Residual Decoder Adapter: ID-Preserving Tokenizer Adaption for Autoregressive Text Rendering
2026-06-01 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors found that visual autoregressive (AR) models have trouble rendering clear and sharp text because the visual tokenizer they use can't capture fine details well. Instead of retraining the whole system, they created the Residual Decoder Adapter (RDA), a tool that improves the tokenizer's output by learning small differences between the generated images and the real images. This method keeps the original token system intact, so it works with existing AR models. Their approach significantly improved text clarity in generated images, shown by big jumps in OCR accuracy on several benchmarks.
Visual Autoregressive ModelsVisual TokenizerImage GenerationText RenderingResidual Decoder AdapterCodebookOCR AccuracyPost-hoc Model ImprovementPixel-space ResidualsTextAtlas Benchmark
Authors
Dongxing Mao, Jinpeng Wang, Jiahao Tang, Kevin Qinghong Lin, Linjie Li, Zhengyuan Yang, Lijuan Wang, Min Li, Jingru Tan
Abstract
Visual Autoregressive (AR) models generate images by predicting discrete tokens that are decoded by a visual tokenizer. Despite demonstrating strong overall image generation ability, they still underperform on text rendering with blur strokes and disrupt letter shapes. In this work, we trace this limitation to the visual tokenizer, which struggles to reconstruct fine-grained detail. Improving the tokenizer is straightforward but expensive, as it necessitates retraining both the tokenizer and the AR model. Can we improve text rendering performance of AR models without retraining the existing tokenizer and AR model? To achieve this, we propose the Residual Decoder Adapter(RDA) that upgrades an existing tokenizer post-hoc without changing its token space. Specifically, it refines the decoder output of the visual tokenizer by introducing two novel components: (i) a paired codebook that shares the token distribution with the original one; (ii) a parallel branch to learn the tiny differences (residual) between the reconstructed image and the ground-truth images in the pixel space. This residual design allows us to enhance the tokenizer non-invasively while preserving compatibility with prior AR models. RDA substantially improves text rendering significantly by a large margin. For instance, we boost finetuned Janus-Pro OCR accuracy rises from 24.52% to 58.26% (TextVisionBlend), from 12.75% to 36.81% (StyledTextSynth) on competitive TextAtlas benchmark. The code is available at https://github.com/CSU-JPG/RDA