POV-Ray : Newsgroups : povray.pov4.discussion.general : Not a v3.8 issue, but a patch is needed for jitter with rtr/kla. Server Time
18 Oct 2021 15:09:05 EDT (-0400)
  Not a v3.8 issue, but a patch is needed for jitter with rtr/kla. (Message 1 to 3 of 3)  
From: William F Pokorny
Subject: Not a v3.8 issue, but a patch is needed for jitter with rtr/kla.
Date: 30 Sep 2021 13:52:25
Message: <6155f959$1@news.povray.org>
Somewhere want to document for v4.0.

It looks as if it's long been the case the jitter value forever shrinks 
as the continuous (-kla) real time rendering (rtr) option runs. The 
user's jitter value needs to be captured aside and restored after each 
'frame' in the continuous render.

I checked and frames rendered by the more usual animation mechanisms are 
OK with respect to the jitter value.

Bill P.


Post a reply to this message

From: Kenneth
Subject: Re: Not a v3.8 issue, but a patch is needed for jitter with rtr/kla.
Date: 1 Oct 2021 00:55:00
Message: <web.61569473d5273f3d4cef624e6e066e29@news.povray.org>
Are you referring to the jitter when using v3.8xx's newer antialiasing method 3
and Stochastic_Seed? Or in a more general AA sense?

I've never noticed the effect you mentioned when running kla/rtr animation in
Windows versions-- but then again, I rarely use any type of AA for such
'real-time rendering' tests (unless I'm doing a specific test that really needs
AA).

So, from your description, it sounds like the jitter is being constantly reduced
(from a given value) *as* the animation runs(?) I'll have to take a closer look.
Thanks for the heads-up.


Post a reply to this message

From: William F Pokorny
Subject: Re: Not a v3.8 issue, but a patch is needed for jitter with rtr/kla.
Date: 1 Oct 2021 06:10:16
Message: <6156de88$1@news.povray.org>
On 10/1/21 12:54 AM, Kenneth wrote:
> Are you referring to the jitter when using v3.8xx's newer antialiasing method 3
> and Stochastic_Seed? Or in a more general AA sense?

Thanks for the question, I wasn't clear.

Only AA methods 1 & 2 with jitter(b) are affected. The new v3.8 method 
3(a) AA is immune to the AA jitter amount issue for certain, but I also 
admit I didn't try that AA method with rtr/kla.

This jitter shrinking issue matters more to the povr branch due both 
having truly random jitter and in supporting big jitter (>1 values) both 
to compensate for another kind of jitter shrink on increasing recursion 
depth and for effects. See "Correcting comments on Anti-alias (AA) modes 
and jitter" post to povray.general from October of last year for more 
detail.

Bill P.

(a) - I perhaps missed a fix for the method 3 sqrt of slightly negative 
color channel values issue, but to my recollection, that's not fixed in 
v3.8 beta 2.

(b) - Even without jitter AA methods 1&2 take additional samples - why I 
too had missed that jitter was shrinking to nothing as rtr/kla ran. 
Well, and that until I started to play with HDRIs environment maps where 
I could run rtr/kla at larger render sizes and good fps rates my 
continuous animation window sizes were smallish.


Post a reply to this message

Copyright 2003-2021 Persistence of Vision Raytracer Pty. Ltd.