Overview
Kupa kwakasiyana kunoita kuti maitiro ekushandura chiitiko che 3D kuita chifananidzo che 2D chinyatsosiyaniswa, saka unogona kuverengera magradients kubva kumapixels akashandurwa kudzokera kumamiriro echiitiko. Izvi zvinokutendera kuti ukwidze geometry, zviwanikwa, mwenje, uye kamera uchishandisa gradient kudzika.
Differentiable Rendering ndeyekombuta-yekuona workflows inodudzira kana kuburitsa midhiya yekuona yekuongorora, mashandiro, uye kugadzira.
Deep Dive
Chinyakare shanduro inzira yenzira imwe chete: dyisa mu geometry, zvishandiswa, mwenje, uye kamera, uye mapikisesi anobuda. Inosiyanisa reverses reverses inoyerera nekombuta kuti yega yega pixel inobuda inoshanduka sei nekuremekedza kune yega yega paramende. Neaya magradients, optimizer inogona kugadzirisa chimiro che 3D kana maumbirwo ayo kudzamara mufananidzo wakapihwa uchienderana nemufananidzo waunonangwa, unova moyo wekupenengura uye kuongorora-ne-synthesis. Chinetso chikuru ndechekuti kupa kunosanganisira discontinuities, kunyanya pachinhu silhouettes uye occlusion edges, uko pixel inosvetuka kubva kumberi ichienda kumashure. Nzira dzakaita senge yakapfava rasterization (SoftRas), edge-sampling (Li et al.'s redner), uye rasterizer muPyTorch3D inobata izvi nekutsetseka kana yakakosha muganho wekubatanidza. NeRF kudzidziswa uye 3D Gaussian splatting akakurumbira maapplication.
Technical Insight
Dambudziko guru nderokuonekwa discontinuities. Pane silhouette yechinhu pixel inotora kubva kumberi kuenda kumashure, saka naive derivative iri zero kunenge kwese kwese uye isina kutsanangurwa kumucheto, isingapi hunobatsira gradient nezve chimiro. Mhinduro dzinogona kupfavisa chivharo kuitira kuti makonanhatu ape kutsvedzerera, kusajeka kwetsoka kune mapikisesi ari pedyo (soft rasterization) kana sampuli zvakajeka mumipendero kuverengera nguva yemuganhu wezvakabatanidzwa (kumucheto sampling).
Kugona Kusiyanisa Kushandura
Kupa kwakasiyana kunoita kuti maitiro ekushandura chiitiko che 3D kuita chifananidzo che 2D chinyatsosiyaniswa, saka unogona kuverengera magradients kubva kumapixels akashandurwa kudzokera kumamiriro echiitiko. Izvi zvinokutendera kuti ukwidze geometry, zviwanikwa, mwenje, uye kamera uchishandisa gradient kudzika. Differentiable Rendering ndeyekombuta-yekuona workflows inodudzira kana kuburitsa midhiya yekuona yekuongorora, mashandiro, uye kugadzira. Kuti uvake kunzwisisa kwakadzama, bata Kupa Kunosiyana semuenzaniso wekushandisa, kwete chinhu chimwe chete: tsanangura zvinodikanwa, kujekesa fungidziro, uye patsanura zvinogona kuitwa nehurongwa hwakavimbika kubva kune zvichiri kuda kutonga kwenyanzvi.
Mukuita, zvikwata zvakasimba zvinoshandisa Differentiable Rendering chiyero chechokwadi nemashandiro anoita semhando yedata, kusiyana kwemwenje, uye kuenderana kwemazita. Ivo vanonyora zvakajeka maitiro ebudiriro, bvunzo vachipokana ne data rechokwadi uye mafambiro ebasa, uye iterate zvichibva pane zvakacherechedzwa maitiro ekutadza kwete kuhwina-nguva imwe chete yebhenji. Apa ndipo apo kunzwisisa kwe theoretical kunoshanduka kuve kugona kwakasimba pane chigadzirwa, mutemo, uye mashandiro.
