CING APTI BUNDLE WITH M EMORY ORY U SA GE P REDICT ION : WITH - - PowerPoint PPT Presentation

cing
SMART_READER_LITE
LIVE PREVIEW

CING APTI BUNDLE WITH M EMORY ORY U SA GE P REDICT ION : WITH - - PowerPoint PPT Presentation

A DAP TIVE R AY AY - BU LE T RACIN CING APTI BUNDLE WITH M EMORY ORY U SA GE P REDICT ION : WITH SAGE DICTIO E FFI FFICIENT G LOBAL AL I LLUMIN ION IN IN L ARGE GE S CENES MINATIO Yusuke ke To Toku kuyoshi (Squar are Enix Co., Ltd.) .)


slide-1
SLIDE 1

ADAP

APTI TIVE RAY AY-BU BUNDLE LE TRACIN CING WITH WITH MEMORY ORY USA SAGE GE PREDICT DICTIO ION:

EFFI

FFICIENT GLOBAL AL ILLUMIN MINATIO ION IN IN LARGE GE SCENES Yusuke ke To Toku kuyoshi Ta Taka kashi Se Seki kine Tiag ago d da Silva Tak akas ashi Kan anai (Squar are Enix Co., Ltd.) .) (Squar are Enix Co., Ltd.) .) (Square Enix Co.

  • ., Ltd., Univ. of
  • f Tok
  • kyo)

(Univ. of

  • f Tok
  • kyo)
slide-2
SLIDE 2

LIGHT

IGHT MAP APS FO FOR LARG ARGE SCENES ES

45.3 M texel light maps Scene with 4.9 km in diameter (3.7 M triangles) Computation time: 1396 secs (2000 sample directions) (GPU: NVIDIA GeForce GTX 580 1.5GB memory)

slide-3
SLIDE 3

INTRODU

DUCTIO ION

  • 1. Intr

Introduc uction

  • 2. Ada

daptive T Til iling for R

  • r Ray

ay-bu bundl dles

  • 3. Expe

perimental Results & Future Work

  • rk
slide-4
SLIDE 4

RAY

AY-BUNDL DLE TRACI RACING

 Set of parallel rays for a sample direction [Sbert96]  Implemented with GPU rasterization [Szirmay-Kalos98, Hachisuka05]  Benefits: HW acceleration, tessellation etc.  Multi-fragment problem is identical to OIT  Per

Per-pixel li l linke ked-lis ist [Yang10]

slide-5
SLIDE 5

LIM

IMIT ITED MEMORY RY CAPACI ACITY OF OF GPU

GPUS

 Uniformly distributed rays  Inhomogeneous light map

density

 High-resolution ray-bundle buffer

is required

Me Memory ov

  • verflow

w of

  • f the li

lists Light lea eaking er error

 Memory usage is unknown

before rendering

 Excessive memory has to be

allocated

Ray ay-bu bundl dle trac racing g is weak ak in lar arge ge scenes

slide-6
SLIDE 6

UNIFO

FORM TIL ILIN ING [THIBIEROZ11]

 Proposed for real-time linked-list OIT  Split a render target into smaller tiled regions  Each tile is rendered separately  Unsuitable for off-line rendering  Overflow is still unpredictable  Scene-dependent parameter tuning

split plit 1 rende der r targe get 8x8 8 rende der r targe gets

slide-7
SLIDE 7

OUR

UR CONT NTRIBUTIONS ONS

 Memory usage prediction for linked-list ray-bundles  Adaptive tile subdivision using the above prediction  Reduce the risk of memory overflow & light leaking error  Avoid over-splitting  Less parameter tuning

Uniform rm tili ling Our adapt ptive tili ling

slide-8
SLIDE 8

ADAPTI

PTIVE TIL ILIN ING FO FOR RAY AY-BUNDLES LES

  • 1. Intr

Introduc uction

  • 2. Ada

daptive T Til iling for R

  • r Ray

ay-bu bundl dles

  • 3. Expe

perimental Results & Future Work

  • rk
slide-9
SLIDE 9

ADAPTI

PTIVE TIL ILIN ING

 Based on adaptive shadow mapping [Fernando01]  Quadtree-based tile subdivision  According to a low-resolution scene analysis  Analysis for memory usage prediction is also added  The overflow risk is reduced dramatically

