than end-to-end outpainting or reference-style superiority. Runtime is approximately six minutes per pair on a single NVIDIA H200 GPU. Project page: https://khangcntn123.github.io/SketchBloom/

Figures & tables

Figure 2: Raster guidance pipeline. ControlNet expands $I_{\text{sketch}}$ into $I_{\text{cnt}}$ , from which $I_{\text{edge}}$ is extracted. DDIM inversion, adaptive latent blending, and style-sourced attention features combine content structure with $I_{\text{ref}}$ , producing the stylized target $I_{\text{sty}}$ .

Figure 3: Vector-native synthesis. A reference-derived stroke vocabulary initializes two passes: structure optimizes geometry against $I_{\text{edge}}$ , whereas texture optimizes stroke attributes against $I_{\text{sty}}$ . Multiply blending places $\mathcal{V}_{\text{struct}}$ over $\mathcal{V}_{\text{tex}}$ to produce the final SVG $\mathcal{V}^{*}$ .

Figure 4: Qualitative comparison under the same raster input. SketchBloom (bottom row) better retains object boundaries, perspective, and scene detail.

Method	Aesthetic [ 22 ] $\uparrow$	DINO [ 5 ] $\uparrow$	DreamSim [ 9 ] $\downarrow$	Edge F1 $\uparrow$
CLIPasso [ 25 ]	4.1690	0.2064	0.7256	0.5262
CLIPascene [ 24 ]	4.3904	0.2022	0.7024	0.5797
SwiftSketch [ 2 ]	3.9784	0.1566	0.8305	0.3139
ControlSketch [ 2 ]	4.1493	0.1470	0.7960	0.3299
DiffSketcher [ 27 ] (adapted)	4.8020	0.3760	0.5351	0.8325
SketchBloom (Ours)	5.2186	0.5053	0.4983	0.8866

Table 1: Quantitative comparison on 64 shared outpainted content images. All methods receive the same

I_{\text{cnt}}

. DINO and DreamSim use

I_{\text{cnt}}

as reference, Edge F1 uses

I_{\text{edge}}

, and Aesthetic is no-reference. Results are computed per content image. Best values are in bold .

Init.	MAE $\downarrow$	SSIM [ 26 ] $\uparrow$	Chamfer [ 4 ] $\downarrow$	Edge F1 $\uparrow$
Random	0.4384	0.1120	0.0006	0.7731
Hybrid	0.2041	0.8068	0.0003	0.8841

Table 2: Structure-only initialization ablation against

I_{\text{edge}}

. Best values are in bold .

Variant	Gram [ 10 ] $\downarrow$	DINO [ 5 ] $\uparrow$	DreamSim [ 9 ] $\downarrow$	Edge F1 $\uparrow$	Aesthetic [ 22 ] $\uparrow$
Texture only	$\mathbf{1.82\times 10^{-4}}$	0.7328	0.1744	0.8067	5.0715
Structure only	$2.76\times 10^{-4}$	0.5800	0.2918	0.8841	4.8718
Fusion	$2.19\times 10^{-4}$	0.7372	0.1712	0.8629	5.3507

Table 3: Fusion ablation. Gram uses

I_{\text{ref}}

, DINO and DreamSim use

I_{\text{sty}}

, and Edge F1 uses

I_{\text{edge}}

. Best values are in bold .

Panel of Figure 5 — Figure 5: Qualitative ablations. Top: hybrid tiling converges earlier than random initialization. Bottom left: patch loss preserves local detail without the gray halos of $L_{2}$ . Bottom right: multiply fusion combines geometric structure with tonal texture.

Vector Graphics Generation

See all 28 papers

SketchForge: Training-Free Scene-to-Vector Sketching via Decoupled Structure and Texture Optimization
Oct 30, 2026Quang-Thinh Nguyen, Duy-Khang Do, Hien Trinh +3Image Vectorization Vector Graphics Generation
University of Science, Ho Chi Minh, Vietnam · Vietnam National University, Ho Chi Minh, Vietnam · University of Dayton, Dayton, Ohio, US
SketchFlow: Zero-Shot Vector Sketch Generation via GMM Prior Flow in CLIP Latent Space
Aug 21, 2026Jin Zhou, Hongliang Yang, Pengfei Xu +1Flow-Based Generative Modeling Image Generation
Guangdong Provincial Key Laboratory of Visual Media and Multidimensional Intelligence, CSSE, Shenzhen University, Shenzhen, China
SketchSense: Learning to Interpret Imperfect Sketch Guidance for Image Inpainting
Aug 13, 2026Zian YangDiffusion Models for Image Restoration Diffusion Model Guidance
Fudan University

Oct 30, 2026J/K move · Enter open · S save

SketchForge: Training-Free Scene-to-Vector Sketching via Decoupled Structure and Texture Optimization

Quang-Thinh Nguyen, Duy-Khang Do, Hien Trinh, Tam V. Nguyen+2

University of Science, Ho Chi Minh, Vietnam · Vietnam National University, Ho Chi Minh, Vietnam · University of Dayton, Dayton, Ohio, US

Image Vectorization Vector Graphics Generation

Open paper AskPDF arXiv

Converting a raster image into a resolution-independent, stroke-editable vector sketch while faithfully preserving both structural contours and tonal texture remains challenging: raster-space rendering pipelines are difficult to edit at the stroke level, while generation-from-scratch vectorization methods struggle to remain faithful to an arbitrary input image. We present SketchForge, a training-free framework that converts an input raster image into a resolution-independent SVG sketch through per-instance test-time optimization of Bezier curves, requiring no additional training or fine-tuning. Strokes are seeded via a contour and attention-guided initialization, then refined by a vector-native differentiable rendering pipeline that decouples structural contour strokes from free-form texture strokes under dedicated objectives, composited via multiply-blend layering. Extensive experiments show that SketchForge consistently outperforms existing vector-based baselines in structural fidelity and generation quality. Project page: https://khangcntn123.github.io/SketchForge

Abstract

Vector Graphics Generation

Explore similar work

SketchForge: Training-Free Scene-to-Vector Sketching via Decoupled Structure and Texture Optimization

SketchFlow: Zero-Shot Vector Sketch Generation via GMM Prior Flow in CLIP Latent Space

SketchSense: Learning to Interpret Imperfect Sketch Guidance for Image Inpainting