Visual AI inogona kuita otomatiki yekuongorora, yekuona, uye yekumaka mabasa pachiyero. Panguva imwecheteyo, kodzero dzeMufananidzo uye kubvumirwa kunogona kuve njodzi dzepamutemo kana hunhu husina kujeka. Nzira yakatsiga ndeyekubatanidza kukurumidza kuyedza nekutonga: mhanyisa vatyairi vendege, tora humbowo, buritsa matanda esarudzo, uye urambe uchivandudza chengetedzo semaitiro emuenzaniso, zvinotarisirwa nemushandisi, uye zvinodikanwa zvekutonga.
Strategic Impact
Visual AI inogona kuita otomatiki yekuongorora, yekuona, uye yekumaka mabasa pachiyero.
Visual AI inogona kuita otomatiki yekuongorora, yekuona, uye yekumaka mabasa pachiyero. Mukutumirwa kwemhando yepamusoro, izvi zvinoshandurirwa kuita mitemo inoyerwa yekushanda, miganhu yevaridzi, uye tsika dzekudzokorora dzinodzokororwa kuitira kuti zvikwata zvikwire kuvimba pane kukwidza kusajeka.
Zvikwata zvekugadzira zvinogona prototype pfungwa nekukurumidza nekudzokororwa kwemaoko mashoma.
Zvikwata zvekugadzira zvinogona prototype pfungwa nekukurumidza nekudzokororwa kwemaoko mashoma. Mukutumirwa kwemhando yepamusoro, izvi zvinoshandurirwa kuita mitemo inoyerwa yekushanda, miganhu yevaridzi, uye tsika dzekudzokorora dzinodzokororwa kuitira kuti zvikwata zvikwire kuvimba pane kukwidza kusajeka.
Mashandisirwo anogona kushandisa masaini emifananidzo nemavhidhiyo ayo aimbove akaoma kugadzirisa.
Mashandisirwo anogona kushandisa masaini emifananidzo nemavhidhiyo ayo aimbove akaoma kugadzirisa. Mukutumirwa kwemhando yepamusoro, izvi zvinoshandurirwa kuita mitemo inoyerwa yekushanda, miganhu yevaridzi, uye tsika dzekudzokorora dzinodzokororwa kuitira kuti zvikwata zvikwire kuvimba pane kukwidza kusajeka.
Real-World Implementation
Kugadzira patsva chimiro uye magadzirirwo echinhu che3D kubva kune mashoma emifananidzo nekugadzirisa modhi kusvika marenders achienderana nemifananidzo (inverse rendering).
Kudzidzira maNeRF uye 3D Gaussian splats, uko gradients kubva kune yakashandurwa maonero anovandudza chimiro chechiitiko.
Kufungidzira zvimiro zvechinhu (hukasha, ratidziro) nekufananidza zvakapihwa mapeji nemufananidzo chaiwo.
Kamera uye pose calibration mumarobhoti, inokodzera inozivikanwa 3D modhi kumufananidzo wekamera kuti utore nzvimbo yayo.
Maitiro Ekuita
Differentiable Rendering mukuita
Kugadzira patsva chimiro uye magadzirirwo echinhu che3D kubva kune mashoma emifananidzo nekugadzirisa modhi kusvika marenders achienderana nemifananidzo (inverse rendering).
Kugadzira patsva chimiro chechinhu che3D uye magadzirirwo kubva kune mashoma emifananidzo nekugadzirisa modhi kusvika mareferensi anoenderana nemifananidzo (inverse rendering) Matimu anowanzo kuwana mhedzisiro iri nani kana achinge atsanangura emhando yepamusoro kumberi, chengetedza nzira yekukwira kwevanhu yemakesi emupendero, uye tarisa zvese zvakawanikwa zvechigadzirwa uye kukanganisa mutengo nekufamba kwenguva.
