ÌìÃÀ´«Ã½

Recent advances at the interface of machine learning and 3D reconstruction, ranging from volumetric methods like NeRF and Plenoxels to point-based primitives such as Gaussian Splats, have enabled high-fidelity 3D modeling from images. Despite this progress, capturing assets through the isolated optimization of segmented objects remains fundamentally under-constrained. Additionally, thin structures pose a difficult challenge for segmentation methods. In this work, we explore necessary constraints and combine priors from monocular depth as well as visual hulls to overcome their respective failure modes, producing high-quality object-centric reconstructions in the face of erroneous segmentation. This is demonstrated using synthetic scenes, all exhibiting fine structure, which we openly release along with coarse and ground truth segmentation masks. Furthermore, we show that segmentation failure can act as a useful signal to guide sampling and further enhance detail preservation.

In recent years, machine learning has led to significant advances in computer graphics and 3D scene understanding, enabling photorealistic novel view synthesis and accurate 3D reconstruction from 2D images. While implicit methods like neural radiance fields (NeRFs) achieve impressive fidelity, explicit approaches such as voxel grids and 3D Gaussian Splatting offer faster training and inference yet remain limited by memory footprint and sensitivity to occlusion. We introduce Cloning-Plenoxels (Clonoxels), a voxel-based framework that automatically detects repeated objects within a scene and enables information sharing between them, improving reconstruction quality and reducing storage requirements. Clonoxels leverages semantic and instance segmentation, rigid alignment, and voxel-by-voxel matching with lighting-aware corrections, producing volumetric representations that are robust to occlusion. Experiments using synthetic and real-world data demonstrate enhanced recovery of hidden structures, lower-memory scene representation, and denoised 3D geometry reconstruction. Our framework offers a practical path toward more efficient and accurate reconstruction of challenging, complex scenes.

Spectroscopy represents the ideal observational method to maximally extract information from galaxies regarding their star formation and chemical enrichment histories. However, absorption spectra of galaxies prove rather challenging at high redshift or in low mass galaxies, due to the need to spread the photons into a relatively large set of spectral bins. For this reason, the data from many state-of-the-art spectroscopic surveys suffer from low signal-to-noise (S/N) ratios, and prevent accurate estimates of the stellar population parameters. In this paper, we tackle the issue of denoising an ensemble by the use of unsupervised Deep Learning techniques trained on a homogeneous sample of spectra over a wide range of S/N. These methods reconstruct spectra at a higher S/N and allow us to investigate the potential for Deep Learning to faithfully reproduce spectra from incomplete data. Our methodology is tested on three key line strengths and is compared with synthetic data to assess retrieval biases. The results suggest a standard Autoencoder as a very powerful method that does not introduce systematics in the reconstruction. We also note in this work how careful the analysis needs to be, as other methods can -- on a quick check -- produce spectra that appear noiseless but are in fact strongly biased towards a simple overfitting of the noisy input. Denoising methods with minimal bias will maximise the quality of ongoing and future spectral surveys such as DESI, WEAVE, or WAVES.

Leveraging machine learning techniques, in the context of object-based media production, could enable provision of personalized media experiences to diverse audiences. To fine-tune and evaluate techniques for personalization applications, as well as more broadly, datasets which bridge the gap between research and production are needed. We introduce and publicly release such a dataset, themed around a UK weather forecast and shot against a blue-screen background, of three professional actors/presenters – one male and one female (English) and one female (British Sign Language). Scenes include both production and research-oriented examples, with a range of dialogue, motions, and actions. Capture techniques consisted of a synchronized 4K resolution 16-camera array, production-typical microphones plus professional audio mix, a 16-channel microphone array with collocated Grasshopper3 camera, and a photogrammetry array. We demonstrate applications relevant to virtual production and creation of personalized media including neural radiance fields, shadow casting, action/event detection, speaker source tracking and video captioning.

Leveraging machine learning techniques, in the context of object-based media production, could enable provision of personalized media experiences to diverse audiences. To fine-tune and evaluate techniques for personalization applications, as well as more broadly, datasets which bridge the gap between research and production are needed. We introduce and publicly release such a dataset, themed around a UK weather forecast and shot against a blue-screen background, of three professional actors/presenters – one male and one female (English) and one female (British Sign Language). Scenes include both production and research-oriented examples, with a range of dialogue, motions, and actions. Capture techniques consisted of a synchronized 4K resolution 16-camera array, production-typical microphones plus professional audio mix, a 16-channel microphone array with collocated Grasshopper3 camera, and a photogrammetry array. We demonstrate applications relevant to virtual production and creation of personalized media including neural radiance fields, shadow casting, action/event detection, speaker source tracking and video captioning.