Saved in:
| Main Authors: | , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2603.17944 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914409167716352 |
|---|---|
| author | Zhang, Fei Zhou, Zijian Tang, Bohao He, Sen Li, Hang Wang, Zhe Sanyal, Soubhik Liu, Pengfei Atliha, Viktar Xiang, Tao Xu, Frost Gunel, Semih |
| author_facet | Zhang, Fei Zhou, Zijian Tang, Bohao He, Sen Li, Hang Wang, Zhe Sanyal, Soubhik Liu, Pengfei Atliha, Viktar Xiang, Tao Xu, Frost Gunel, Semih |
| contents | We introduce the first method, to the best of our knowledge, for adapting image-to-video models to layer-aware text (glyph) animation, a capability critical for practical dynamic visual design. Existing approaches predominantly handle the transparency-encoding (alpha channel) as an extra latent dimension appended to the RGB space, necessitating the reconstruction of the underlying RGB-centric variational autoencoder (VAE). However, given the scarcity of high-quality transparent glyph data, retraining the VAE is computationally expensive and may erode the robust semantic priors learned from massive RGB corpora, potentially leading to latent pattern mixing. To mitigate these limitations, we propose TransText, a framework based on a novel Alpha-as-RGB paradigm to jointly model appearance and transparency without modifying the pre-trained generative manifold. TransText embeds the alpha channel as an RGB-compatible visual signal through latent spatial concatenation, explicitly ensuring strict cross-modal (RGB-and-Alpha) consistency while preventing feature entanglement. Our experiments demonstrate that TransText significantly outperforms baselines, generating coherent, high-fidelity transparent animations with diverse, fine-grained effects. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_17944 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | TransText: Alpha-as-RGB Representation for Transparent Text Animation Zhang, Fei Zhou, Zijian Tang, Bohao He, Sen Li, Hang Wang, Zhe Sanyal, Soubhik Liu, Pengfei Atliha, Viktar Xiang, Tao Xu, Frost Gunel, Semih Computer Vision and Pattern Recognition We introduce the first method, to the best of our knowledge, for adapting image-to-video models to layer-aware text (glyph) animation, a capability critical for practical dynamic visual design. Existing approaches predominantly handle the transparency-encoding (alpha channel) as an extra latent dimension appended to the RGB space, necessitating the reconstruction of the underlying RGB-centric variational autoencoder (VAE). However, given the scarcity of high-quality transparent glyph data, retraining the VAE is computationally expensive and may erode the robust semantic priors learned from massive RGB corpora, potentially leading to latent pattern mixing. To mitigate these limitations, we propose TransText, a framework based on a novel Alpha-as-RGB paradigm to jointly model appearance and transparency without modifying the pre-trained generative manifold. TransText embeds the alpha channel as an RGB-compatible visual signal through latent spatial concatenation, explicitly ensuring strict cross-modal (RGB-and-Alpha) consistency while preventing feature entanglement. Our experiments demonstrate that TransText significantly outperforms baselines, generating coherent, high-fidelity transparent animations with diverse, fine-grained effects. |
| title | TransText: Alpha-as-RGB Representation for Transparent Text Animation |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2603.17944 |