Ntuziaka Visual AI

Ntugharị dị iche

Nsụgharị dị iche iche na-eme ka usoro ịtụgharị ihe nkiri 3D ka ọ bụrụ onyonyo 2D dị iche iche, yabụ ị nwere ike gbakọọ gradients site na pikselụ atụgharịa azụ na mpaghara ọnọdụ.

2 nkeji na-agụEmelitere ikpeazụ

Nchịkọta

This lets you optimize geometry, materials, lighting, and camera using gradient descent.

Ime miri emi

Nsụgharị ọdịnala bụ okporo ụzọ nwere otu ụzọ: nri na geometry, ihe, ọkụ, na igwefoto, na pikselụ pụta. Ntugharị dị iche iche na-atụgharị na-aga site na ịgbakọ ka pikselụ mmepụta ọ bụla si agbanwe n'ihe gbasara oke ntinye ọ bụla. Site na gradients ndị ahụ, onye na-ebuli elu nwere ike ịhazigharị ọdịdị 3D ma ọ bụ textures ya ruo mgbe onyonyo egosipụtara dabara na foto ebumnuche, nke bụ obi ntụgharị ntụgharị na nyocha site na njikọ. Isi ihe siri ike bụ na nsụgharị na-agụnye nkwụsịtụ, karịsịa na silhouettes ihe na n'akụkụ oghere, ebe pikselụ na-amali na mberede site n'ihu ruo n'azụ. Ụzọ dị ka nro rasterization (SoftRas), sample-sampling (Li et al.'s redner), na rasterizer na PyTorch3D na-eji ire ụtọ ma ọ bụ ihe jikọrọ oke pụrụ iche. Ọzụzụ NeRF na 3D Gaussian splatting bụ ngwa ewu ewu.

Nghọta nka nka

Isi ihe ịma aka bụ nkwụsị visibiliti. N'elu silhouette ihe, pikselụ na-esi n'ihu gaa n'azụ, yabụ, usoro enweghị isi bụ efu ihe fọrọ nke nta ka ọ bụrụ ebe niile na enweghị nkọwa ya na nsọtụ, na-enye enweghị gradient bara uru gbasara udi. Ngwọta na-eme ka mkpuchi dị nro ka triangles na-atụnye aka n'ụkwụ na-adịghị mma na pikselụ dị nso (rasterization dị nro) ma ọ bụ mee nlele n'ụzọ doro anya n'akụkụ ọnụ iji gbakọọ oge njedebe nke ihe nrụnye (nlekọta ọnụ).

Mmetụta atụmatụ

Ọsọ na ọnụ ọgụgụ

Visual AI nwere ike megharịa nyocha, nchọpụta na mkpado ọrụ n'ọtụtụ.

Mee nhọrọ

Otu ndị na-emepụta ihe nwere ike imepụta echiche ngwa ngwa site na ngbanwe akwụkwọ ntuziaka ole na ole.

Team na usoro ọrụ

Ọrụ nwere ike iji onyonyo na akara vidiyo siri ike ịhazi.

Ọdịnihu nke ntụgharị dị iche iche

Nsụgharị dị iche iche na-aghọ anụ ahụ jikọtara n'etiti eserese na mmụta miri emi. Dị ka ezigbo oge dị iche iche renderers na Gaussian-splating pipeline tozuru okè, na-atụ anya tighter loops maka 3D nwughari site na foto, neural ihe onwunwe simulation, robotics simulation na mmụta physics, na ọgwụgwụ-na-ọgwụgwụ usoro ebe otu ọnwụ na-esi na ikpeazụ image ruo n'ókè na-egosi parameters. Ịchọ ụzọ dị iche iche maka ọkụ zuru ụwa ọnụ bụ oke nyocha na-arụ ọrụ na-aga n'ihu na ịdị irè.

Mmejuputa n'ezie n'ụwa

Ịmegharị udi na udidi ihe 3D site na foto ole na ole site n'ịkwalite ihe nlereanya ruo mgbe nsụgharị dabara na onyonyo (nsụgharị ntụgharị).

