Meta tags:
author= James Tompkin;
description= James Tompkin—Associate Professor at Brown University. Visual computing research: computer vision, graphics, and HCI; multi-camera and dynamic scene reconstruction.;
Headings (most frequently used words):
papers, authors, in, this, thread, reconstruction, students, associate, professor, visual, computing, contact, acknowledgements, scene, dynamic, 3d, related, and, courses, james, tompkin, practical, object, monocular, active, illumination, for, controllable, generative, models, light, fields, from, display, to, 4d, algorithms, editing, video, by, recovering, structure, workshops, university, doctoral, masters, undergraduate, extended, family, phds, biography, 2024, academic, lineage, siggraph, 50th, 2023, faculty, tenure, application, materials, exhibitions,
Text of the page (most frequently used words):
and (210), the (146), for (133), james (102), #tompkin (98), from (90), scene (79), with (71), computer (66), video (61), vision (58), conference (54), dynamic (53), light (50), view (49), fields (47), thread (47), kim (45), reconstruction (44), depth (43), image (40), 2021 (37), graphics (37), 2020 (35), time (34), 2024 (33), publications (33), object (33), multi (30), siggraph (29), christian (29), 2025 (29), that (29), 2023 (28), neural (28), field (27), transactions (27), synthesis (25), 2022 (25), 2018 (24), editing (24), interactive (24), 2019 (24), gaussian (24), hanspeter (23), pfister (23), international (23), visualization (22), cvpr (22), kwang (22), controllable (21), monocular (21), 2015 (20), images (20), research (19), theobalt (19), this (19), camera (19), brown (18), 2017 (18), visual (18), radiance (18), geometry (18), via (18), real (18), models (18), huang (18), estimation (18), jan (17), kautz (17), thesis (17), gan (17), can (17), structure (17), content (16), generation (16), consistent (16), material (15), human (15), large (15), scenes (15), reality (15), learning (15), rendering (15), model (15), flight (15), face (15), differentiable (15), consistency (15), how (15), european (15), motion (15), university (14), phd (14), segmentation (14), quality (14), lighting (14), temporal (14), 2012 (14), display (14), into (14), illumination (14), new (13), work (13), computing (13), scalable (13), shape (13), diffusion (13), active (13), objects (13), recovering (13), min (13), asia (13), pattern (13), papers (13), using (12), practical (12), 2014 (12), 2016 (12), systems (12), based (12), translation (12), fast (12), splatting (12), eccv (12), project (12), recognition (12), generative (12), computational (11), attention (11), machine (11), sensor (11), wang (11), dense (11), are (11), numair (11), khan (11), information (11), per (11), what (11), moving (10), jeff (10), aaron (10), gokaslan (10), cameras (10), perception (10), gaussians (10), tof (10), flow (10), towards (10), baseline (10), free (10), tracing (10), path (10), aware (10), zhang (10), algorithms (10), iccv (10), richardt (10), across (10), captured (10), processing (9), deformation (9), removing (9), spatio (9), appearance (9), now (9), displays (9), paper (9), blind (9), representation (9), than (9), then (9), authors (8), our (8), videos (8), unsupervised (8), generating (8), unstructured (8), data (8), novel (8), physically (8), qian (8), 2013 (8), creation (8), code (8), british (8), input (8), sparse (8), background (8), david (8), also (7), its (7), interaction (7), methods (7), 2011 (7), guided (7), automatic (7), daniel (7), local (7), density (7), superpixel (7), style (7), decomposition (7), laidlaw (7), not (7), like (7), networks (7), texture (7), analysis (7), benjamin (7), attal (7), yiwen (7), brittle (7), shot (7), surface (7), page (7), eurographics (7), uist (7), related (7), nicolas (7), bonneel (7), bmvc (7), two (7), over (7), where (7), single (7), raw (7), google (6), augmented (6), science (6), wojciech (6), matusik (6), harvard (6), scale (6), pan (6), zoom (6), michail (6), schwab (6), stamps (6), gaussigan (6), unposed (6), silhouettes (6), stereo (6), adaptive (6), through (6), 360 (6), tax (6), social (6), control (6), long (6), live (6), modern (6), supervision (6), structured (6), michael (6), svbrdf (6), disentangling (6), portrait (6), marc (6), complexity (6), edges (6), rgb (6), 2026 (6), zero (6), yiqing (6), liang (6), mikhail (6), okunev (6), user (6), process (6), interface (6), directly (6), media (6), forum (6), pen (6), technology (6), noise (6), sun (6), neurips (6), one (6), plus (6), capture (6), ai4cc (6), matthew (6), collapse (6), array (6), frame (6), lenslet (6), measurements (6), meta (5), lab (5), college (5), london (5), student (5), institute (5), techniques (5), led (5), design (5), youssef (5), mejjati (5), xiuchao (5), hakyeong (5), improving (5), 6dof (5), törf (5), eliot (5), dead (5), flowed (5), shen (5), reconstructing (5), vikas (5), thamizharasan (5), controlled (5), rules (5), runfeng (5), bidirectional (5), gaufre (5), improve (5), area (5), course (5), csci (5), virtual (5), symposium (5), factors (5), chi (5), miguel (5), granados (5), oliver (5), emerging (5), technologies (5), match (5), short (5), ieee (5), applications (5), toole (5), weiwei (5), jiamin (5), clip (5), gradient (5), space (5), filter (5), rather (5), reflectance (5), regions (5), optimisation (5), only (5), first (5), derived (5), separate (5), without (5), forward (5), hosted (4), github (4), com (4), professor (4), dec (4), art (4), students (4), statement (4), low (4), prof (4), informatics (4), dynamics (4), edit (4), contact (4), gaurav (4), bharaj (4), connectomics (4), evaluating (4), haehn (4), non (4), indoor (4), bundle (4), adjusting (4), lumigraph (4), omnidirectional (4), handwriting (4), descriptors (4), semantic (4), masters (4), both (4), 3dmm (4), biases (4), slant (4), matryodshka (4), sphere (4), smartphone (4), decompositions (4), language (4), optimizing (4), indirectly (4), edge (4), supervised (4), many (4), production (4), cvmp (4), cinemagraphs (4), peter (4), removal (4), crowded (4), demo (4), joint (4), alexa (4), painting (4), graph (4), places (4), resolution (4), effect (4), software (4), regularization (4), wave (4), sylvain (4), paris (4), deqing (4), kalyan (4), sunkavalli (4), intrinsic (4), multicut (4), stefanie (4), tellex (4), robot (4), arrays (4), any (4), michelle (4), borkin (4), 3dv (4), qixing (4), hujun (4), bao (4), wacv (4), shading (4), measurement (4), domain (4), pose (4), other (4), under (4), every (4), makes (4), loss (4), point (4), lets (4), uses (4), jointly (4), fuses (4), colour (4), each (4), email (4), bluesky (4), bsky (4), brownvc (4), theme (3), open (3), source (3), ryan (3), intel (3), foundation (3), industry (3), bbc (3), collaborators (3), here (3), tenure (3), position (3), york (3), city (3), vivek (3), ramanujan (3), program (3), have (3), teaching (3), find (3), king (3), post (3), max (3), planck (3), applied (3), associate (3), his (3), doctoral (3), has (3), level (3), characters (3), creating (3), database (3), segmentations (3), sliders (3), vector (3), scanerf (3), mixtures (3), floatingfusion (3), stabilized (3), omnisdf (3), signed (3), distance (3), functions (3), binoctrees (3), residency (3), lucas (3), kasser (3), henry (3), stone (3), isa (3), milefchik (3), decoupled (3), atsunobu (3), kotani (3), remote (3), fusion (3), deep (3), zhiqiu (3), xinjie (3), convolutional (3), computers (3), processes (3), use (3), eye (3), benchmark (3), during (3), exploring (3), mask (3), hand (3), manipulation (3), purvi (3), goel (3), scholar (3), mapeke (3), beyond (3), incomplete (3), self (3), volume (3), wild (3), see (3), spatial (3), visibility (3), kwon (3), spring (3), workshop (3), imaging (3), projects (3), hci (3), group (3), george (3), surfaces (3), fabrizio (3), pece (3), moment (3), imagery (3), hans (3), seidel (3), includes (3), dataset (3), samuel (3), muff (3), mccann (3), collections (3), supplemental (3), inpainting (3), transition (3), ucl (3), lenticular (3), prints (3), device (3), panoramic (3), context (3), character (3), efficient (3), high (3), order (3), examples (3), gary (3), mdpi (3), pdf (3), continuous (3), line (3), doi (3), produces (3), occlusion (3), won (3), dong (3), jang (3), embedding (3), book (3), immersive (3), yang (3), andreas (3), meuleman (3), yujun (3), sensing (3), aerial (3), gap (3), idea (3), wide (3), small (3), frames (3), cut (3), against (3), recovery (3), harder (3), produce (3), moves (3), pixels (3), sub (3), set (3), views (3), given (3), second (3), enable (3), conditioned (3), problem (3), applies (3), specular (3), few (3), identity (3), sacrificing (3), including (3), but (3), synthetic (3), recover (3), street (3), reconstruct (3), especially (3), complex (3), researcher (3), pages (2), jekyll (2), basic (2), orderedlist (2), web (2), munity (2), gara (2), mond (2), typo (2), graphy (2), joshua (2), hibbert (2), johnston (2), html5 (2), boiler (2), plate (2), activision (2), adobe (2), amazon (2), cognex (2), snap (2), epsrc (2), nsf (2), darpa (2), nasa (2), funding (2), intrepid (2), acknowledgements (2), experience (2), used (2), case (2), festival (2), november (2), museum (2), supported (2), pair (2), exhibition (2), tried (2), faculty (2), application (2), 50th (2), was (2), poster (2), about (2), msci (2), engd (2), doc (2), engineering (2), tools (2), create (2), recent (2), developed (2), performances (2), walking (2), automata (2), metallophone (2), sounds (2), proofreading (2), graphical (2), cnns (2), umass (2), booc (2), education (2), system (2), hierarchical (2), concept (2), maps (2), linear (2), plans (2), easypz (2), binding (2), visualizations (2), timelines (2), visconnect (2), distributed (2), event (2), synchronization (2), collaborative (2), svgs (2), multithread (2), vdom (2), allen (2), transparent (2), latent (2), common (2), sense (2), machines (2), centric (2), verification (2), combat (2), deepfakes (2), why (2), conditional (2), cmu (2), yuanhao (2), typer (2), gaze (2), behavior (2), typing (2), unpaired (2), facebook (2), portal (2), ble (2), intuitive (2), unbounded (2), mit (2), zejiang (2), modeling (2), martin (2), snower (2), deferred (2), constrained (2), eleanor (2), tursman (2), selena (2), ling (2), nick (2), labs (2), scientist (2), hdr (2), four (2), intelligence (2), variation (2), labels (2), hojung (2), zhu (2), photography (2), courses (2), tutorial (2), instrument (2), viewer (2), jennifer (2), beste (2), yuksel (2), donnerer (2), anthony (2), steed (2), brain (2), bci (2), kartic (2), subr (2), photos (2), seen (2), early (2), videoscapes (2), useful (2), building (2), graphs (2), matching (2), construction (2), textures (2), viewpoint (2), printing (2), angular (2), tap (2), artifact (2), transitions (2), tasks (2), gestures (2), helge (2), rhodin (2), mapping (2), jin (2), super (2), semi (2), label (2), zheng (2), upon (2), direct (2), pacific (2), evgeny (2), levinkov (2), john (2), brian (2), lee (2), verena (2), kaynig (2), lichtman (2), robotics (2), eric (2), rosen (2), whitney (2), elizabeth (2), phillips (2), chien (2), konidaris (2), communicating (2), arm (2), intent (2), mixed (2), head (2), mounted (2), follows (2), filtering (2), criteria (2), ranking (2), bundles (2), ritchie (2), journal (2), controlling (2), svg (2), insertion (2), jing (2), piecewise (2), planar (2), donglai (2), wei (2), saffo (2), shash (2), sinha (2), cody (2), dunne (2), digital (2), world (2), tvcg (2), changil (2), fumeng (2), lane (2), harrison (2), effects (2), accuracy (2), scatterplots (2), cvm (2), han (2), hyunho (2), kang (2), flash (2), srinath (2), sridhar (2), correlation (2), aarrushi (2), shandilya (2), arxiv (2), nguyen (2), lei (2), xiao (2), douglas (2), lanman (2), winter (2), difficult (2), hyeongwoo (2), tmlr (2), adam (2), harley (2), leonidas (2), guibas (2), volodymyr (2), kuleshov (2), yifan (2), peng (2), guo (2), orazio (2), gallo (2), yaoan (2), gao (2), huamin (2), belal (2), shaheen (2), treedgs (2), distant (2), dbh (2), dongyoung (2), choi (2), splat (2), seamless (2), clear (2), adds (2), runs (2), propagates (2), magnitude (2), posed (2), enough (2), multiple (2), flicker (2), actually (2), decomposes (2), aligning (2), required (2), visible (2), authoring (2), tool (2), pipelines (2), stabilisation (2), chosen (2), mpi (2), inf (2), stabilised (2), stabilising (2), takes (2), same (2), photograph (2), edits (2), person (2), just (2), approach (2), decompose (2), cross (2), plausible (2), modifications (2), possible (2), central (2), propagated (2), aperture (2), way (2), epi (2), direction (2), points (2), whose (2), weights (2), reprojection (2), all (2), superpixels (2), stay (2), output (2), treats (2), optimise (2), targeted (2), glasses (2), captures (2), fills (2), record (2), rays (2), present (2), humans (2), interact (2), computationally (2), relativistic (2), stylegan2 (2), tricks (2), fid (2), unconditional (2), stylegan (2), map (2), diffuse (2), downstream (2), coarse (2), alone (2), generator (2), writer (2), out (2), samples (2), trains (2), sits (2), principled (2), sample (2), exactly (2), requires (2), efficiently (2), want (2), preserving (2), carries (2), sensors (2), 3dgs (2), when (2), volumetric (2), baselines (2), training (2), 100 (2), processed (2), better (2), nerf (2), rendered (2), integrate (2), spacetime (2), modalities (2), reduce (2), resolve (2), ambiguities (2), interference (2), estimate (2), modelling (2), avoid (2), errors (2), feed (2), trained (2), casual (2), apples (2), method (2), warping (2), focuses (2), reconstructs (2), semantics (2), ambiguity (2), ordinary (2), rig (2), fits (2), regime (2), matches (2), shadow (2), reflections (2), global (2), these (2), tile (2), tiles (2), phone (2), often (2), conditions (2), photographs (2), center (2), room (2), 547 (2), 115 (2), waterman (2), providence (2), 02912 (2), folks (2), save (2), gcal (2), include (2), agenda (2), instructions (2), office (2), hours (2), weds (2), 1300 (2), appointment (2), orcid (2), jamestompkin (2), javascript (2), address (2), onboarding (2), share, sent, apply, track, assistant, february, march, luxembourg, film, may, imaginales, epinal, france, iscp, april, desom, rear, window, bad, discover, club, risd, arts, put, contribute, myself, exhibitions, sept, materials, chair, posters, which, coincidentally, running, 20th, iteration, too, outstanding, contributors, res, png, history, please, summary, newer, tab, ian, mackie, sponsored, univer, sity, veiv, intern, disney, cambridge, vci, paulson, school, sciences, academic, lineage, intersection, helps, develop, experiences, creates, exploration, creative, postdoctoral, helped, within, biography, defender, boston, synthesia, alibaba, extended, family, phds, lightfield, voxelized, representations, berkeley, environment, telerobotics, more, jacob, rethinking, jayden, harry, deblurring, exposure, astrophotography, general, troy, conklin, agency, perceptions, creativity, artistic, agents, anika, bahl, undergraduate, spectrum, cornell, fauxtoshop, operations, kernel, prediction, parameter, blocks, ziyin, unaligned, domains, pathtracing, stanford, untrusted, choreography, clustering, congressional, innovation, utoronto, amherst, rice, rasterizer, irradiance, niiv, implicit, isotropic, dimensions, spatiotemporal, luma, ashley, joel, salzman, summer, cs171, gisp, 0002, nfts, blockchain, ally, nikolas, lazar, mentoring, 1950, game, engines, fall, 2000, 2951, 1290, 1430, introduction, videography, england, conventional, physics, inspired, workshops, dissertation, 2006, venues, networked, webpage, contains, events, schematics, webgl, annual, people, bcs, 2009, sheridan, abel, maciel, roussos, diy, 2010, touch, p300, interfaces, select, physical, feng, yebin, liu, carsten, stoll, qionghai, dai, henrik, lieng, perspective, grau, later, publication, cdfg, stanislav, jakuschevskij, correspondence, search, theobald, databases, alt, title, philippe, levieux, archived, discovery, varying, simon, heinzle, preference, study, rajvi, shah, shahram, izadi, contexts, curvature, riemannian, submanifolds, ghost, generalized, kiran, varanasi, vidicontexts, tpami, younghee, hyung, compression, manifolds, propagation, explicit, relationship, animation, sca, stelian, coros, bernhard, thomaszewski, bernd, bickel, levin, yun, fei, changxi, builds, edilson, aguiar, generalizing, steffen, kirchhoff, bjoern, andres, hendrik, strobelt, colin, fredericks, connor, huff, dana, higgins, anton, strezhnev, maya, komisarchik, special, issue, hoffer, matejek, adi, suissa, peleg, ali, awami, kamentsky, felix, gonda, eagon, meng, william, richard, schalek, alyssa, wilson, toufiq, parag, johanna, beyer, thouis, jones, markus, hadwiger, isrr, hri, serena, booth, krzysztof, gajos, jim, waldo, radhika, nagpal, piggybacking, robots, overtrust, dormitory, security, predictor, combination, test, could, called, separately, tracking, etra, alexandra, papoutsaki, yuze, juhyun, park, tensor, attribute, darren, cosker, ijrr, injecting, bookmark, hosts, docs, jsfiddle, ocks, org, sicheng, hao, olga, vitek, 1145, 3359997, 3365708, continuum, vrcai, joanna, tarko, jiaju, xiangyu, haoming, lai, hughes, byproduct, medical, assisted, intervention, miccai, salma, magid, denis, schapiro, sorger, channel, informative, protein, identification, highly, multiplexed, yixuan, cristina, nita, rotaru, chapter, springer, gordon, wetzstein, loudon, cohen, guesman, nicholas, bond, testing, privileged, adapting, features, statistical, dependence, cue, classification, task, appeared, beatrix, emőke, fülöp, balogh, julie, digne, curtis, wigington, tong, healey, dually, noted, layout, annotations, iccp, hyun, joo, dahyun, acquisition, state, report, yiheng, xie, towaki, takikawa, shunsuke, saito, litany, shiqin, yan, federico, tombari, vincent, sitzmann, zihan, biomedical, isbi, xingxuan, leahy, helen, dalit, ben, yosef, needleman, quantized, amodal, blastomere, instance, yuxin, aid, three, studies, judgments, celine, aubuchon, jovan, kemp, fulvio, domini, ijcv, alexander, meyerowitz, xin, zhengqin, thu, phuoc, yue, gives, insight, prompt, akin, caliskan, berkay, kicanaoglu, hyeonjoong, yuanbo, xiaohan, praccho, muna, mcquay, liuren, yin, mikaela, angelina, openreview, public, reviews, response, zixuan, anh, duong, abhishek, badki, hang, chung, tongyu, zhou, ivy, chen, kevin, hsu, rossi, siu, shunan, franck, dernoncourt, infovids, reimagining, alternative, presenter, relationships, changqing, zou, minh, hieu, bach, thuan, bui, shubham, tim, fairley, zane, jaemin, cho, woohyun, filters, catalogue, perhaps, collaboration, genai, selected, extends, rigs, transfer, scheme, expensive, subset, saving, coloured, scribbles, supervoxel, solved, feel, responsive, once, boundaries, chained, binary, stabilises, borrowing, regularity, unprocessed, agnostic, demonstrated, stylisation, hybrid, split, orders, prior, support, refinement, compositing, inpaints, revealed, candidate, homographies, sidestepping, full, earlier, filling, hole, occluded, pitched, occlusions, previous, had, attempted, selection, loop, detection, looping, clips, region, spans, postdoc, years, liris, cnrs, earliest, piece, isolated, extension, different, angle, right, before, changes, must, propagate, consistently, separating, require, understanding, underlying, manipulating, inverse, disoccluded, completed, diffused, obtained, diffusing, positions, depths, differentiably, optimised, scales, 