Differentiable Rendering mukuita
Kudzidzira maNeRF uye 3D Gaussian splats, uko gradients kubva kune yakashandurwa maonero anovandudza chimiro chechiitiko.
Kudzidzisa maNeRFs uye 3D Gaussian splats, uko gradients kubva kune yakapihwa maonero anovandudza chiitiko chinomiririra Matimu anowanzo kuwana mhedzisiro iri nani kana achinge atsanangura emhando yepamusoro kumberi, chengetedza nzira yekukwira kwevanhu yemakesi ekumucheto, uye kuteedzera zvese zvakawanikwa zvechigadzirwa uye mutengo wekukanganisa nekufamba kwenguva.
Differentiable Rendering mukuita
Kufungidzira zvimiro zvechinhu (hukasha, ratidziro) nekufananidza zvakapihwa mapeji nemufananidzo chaiwo.
Kufungidzira zvinhu zvechinhu chechinhu (hukasha, ratidziro) nekufananidza zvakapihwa mapikicha chaiwo Matimu anowanzo kuwana mhedzisiro iri nani kana achinge atsanangura emhando yepamusoro kumberi, chengetedza nzira yekukwira kwevanhu yemakesi emupendero, uye kuteedzera zvese zvakawanikwa zvechigadzirwa uye mutengo wekukanganisa nekufamba kwenguva.
Differentiable Rendering mukuita
Kamera uye pose calibration mumarobhoti, inokodzera inozivikanwa 3D modhi kumufananidzo wekamera kuti utore nzvimbo yayo.
Kamera uye positi calibration mumarobhoti, inokodzera inozivikanwa 3D modhi kumufananidzo wekamera kuti udzore chinzvimbo chayo Matimu anowanzo kuwana mhedzisiro iri nani kana achinge atsanangura emhando yepamusoro kumberi, chengetedza nzira yekukwira kwevanhu yemakesi emupendero, uye kuteedzera zvese zvakawanikwa zvechigadzirwa uye mutengo wekukanganisa nekufamba kwenguva.
Njodzi & Guardrails
Kodzero dzemifananidzo uye kubvumirwa kunogona kuve njodzi dzepamutemo kana provenance isina kujeka.
Kuita kwemuenzaniso kunogona kusiyanisa kupenya, huwandu hwevanhu, uye nharaunda.
Manyepo enhema anogona kusacherechedzwa kunze kwekunge zvikumbaridzo zvekuvimba zvikatariswa.
Implementation Roadmap
Tsanangura maitiro ekugamuchirwa echokwadi, kurangarira, uye mutengo wekukanganisa.
Tsanangura maitiro ekugamuchirwa echokwadi, kurangarira, uye mutengo wekukanganisa. Bata nhanho yega yega segedhi rehumbowo: kana maitiro asina kusangana, imbomira kuburitsa, vhara gaka, uye wobva wawedzera kushandiswa.
Edzai nedata rinoenderana nemamiriro chaiwo ekugadzira.
Edzai nedata rinoenderana nemamiriro chaiwo ekugadzira. Bata nhanho yega yega segedhi rehumbowo: kana maitiro asina kusangana, imbomira kuburitsa, vhara gaka, uye wobva wawedzera kushandiswa.
Wedzera ongororo yemunhu kune yakaderera-kusavimbika kana yakakwirira-inokanganisa kufanotaura.
Wedzera ongororo yemunhu kune yakaderera-kusavimbika kana yakakwirira-inokanganisa kufanotaura. Bata nhanho yega yega segedhi rehumbowo: kana maitiro asina kusangana, imbomira kuburitsa, vhara gaka, uye wobva wawedzera kushandiswa.
Tevera modhi kudonha uye simbisa mushure mekuchinja kwekamera kana dataset.
Tevera modhi kudonha uye simbisa mushure mekuchinja kwekamera kana dataset. Bata nhanho yega yega segedhi rehumbowo: kana maitiro asina kusangana, imbomira kuburitsa, vhara gaka, uye wobva wawedzera kushandiswa.