We quantify the upper bound on the size of the implicit neural representation (INR) model from a digital perspective. The upper bound of the model size increases exponentially as the required bit-precision increases. To this end, we present a bit-plane decomposition method that makes INR predict bit-planes, producing the same effect as reducing the upper bound of the model size. We validate our hypothesis that reducing the upper bound leads to faster convergence with constant model size. Our method achieves lossless representation in 2D image and audio fitting, even for high bit-depth signals, such as 16-bit, which was previously unachievable. We pioneered the presence of bit bias, which INR prioritizes as the most significant bit (MSB). We expand the application of the INR task to bit depth expansion, lossless image compression, and extreme network quantization. Our source code is available at https://github.com/WooKyoungHan/LosslessINR.

Despite significant advances in learning-based lossy compression algorithms, standardizing codecs remains a critical challenge. In this paper, we present the JPEG Processing Neural Operator (JPNeO), a next-generation JPEG algorithm that maintains full backward compatibility with the current JPEG format. Our JPNeO improves chroma component preservation and enhances reconstruction fidelity compared to existing artifact removal methods by incorporating neural operators in both the encoding and decoding stages. JPNeO achieves practical benefits in terms of reduced memory usage and parameter count. We further validate our hypothesis about the existence of a space with high mutual information through empirical evidence. In summary, the JPNeO functions as a high-performance out-of-the-box image compression pipeline without changing source coding's protocol. Our source code is available at https://github.com/WooKyoungHan/JPNeO.

Existing frameworks for image stitching often provide visually reasonable stitchings. However, they suffer from blurry artifacts and disparities in illumination, depth level, etc. Although the recent learning-based stitchings relax such disparities, the required methods impose sacrifice of image qualities failing to capture high-frequency details for stitched images. To address the problem, we propose a novel approach, implicit Neural Image Stitching (NIS) that extends arbitrary-scale super-resolution. Our method estimates Fourier coefficients of images for quality-enhancing warps. Then, the suggested model blends color mismatches and misalignment in the latent space and decodes the features into RGB values of stitched images. Our experiments show that our approach achieves improvement in resolving the low-definition imaging of the previous deep image stitching with favorable accelerated image-enhancing methods. Our source code is available at https://github.com/minshu-kim/NIS.

Modern displays and contents support more than 8bits image and video. However, bit-starving situations such as compression codecs make low bit-depth (LBD) images (<8bits), occurring banding and blurry artifacts. Previous bit depth expansion (BDE) methods still produce unsatisfactory high bit-depth (HBD) images. To this end, we propose an implicit neural function with a bit query to recover de-quantized images from arbitrarily quantized inputs. We develop a phasor estimator to exploit the information of the nearest pixels. Our method shows superior performance against prior BDE methods on natural and animation images. We also demonstrate our model on YouTube UGC datasets for de-banding. Our source code is available at https://github.com/WooKyoungHan/ABCD

Preintegrated volume rendering produces high-quality renderings without increased sampling rates. However, a look-up table of a conventional preintegrated volume rendering requires a dimensionality of two, which disturbs interactive renderings when the transfer function is changed. Furthermore, as the resolution of the volume data set increases, the memory space required is impractical or inefficient, especially on GPUs. In the past, several approximation methods have been proposed to reduce the complexity of both the time and memory requirement, but most of them do not correctly present thin opaque structures within slabs and ignore the self-attenuation. We propose an advanced interactive preintegrated volume rendering algorithm that achieves not only high-quality renderings comparable to the conventional ones, but also $(O(n))$ time and memory space requirements even with the self-attenuation within the slabs applied. The algorithm proposed in this paper decomposes the exponential term of the ray integration equation into a power series of a finite order in the form of a linear combination to build a one-dimensional look-up table. Moreover, the proposed algorithm effectively applies the self-attenuation that is caused by fully opaque isosurfaces, by introducing an opaque prediction table. Experimental results demonstrate that the proposed algorithm offers renderings visibly identical to existing preintegrated volume renderings without degrading rendering speed.

Volume ray-casting with a higher order reconstruction filter and/or a higher sampling rate has been adopted in direct volume rendering frameworks to provide a smooth reconstruction of the volume scalar and/or to reduce artifacts when the combined frequency of the volume and transfer function is high. While it enables high-quality volume rendering, it cannot support interactive rendering due to its high computational cost. In this paper, we propose a fast high-quality volume ray-casting algorithm which effectively increases the sampling rate. While a ray traverses the volume, intensity values are uniformly reconstructed using a high-order convolution filter. Additional samplings, referred to as virtual samplings, are carried out within a ray segment from a cubic spline curve interpolating those uniformly reconstructed intensities. These virtual samplings are performed by evaluating the polynomial function of the cubic spline curve via simple arithmetic operations. The min max blocks are refined accordingly for accurate empty space skipping in the proposed method. Experimental results demonstrate that the proposed algorithm, also exploiting fast cubic texture filtering supported by programmable GPUs, offers renderings as good as a conventional ray-casting algorithm using high-order reconstruction filtering at the same sampling rate, while delivering 2.5x to 3.3x rendering speed-up.