ringing pattern generated by path tracing about pbrt-v2 HOT 3 CLOSED

My guess would be that you're running out of floating point precision--a
32-bit float may not have quite enough precision to accumulate the sum of
200k samples per pixel.

One way to test this would be to:

1. Modify the Pixel struct in film/image.h to use 'double' instead of
   'float' for all members.
2. Remove the AtomicAdd() stuff in film/image.cc and just update the Pixel
   fields directly in AddSample() and Splat().
3. Render the scene using one thread (to be safe), with --ncores 1 on the
   command-line. (Or don't worry about this--it's probably fine.)

And see if you still see these errors. If not, it's floating-point
precision. If you do, then it's something else. (And we can work on it
from there...)

Thanks,
Matt

On Tue, Apr 22, 2014 at 1:58 PM, HoverDong [email protected] wrote:

hi guys,

When I use path tracing to generate a very simple scene image, I got some
ringing patterns. The scene is just a Cornell box with back, floor and left
wall and a point light source. The material of all the walls are diffuse
material. I used 200k samples/pixel to generate the scene. I set the
maxDepth=0, which means the generated image only shows the direct light.
When I use less samples number, the ringing pattern becomes less obvious.
The attached image showed the R channel of the generated image. Any ideas?
[image: point_gt_d]https://cloud.githubusercontent.com/assets/7377509/2770612/b31da966-ca60-11e3-93d0-6b6d0abc007a.jpg

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/29
.

from pbrt-v2.

hey Matt, Thanks for the quick reply. I really appreciate it!!! I changed the code as you suggested and tried with 200k sample numbers again. The ringing pattern has gone:) So, it is the precision problem as you said.

Hover

Date: Tue, 22 Apr 2014 19:32:02 -0700
From: [email protected]
To: [email protected]
CC: [email protected]
Subject: Re: [pbrt-v2] ringing pattern generated by path tracing (#29)

My guess would be that you're running out of floating point precision--a

32-bit float may not have quite enough precision to accumulate the sum of

200k samples per pixel.

One way to test this would be to:

1. Modify the Pixel struct in film/image.h to use 'double' instead of

'float' for all members.

1. Remove the AtomicAdd() stuff in film/image.cc and just update the Pixel

fields directly in AddSample() and Splat().

1. Render the scene using one thread (to be safe), with --ncores 1 on the

command-line. (Or don't worry about this--it's probably fine.)

And see if you still see these errors. If not, it's floating-point

precision. If you do, then it's something else. (And we can work on it

from there...)

Thanks,

Matt

On Tue, Apr 22, 2014 at 1:58 PM, HoverDong [email protected] wrote:

hi guys,

When I use path tracing to generate a very simple scene image, I got some

ringing patterns. The scene is just a Cornell box with back, floor and left

wall and a point light source. The material of all the walls are diffuse

material. I used 200k samples/pixel to generate the scene. I set the

maxDepth=0, which means the generated image only shows the direct light.

When I use less samples number, the ringing pattern becomes less obvious.

The attached image showed the R channel of the generated image. Any ideas?

[image: point_gt_d]https://cloud.githubusercontent.com/assets/7377509/2770612/b31da966-ca60-11e3-93d0-6b6d0abc007a.jpg

—

Reply to this email directly or view it on GitHubhttps://github.com//issues/29

.

—
Reply to this email directly or view it on GitHub.

from pbrt-v2.

Closing this out. (And note that pbrt-v3 offers the ability to select double-precision for all floating-point computations at compile time.)

from pbrt-v2.