50k, needed, trivial, segments, horizontal, vertical, epis, clusters, respect, shifts, being, outwards, does, double, duty, orientation, 150, millimetre, something, fix, advance, solve, size, arrangement, trade, matters, validated, resulting, dual, purpose, senses, precursor, write, arcs, sit, reshape, turns, kaist, epipolar, plane, routine, function, fully, describes, refocusing, parallax, displayed, offer, challenge, stresses, devices, hardware, regularised, proven, convergence, minimalist, architecture, stripped, usual, beat, ffhq, imagenet, cifar, stacked, mnist, compete, disentangles, frozen, flame, morphable, identifies, noisy, directions, residual, entanglement, damps, them, stochastic, jacobian, regulariser, formalises, maths, fixes, close, controllability, longer, costs, estimates, normals, albedo, roughness, frequency, editor, relight, learns, anisotropic, masks, drive, posing, splits, class, bounding, box, inserted, diverse, surroundings, letting, generate, held, generators, discriminators, localise, disturbing, inter, compositional, controls, primitives, matter, taxfreegan, closes, conditioning, shows, disentanglement, strategy, kills, r3gan, alongside, arc, architectural, reset, drop, bag, impressive, mind, axes, articulate, photorealism, brought, relevance, place, there, usually, versus, tradeoff, manage, optically, main, floating, lens, unknown, calibrates, snapshot, ambient, recovers, higher, fidelity, commodity, partially, backbone, heuristics, stabilise, otherwise, measured, comparable, while, faster, vectors, estimated, phasors, coherent, less, error, establishes, phasor, added, showing, handling, rethinks, heterogeneous, built, robustness, since, resample, swinging, baseball, bat, slow, typically, their, pixel, treat, simplifying, assumptions, flying, artifacts, needs, readings, further, predicts, million, recipe, generalises, davis, robotap, categorised, differences, resolvable, get, swamped, uniformly, casts, canonical, template, static, component, initialised, absorb, roughly, twenty, minutes, renders, carrying, latter, foreground, approached, ways, fit, consider, regularisations, help, monodygaubench, additional, might, optical, saff, msf, driven, millions, transfers, fitting, observing, deforming, tries, fundamentally, instant, progress, depends, well, signals, together, trunks, span, altitude, extracts, opacity, weighted, isolates, trunk, solid, circles, diameter, breast, height, rmse, below, lidar, tackles, drops, anchors, intermediate, viewpoints, smooth, circular, sweep, placing, sdf, inside, adaptively, subdivided, spherical, binoctree, setting, keeps, memory, detail, replaces, lights, differentiates, linearly, transformed, cosines, weighting, relighting, psnr, fifth, targets, curved, refractions, break, mixture, along, ray, warps, volumes, learned, blending, pushes, tiled, hash, grid, mlps, admm, reaches, consensus, giving, poses, assigns, mlp, dependent, branch, distributes, gpus, stays, square, metres, interreflection, refine, mesh, facet, disambiguated, consumer, tackle, problems, indoors, handheld, airborne, interreflect, refract, lit, sources, contribution, specific, matched, own, drone, synthesising, measuring, properties, great
Text of the page (random words):
ly for unstructured capture complex appearance and large places 2020 now practical scene object reconstruction how do we reconstruct scenes and objects from photographs especially for unstructured capture complex appearance and large places in my lab and with our great collaborators we reconstruct scenes and objects from images each project fits a representation of 3d shape and appearance so that the model s rendered images match the captured ones with many images in controlled conditions this problem is solvable what makes it interesting is when we work with few casual captures or with challenging appearance or in large and uncontrolled conditions these cases often contain visual ambiguities and rarely have a single solution we ve tried to tackle this space of problems with large scenes both indoors