 It is not completely eliminated, however

Our contribu bution ions

slide-10
SLIDE 10

IMPORTAN

ANCE CE & F

& FRAG

RAGMENT COUNT NT ANALY LYSIS

 Render two mipmaps from the ray-bundle direction  Pixels as quadtree nodes (resolution: 2n)

render

Importance mipmap Fragment count mipmap (re required ray ray de density) (memory usa sage pe per r ray ray)

slide-11
SLIDE 11

RECU

CURS RSIVE TIL ILE SUBDIV DIVIS ISION

 Start from the top mip level (root of the quadtree)  A tile is subdivided when overflow is predicted

Requ quire ired r d ray-bu bundl dle pix ixel l count Estima imated u d upper r bound

computed with importance mipmap computed with fragm gment nt c count unt m mipm pmap

Subdi bdivision c con

  • ndi

dition

for each tile

slide-12
SLIDE 12

EXPER

ERIMEN ENTAL RES ESULT ULTS

& & CONCLUSI

SIONS ONS

  • 1. Intr

Introduc uction

  • 2. Ada

daptive T Til iling for R

  • r Ray

ay-bu bundl dles

  • 3. Expe

perimental Results & Future Work

  • rk
slide-13
SLIDE 13

NO TIL

ILIN ING

100 100 secs MSE: SE: 2. 2.04 0435e-2

  • v
  • verflow ratio: 0%

GPU: NVIDIA GeForce GTX 580 with 1.5GB memory

2000 sample directions Ray-bundle resolution: 10242 Node buffer size: 5M nodes Analysis resolution: 10242

Grou Ground tru ruth

slide-14
SLIDE 14

35 35X35 U 35 UNIFO

FORM TIL ILIN ING

1381 1381 secs MSE: SE: 3. 3.33 3344e-3

  • v
  • verflow ratio: 6.96%

GPU: NVIDIA GeForce GTX 580 with 1.5GB memory

2000 sample directions Ray-bundle resolution: 10242 Node buffer size: 5M nodes Analysis resolution: 10242

Grou Ground tru ruth

slide-15
SLIDE 15

OUR

UR ADAPTI PTIVE TIL ILIN ING

(172.7 tiles / direction)

1396 1396 secs MSE: SE: 2. 2.53 5349e-4

  • verflow

w ratio: 1.27e 7e-2%

2000 sample directions Ray-bundle resolution: 10242 Node buffer size: 5M nodes Analysis resolution: 10242

GPU: NVIDIA GeForce GTX 580 with 1.5GB memory

Grou Ground tru ruth

slide-16
SLIDE 16

COMPU

MPUTA TATI TION TIM IMES PER ER SAMPLE LE DIRE RECT CTION

Analysi sis Ren ender ering

7.9 13.3 2.4 7.5 12.5

Mipm pmappi pping

0.3 0.5 0.3 0.2 0.3

Tile S Subdivisi sion

0.3 0.2 0.3 0.4 0.4

GPU GPU-CPU D Data Copy Copy

0.7 0.8 0.6 0.8 0.7

Ray ay-bu bundl dle Cre Creation

291.6 405.7 69.9 269.8 418.9

Light M Map Update te

180.3 274.7 63.8 217.4 286.9 (ms)

2% o

  • verhead

head

GPU: NVIDIA GeForce GTX 580 with 1.5GB memory

slide-17
SLIDE 17

CONCLUSI

SIONS ONS

 Adaptive tiling for linked-list ray-bundles  A tiles is subdivided when overflow is predicted  The risk of memory overflow is reduced dramatically  Less parameter tuning  Memory usage prediction  Using the fragment count mipmap  Demonstrated baking light maps of large scenes  With a limited memory capacity

slide-18
SLIDE 18

FUT

UTUR URE WORK RK

 Improving the analysis accuracy  Supersampling  Conservative rasterization [Hasselgren05]  Ray-bundle warping  Rectilinear texture warping [Rosen12]  Real-time linked-list OIT  For an arbitrary node buffer size

Warpi rping

slide-19
SLIDE 19

THA

HANK YOU OU