Intel ayaa daabacday siideynta mashruuca oidn 2.0 (Open Image Denoise), kaas oo soo saara ururinta filtarrada si ay u diidaan sawirada loo diyaariyey iyadoo la adeegsanayo hababka samaynta rayiga. Fur Denoise Image Denoise waxaa loo soo saaray qayb ka mid ah mashruuca qalabaynta hal API-ga weyn ee loogu talagalay in lagu horumariyo agabka aragga software ee sayniska (SDVis (Software Defined Visualization)), oo ay ku jiraan maktabadda raadraaca Embree, habka sawir-qaadista sawir-qaadista ee GLuRay, OSPRay waxay qaybisay madal raadinta raajo. , iyo nidaamka rasterization software OpenSWR Koodhku wuxuu ku qoran yahay C++ waxaana lagu daabacay shatiga Apache 2.0.
Hadafka mashruucu waa in la bixiyo sifooyin diidmo oo tayo sare leh, hufan, oo si sahlan loo isticmaali karo kuwaas oo lagu dabaqi karo si loo hagaajiyo tayada natiijooyinka raadinta raajada. Shaandhooyinka la soo jeediyay ayaa oggolaanaya, ku salaysan natiijada wareegga raadraaca raajada oo la gaabiyay, in la helo heerka ugu dambeeya ee tayada oo la barbardhigi karo natiijada habka kharash badan iyo waqti-qaadashada ee bixinta faahfaahsan.
Furitaanka Sawirka Denoise wuxuu shaandheeyaa qaylada aan sugnayn sida Monte Carlo Nambarada Is-dhexgalka (MCRT) raadinta raajo. Si loo gaaro wax-soo-saarka tayada sare leh ee algorithms-yadan oo kale, raadinta tiro aad u badan oo fallaadho ah ayaa loo baahan yahay, haddii kale farshaxannada la arki karo ee qaabka qaylada random ayaa ka muuqda sawirka natiijada.
Isticmaalka Denoise Image Furan wuxuu kuu ogolaanayaa inaad hoos u dhigto tirada xisaabinta lagama maarmaanka ah dhowr amar oo waaweyn markaad xisaabinayso pixel kasta. Natiijo ahaan, waxaa suurtogal ah in la soo saaro sawirka hore ee buuqa badan si dhakhso ah, laakiin ka dibna u keen tayada la aqbali karo iyadoo la isticmaalayo algorithms yaraynta dhawaaqa degdega ah. Qalabka ku habboon, qalabka la soo jeediyay ayaa xitaa loo isticmaali karaa raadinta shucaaca is-dhexgalka leh ee ciribtirka dhawaaqa duulista.
Maktabadu waxa loo isticmaali karaa noocyo kala duwan oo qalab ah, laga bilaabo laptops iyo kombuyuutar ilaa noodhadhka kooxaha. Hirgelinta waxaa loo habeeyay fasalo kala duwan oo ah 64-bit Intel CPUs oo taageero u ah SSE4, AVX2, AVX-512 iyo XMX (Xe Matrix Extensions) tilmaamaha, Apple Silicon chips iyo nidaamyada leh Intel Xe GPUs (Arc, Flex iyo Max taxane), NVIDIA (ku salaysan Volta, Turing, Ampere, Ada Lovelace iyo Hopper architectures) iyo AMD (ku salaysan RDNA2 (Navi 21) iyo RDNA3 (Navi 3x) naqshadaha). Taageerada SSE4.1 waxaa lagu dhawaaqay inay tahay shuruudaha ugu yar.
Isbeddellada muhiimka ah ee siidaynta Muuqaalka Furan ee Denoise 2.0:
- Taageerada dardargelinta hawlaha dhimista buuqa iyadoo la isticmaalayo GPU-da. Taageerada la hirgeliyay ee GPU-da-dejinta oo leh SYCL, CUDA, iyo HIP nidaamyada kuwaas oo loo isticmaali karo GPU-yada ku salaysan Intel Xe architecture, AMD RDNA2, AMD RDNA3, NVIDIA Volta, NVIDIA Turing, NVIDIA Ampere, NVIDIA Ada Lovelace, iyo NVIDIA Hopper.
- Maaraynta kaydinta cusub ee API ayaa lagu daray, taasoo kuu ogolaanaysa inaad doorato nooca kaydinta, koobi ka samayso xogta martida loo yahay, oo aad ka soo dejiso kaydka dibadeed ee garaafyada API-yada sida Vulkan iyo Direct3D 12.
- Taageero lagu daray qaabka fulinta asynchronous (shaqooyinka oidnExecuteFilterAsync iyo oidnSyncDevice).
- Waxaa lagu daray API si loogu diro codsiyada aaladaha jireed ee ku jira nidaamka.
- Shaqada oidnNewDeviceByID oo lagu daray si loo abuuro qalab cusub oo ku salaysan aqoonsiga aaladda, sida UUID ama cinwaanka PCI.
- Sifooyinka lagu daray ee la qaadi karo ee SYCL, CUDA iyo HIP.
- Waxaa lagu soo daray ikhtiyaarada iskaanka aaladaha cusub (SystemMemorySupported,MemorySupported la maamulay,MemoryTypes dibadda ah).
- Lagu daray halbeeg si loo dejiyo heerka tayada filtarrada.
Source: opennet.ru