and out with handheld 360 or airborne cameras with sparse and wide baselines with surfaces that interreflect refract and are lit by many sources across the projects the contribution often focuses on what specific information can reduce the ambiguity a physically based rendering model that better matches real light and cameras or a scene representation matched to the scene s own structure this idea applies from a single object up to a building or street scene from a phone to a drone and from synthesising new views of a scene to measuring its properties authors hujun bao dongyoung choi yaoan gao purvi goel qixing huang hakyeong kim min h kim yifan peng belal shaheen yujun shen vikas thamizharasan huamin wang xiuchao wu jiamin xu weiwei xu papers in this thread shape from tracing towards reconstructing 3d object geometry and svbrdf material from images via differentiable path tracing international conference on 3d vision 3dv 2020 uses differentiable path tracing with global illumination effects like interreflection in the forward model to refine a coarse mesh and its per facet svbrdf so shading shadow and material are jointly disambiguated from images captured by phone and consumer 360 camera scalable neural indoor scene rendering transactions on graphics siggraph 2022 tiles a large indoor scene and assigns a small mlp per tile with a separate view dependent branch for reflections so training distributes across gpus and rendering stays interactive on scenes over 100 square metres scanerf scalable bundle adjusting neural radiance fields for large scale scene rendering transactions on graphics siggraph asia 2023 pushes the tiled nerf idea into the bundle adjusting regime each tile carries a hash grid plus diffuse and specular mlps and admm reaches camera pose consensus across tiles with a specular aware warping loss giving the poses a second optimisation path local gaussian density mixtures for unstructured lumigraph rendering siggraph asia conference papers 2024 targets curved surface reflections and refractions exactly where view consistent global density models break with a per view gaussian mixture density along each ray then warps and fuses these local volumes with learned blending weights for unstructured lumigraph rendering efficient object reconstruction with differentiable area light shading siggraph asia conference papers 2025 replaces point lights with active area lighting during capture then differentiates through linearly transformed cosines plus shadow visibility weighting for shading recovering material at 3 db relighting psnr or matching point light quality from a fifth of the photos related papers omnisdf scene reconstruction using omnidirectional signed distance functions and adaptive binoctrees computer vision and pattern recognition cvpr 2024 reconstructs scenes captured by a small baseline circular sweep of a 360 camera by placing an sdf inside an adaptively subdivided spherical binoctree whose geometry matches the capture setting and keeps memory in line with detail splat based gradient domain fusion for seamless view transition international conference on 3d vision 3dv 2026 tackles the sparse view wide baseline regime where gaussian splatting drops geometry anchors it with two view stereo fills intermediate viewpoints via reprojection and fuses in the gradient domain so colour transitions stay smooth across views treedgs aerial gaussian splatting for distant dbh measurement mdpi remote sensing 2026 treats aerial gaussian splatting as a measurement instrument trunks span only a few pixels from altitude so the method extracts a dense opacity weighted point set isolates trunk samples and fits solid circles to estimate diameter at breast height 4 79 cm rmse below a lidar baseline collapse monocular dynamic 3d reconstruction with only ordinary rgb video no depth sensor no rig can we recover dynamic 3d scene geometry and motion 2023 now monocular dynamic 3d reconstruction with only ordinary rgb video no depth sensor no rig can we recover dynamic 3d scene geometry and motion monocular dynamic 3d reconstruction takes a single moving camera observing a deforming scene and tries to recover a 4d representation including geometry appearance and motion the problem is fundamentally under constrained at any one instant and progress depends on how well the chosen scene representation and supervision signals work together we ve approached this in two ways first per scene methods fit a representation to a single video we consider what motion models and regularisations can help gaufre monodygaubench and what additional information might resolve the ambiguity e g semantics attention and optical flow supervision saff second zero msf is data driven