ToonAging: Face Re-Aging upon Artistic Portrait Style Transfer

TL;DR

ToonAging tackles the challenge of editing age in non-photorealistic portraits while applying diverse artistic styles in a single generative pass. It achieves this by fusing age-related latent codes from a re-aging network (SAM) with an exemplar-based style latent from DualStyleGAN in a controlled, layer-wise manner using a weight vector $w$, operating in the $\mathcal{W}^+$ space for aging and preserving the extrinsic style path. The method avoids domain-specific training and relies on an exemplar-based scheme to transfer styles across arbitrary NPR domains, improving both aging realism and stylistic fidelity. Experimental results—qualitative visuals and a user study across multiple NPR domains—demonstrate superior naturalness, attribute preservation, and user preference, highlighting ToonAging's practical impact for entertainment media and NPR-based facial edits.

Abstract

Face re-aging is a prominent field in computer vision and graphics, with significant applications in photorealistic domains such as movies, advertising, and live streaming. Recently, the need to apply face re-aging to non-photorealistic images, like comics, illustrations, and animations, has emerged as an extension in various entertainment sectors. However, the lack of a network that can seamlessly edit the apparent age in NPR images has limited these tasks to a naive, sequential approach. This often results in unpleasant artifacts and a loss of facial attributes due to domain discrepancies. In this paper, we introduce a novel one-stage method for face re-aging combined with portrait style transfer, executed in a single generative step. We leverage existing face re-aging and style transfer networks, both trained within the same PR domain. Our method uniquely fuses distinct latent vectors, each responsible for managing aging-related attributes and NPR appearance. By adopting an exemplar-based approach, our method offers greater flexibility compared to domain-level fine-tuning approaches, which typically require separate training or fine-tuning for each domain. This effectively addresses the limitation of requiring paired datasets for re-aging and domain-level, data-driven approaches for stylization. Our experiments show that our model can effortlessly generate re-aged images while simultaneously transferring the style of examples, maintaining both natural appearance and controllability.

Figures (15)

• Figure 1: We propose ToonAging, which can perform face re-aging and portrait style transfer in a single generation step. Since we adopted an exemplar-based approach, portraits can be transferred to various domains, enabling plausible re-aging progression simultaneously.
• Figure 2: Comparison of our method with naive sequential approaches. (a) ToonAging perceptually reflects the apparent age while simultaneously effectively transferring the style of $S$. (b) The naive RS (re-aging then style transfer) often loses age-related attributes, resulting in the image lacking a discernible target age and missing the facial attributes of the input (e.g., expression; here, the input face has a smiling expression, but the generated image does not.) (c) Conversely, SR (style transfer then re-aging) maintains a realistic style but fails to retain the input $S$'s style. Yellow boxes denote the intermediate results of the first stage in 2-stage approaches.
• Figure 3: Our ToonAging architecture. (Left) Schematics, (Right) Data-level architecture. Under ToonAging, age-related attributes and age code are obtained via GAN inversion. Latent vector for reconstruction is embedded by $E_{\text{inv, w+}}$, and the age-related latent vector is embedded by $E_{\text{age}}$ as residuals to be added to the reconstruction vector. The example style $S$ is obtained by $E_{\text{inv, z+}}$ in $\mathcal{Z}+$ space, then converted into $\mathcal{W}+$ space with a learned MLP layer. Finally, the two latent vectors are used in the generator of DualStyleGAN.
• Figure 4: Latent vector behavior according to partial convolution in the generator for face re-aging. SAM was employed for latent analysis.
• Figure 5: Latent behavior for exemplar-based style transfer. DualStyleGAN was used for latent analysis.