Ọzụzụ NeRFs na 3D Gaussian splats, ebe gradients sitere na echiche egosipụtara na-emelite ihe ngosi ihe nkiri ahụ.

Na-eme atụmatụ njirimara ihe (mkpụrụ siri ike, nlegharị anya) site na ijikọ ihe ndị egosipụtara na ezigbo foto.

Igwefoto na nhazi ọkwa na robotics, dabara ụdị 3D amaara na onyonyo igwefoto iji nwetaghachi ọnọdụ ya.

Ihe ize ndụ & okporo ụzọ nche

Ikike onyonyo na nkwenye nwere ike bụrụ ihe egwu dị n'iwu ma ọ bụrụ na edoghị anya.

Ọrụ nlereanya nwere ike ịdịgasị iche n'ofe ọkụ, igwe mmadụ, na gburugburu.

Enwere ike ghara ịhụ ihe dị mma ma ọ bụrụ na enyochaghị oke ntụkwasị obi.

Map mmejuputa

1

Kọwaa ụkpụrụ nnabata maka nkenke, icheta, na ụgwọ njehie.

2

Nwalee na data dabara na ọnọdụ mmepụta n'ezie.

3

Tinye nyocha mmadụ maka obere obi ike ma ọ bụ amụma mmetụta dị elu.

4

Sochie ihe nlere anya wee megharịa ka emechara mgbanwe igwefoto ma ọ bụ dataset.

Nọgide na-eme nchọpụta

Free newsletter

Get the daily AI briefing

Three verified AI stories every weekday morning, written in plain English. Free forever, no ads.

One email each weekday. Unsubscribe in one click. We never sell or share your address.

Test yourself

Take the Differentiable Rendering quiz

Instant feedback on every answer, and a shareable certificate with a verifiable ID once you pass a course.

Malite ajụjụ

Support free AI education. AI Understanding is a 501(c)(3) nonprofit — no ads, no paywall, ever. Make a donation

Ntuziaka na-esote

DDSP Synthesis ọdịyo dị iche

Ajụjụ a na-ajụkarị

What is Differentiable Rendering?

Nsụgharị dị iche iche na-eme ka usoro ịtụgharị ihe nkiri 3D ka ọ bụrụ onyonyo 2D dị iche iche, yabụ ị nwere ike gbakọọ gradients site na pikselụ atụgharịa azụ na mpaghara ọnọdụ. Nke a na-enye gị ohere ịkwalite geometry, ihe, ọkụ na igwefoto site na iji mgbada gradient.

Kedu ihe nsụgharị dị iche na-ahapụ gị ịgbakọ?

Site n'ime ka nsụgharị dị iche, ị na-enweta gradients na-esi na pikselụ laghachi azụ na ntinye ọnọdụ, na-eme ka nkwalite dabere na gradient.

Kedu ihe kpatara silhouettes ihe bụ isi ihe siri ike maka nsụgharị dị iche?

N'elu silhouette, pikselụ na-adaba n'etiti ihu na azụ, yabụ, usoro enweghị isi bụ efu ma ọ bụ enweghị nkọwa na enweghị ozi ọ bụla.

Kedu ka rasterization dị nro si edozi nsogbu nkwụsị?

Rasterization dị nro (dịka, SoftRas) na-eme ka mkpuchi dị nro ka triangle ọ bụla na-emetụta pikselụ gbara ya gburugburu, na-emepụta gradients ndị nwere ike iji.

Kedu ihe bụ 'inverse rendering'?

Nsụgharị ntụgharị na-eji nsụgharị dị iche iche na-ebuli paramita ọnọdụ ruo mgbe onyonyo arụpụtara dabara na foto a hụrụ.

Kedu usoro ọgbara ọhụrụ na-adabere na nsụgharị dị iche iche maka ọzụzụ?

Ma NeRFs na 3D Gaussian splatting na-eme ka ọ dịkwuo mma site n'ịgbasa ihe nhụsianya oghere-ohere site na onye nrụpụta dị iche.