a feed forward model trained on millions of synthetic examples that transfers zero shot to real video with no per scene fitting authors orazio gallo leonidas j guibas adam harley numair khan eliot laidlaw douglas lanman yiqing liang runfeng li mikhail okunev lei xiao papers in this thread semantic attention flow fields for monocular dynamic scene decomposition international conference on computer vision iccv 2023 reconstructs a 4d neural volume carrying not just colour and density but also scene flow semantics and attention then uses the latter two to decompose foreground objects from background across spacetime without supervision gaufre gaussian deformation fields for real time dynamic novel view synthesis arxiv dec 2023 wacv 2025 casts monocular dynamic reconstruction as a canonical gaussian template plus a forward warping deformation field with a separate static component initialised to absorb non moving regions so the deformation focuses on what actually moves trains in roughly twenty minutes and renders in real time monocular dynamic gaussian splatting fast brittle and scene complexity rules transactions on machine learning research tmlr 2025 an apples to apples benchmark of monocular dynamic gaussian splatting methods categorised by motion representation method differences are resolvable on synthetic data but get swamped by real world scene complexity and the optimisation is uniformly brittle zero shot monocular scene flow estimation in the wild computer vision and pattern recognition cvpr 2025 a feed forward model that jointly predicts geometry and scene flow trained on a one million sample synthetic recipe generalises zero shot to casual davis video and robotap manipulation scenes no per scene optimisation required collapse active illumination for dynamic 3d reconstruction can physically modelling active illumination directly from raw sensor measurements improve scene estimation and avoid errors from derived depth 2021 now active illumination for dynamic 3d reconstruction can physically modelling active illumination directly from raw sensor measurements improve scene estimation and avoid errors from derived depth time of flight and structured light cameras are typically used as depth sensors their raw measurements are processed into a per pixel depth map and downstream reconstruction methods treat that depth as input but depth processing makes simplifying assumptions about the scene creating noise in low reflectance regions flying pixels under multi path interference and motion artifacts in fast moving scenes each depth estimate needs multiple illumination readings further derived depth is difficult to integrate with other sensor modalities like colour cameras our work rethinks reconstruction for heterogeneous multi shot imaging processes built upon a differentiable forward model of how the active illumination produces the raw sensor output for a given scene our methods optimise a 4d volumetric scene representation like nerf or 3dgs so that rendered measurements match what the sensor captured this lets us integrate sensor measurements over spacetime in a principled way including across modalities to reduce noise resolve ambiguities in multi shot sensing and improve robustness to multi path interference and since we model motion over time we can resample fast motion a swinging baseball bat into slow motion authors benjamin attal aaron gokaslan changil kim hakyeong kim eliot laidlaw runfeng li marc mapeke andreas meuleman matthew o toole mikhail okunev christian richardt aarrushi shandilya papers in this thread törf time of flight radiance fields for dynamic scene view synthesis neural information processing systems neurips 2021 establishes that a 4d scene can be supervised directly by continuous wave tof phasor measurements rather than processed depth with added colour cameras showing low noise super resolution and better multi path handling flowed time of flight radiance fields european conference on computer vision eccv 2024 adds motion vectors that are jointly estimated with geometry uses four raw frames not phasors captured over time from a continuous wave tof sensor to create a coherent dynamic reconstruction 20 less depth error on dynamic objects than the c tof baseline time of the flight of the gaussians optimizing depth indirectly in dynamic radiance fields computer vision and pattern recognition cvpr 2025 applies raw tof supervision to a gaussian splatting backbone with two heuristics that stabilise the otherwise brittle 3dgs optimisation when depth is not directly measured comparable quality to neural volumetric baselines while training 100 faster related papers neural fields for structured lighting international conference