Denoising Deep Monte Carlo Renderings

Delio Vicini David Adler Jan Novák Fabrice Rousselle Brent Burley
In Computer Graphics Forum

A pro­duc­tion scene (a) rendered as a deep noisy im­age (b), de­noised us­ing our meth­od (c), and com­pos­ited (d) with a deep volume. Our al­gorithm re­moves a sig­ni­fic­ant amount of the in­put Monte Carlo noise, while pre­serving the depth sep­ar­a­tion of the deep im­age. (c) Dis­ney


We present a nov­el al­gorithm to de­noise deep Monte Carlo ren­der­ings, in which pixels con­tain mul­tiple col­our val­ues, each for a dif­fer­ent range of depths. Deep im­ages are a more ex­press­ive rep­res­ent­a­tion of the scene than con­ven­tion­al flat im­ages. However, since each depth bin re­ceives only a frac­tion of the flat pixel's samples, de­nois­ing the bins is harder due to the less ac­cur­ate mean and vari­ance es­tim­ates. Fur­ther­more, deep im­ages lack a reg­u­lar struc­ture in depth—the num­ber of depth bins and their depth ranges vary across pixels. This pre­vents a straight­for­ward ap­plic­a­tion of patch-based dis­tance met­rics fre­quently used to im­prove the ro­bust­ness of ex­ist­ing de­nois­ing fil­ters. We ad­dress these con­straints by com­bin­ing a flat im­age-space non-loc­al means fil­ter op­er­at­ing on pixel col­ours with a deep cross-bi­lat­er­al fil­ter op­er­at­ing on aux­il­i­ary fea­tures (al­bedo, nor­mal, etc.). Our ap­proach sig­ni­fic­antly re­duces noise in deep im­ages while pre­serving their struc­ture. To our best know­ledge, our al­gorithm is the first to en­able ef­fi­cient deep-com­pos­it­ing work­flows with de­noised Monte Carlo ren­der­ings. We demon­strate the per­form­ance of our fil­ter on a range of scenes high­light­ing the chal­lenges and ad­vant­ages of de­nois­ing deep im­ages.


Text citation

Delio Vicini, David Adler, Jan Novák, Fabrice Rousselle, and Brent Burley. 2019. Denoising Deep Monte Carlo Renderings. In Computer Graphics Forum 38(1). 316-327.

    author = {Delio Vicini and David Adler and Jan Novák and Fabrice Rousselle and Brent Burley},
    title = {Denoising Deep Monte Carlo Renderings},
    journal = {Computer Graphics Forum},
    volume = {38},
    number = {1},
    pages = {316-327},
    year = {2019},
    month = feb,
    doi = {10.1111/cgf.13533}