on computer vision iccv 2023 carries the supervision by raw measurement approach from tof over to structured light and lets us separate direct and ambient illumination recovers higher fidelity depth on objects than commodity structured light sensors including for partially transparent surfaces floatingfusion depth from tof and image stabilized stereo cameras european conference on computer vision eccv 2022 fuses tof depth with stereo from a smartphone s optically stabilised main rgb camera where the floating lens has unknown pose self calibrates the multi sensor geometry from a single snapshot then fuses via a correlation volume collapse controllable generative models how do we efficiently control generative models to produce what we want preserving identity 3d structure style without sacrificing quality 2018 now controllable generative models how do we efficiently control generative models to produce what we want preserving identity 3d structure style without sacrificing quality a generative model that can sample new content is impressive one that produces exactly what a user has in mind is useful controlling generation requires aligning the model s latent structure with axes a person can articulate identity pose style lighting geometry without sacrificing the photorealism that brought the model to relevance in the first place there is usually a quality versus control tradeoff to manage this thread runs from youssef mejjati s phd work on unsupervised attention for image to image translation through compositional controls object stamps gaussigan s 3d gaussian primitives from silhouettes alone into 3dmm conditioned face generation where yiwen huang s phd now sits two recent moves matter taxfreegan closes the fid gap to unconditional stylegan under 3dmm conditioning and our disentangling 3d work shows that the noise in clip s embedding space not the disentanglement strategy is what kills quality r3gan sits alongside this arc as our architectural reset a principled relativistic loss that lets the modern gan drop its bag of tricks authors aaron gokaslan yiwen huang hyeongwoo kim kwang in kim atsunobu kotani volodymyr kuleshov youssef a mejjati isa milefchik zejiang shen michael snower stefanie tellex vikas thamizharasan oliver wang xinjie yi zhiqiu yu qian zhang papers in this thread unsupervised attention guided image to image translation neural information processing systems neurips 2018 jointly trains attention with generators and discriminators so unsupervised image to image translation can localise edits to objects without disturbing background or inter object structure generating handwriting via decoupled style descriptors european conference on computer vision eccv 2020 factors handwriting style into separate character level and writer level descriptors letting the model generate new characters in a held out writer s hand from only a few samples generating object stamps ai for content creation ai4cc cvpr 2020 splits conditional object insertion into a mask generator shape given a class and bounding box and a texture generator appearance conditioned on the background so the inserted object is both diverse in shape and consistent with its surroundings gaussigan controllable image synthesis with 3d gaussians from unposed silhouettes british machine vision conference bmvc ai for content creation ai4cc cvpr 2021 learns a coarse 3d object representation as a set of self supervised anisotropic 3d gaussians from unposed 2d masks alone then uses it to drive controllable mask and texture synthesis with interactive posing learning physically based material and lighting decompositions for face editing international conference on computational visual media cvm 2022 estimates per portrait surface normals albedo roughness and a high frequency lighting map and decomposes diffuse and specular reflectance so a downstream editor can relight a face from a single photograph removing the quality tax in controllable face generation winter conference on applications of computer vision wacv ai for content creation ai4cc cvpr 2024 formalises 3dmm conditioned face generation as a maths problem then applies targeted fixes that close the fid gap to unconditional stylegan so controllability no longer costs visible image quality disentangling 3d from large vision language models for controlled portrait generation 2025 disentangles 3d portrait generation from a frozen clip plus a flame morphable model then identifies clip s noisy embedding directions as the residual source of entanglement and damps them with a stochastic jacobian regulariser the gan is dead long live the gan a modern gan baseline neural information processing systems neurips 2024 a regularised relat...
|
