"Bald Eagle" <cre### [at] netscapenet> wrote:

> Next post will fix that with fabs(r0)

As you can see, there is a lot more red in this render, showing the removal of
erroneous bounding intersection positives, and the rays fail on both the left
and the right now, with only the rays that actually intersect the bounding
cylinder triggering the root-solving phase.

Post a reply to this message

Attachments:
Download 'sor_diagram1.png' (522 KB)

Preview of image 'sor_diagram1.png'

And here is a much simplified diagram showing only 3 rays, with all extraneous
lines removed.

Only the rays and the perpendicular distances are shown, color coded for
bounding intersection test state.

Post a reply to this message

Attachments:
Download 'sor_diagram1.png' (134 KB)

Preview of image 'sor_diagram1.png'

Desmos live graphing

Move the ray origin or the direction vector endpoint.
Very useful for working things out.

https://www.desmos.com/calculator/o50ybsmkw3

- BE

Post a reply to this message

On 10/9/25 22:08, Bald Eagle wrote:
> And here is a much simplified diagram showing only 3 rays, with all extraneous
> lines removed.
> 
> Only the rays and the perpendicular distances are shown, color coded for
> bounding intersection test state.
> 

I'm loosely following your sor updates - good work. I'll try and carve 
out some time to code up your suggested changes in my working yuqk code 
& do some testing.

Bill P.

Post a reply to this message

William F Pokorny <ano### [at] anonymousorg> wrote:

> I'm loosely following your sor updates - good work. I'll try and carve
> out some time to code up your suggested changes in my working yuqk code
> & do some testing.

Thank you, Sir.
I'm hoping that at least a little bit of improvement will come of all of this.

Interested in your thoughts about the possibility of breaking the sor {} up into
individual spline segments for rendering to optimize the cylindrical bounding.

- BW

Post a reply to this message

2025 paper, if someone can access the full article.

https://www.researchgate.net/publication/3904359_Ray_tracing_surfaces_of_revolution_An_old_problem_with_a_new_perspecti
ve

Ray tracing surfaces of revolution: An old problem with a new perspective
February 2001Proceedings of Computer Graphics International Conference, CGI

Abstract
We present a new subdivision scheme that is shown to improve the performance of
ray tracing surfaces of revolution over Kajiya's (1983) classical work. This is
based on a monotonic interval partitioning of a generatrix of a surface of
revolution. The algorithm has a search complexity upper bound of O(log(m n)) for
m monotonic intervals and n subdivisions for each interval and runs up to three
times faster on large scenes. This method also suggests a novel hybrid bounding
volume scheme that reduces this number of intersection tests between a ray and
the actual object surface

Post a reply to this message

"Bald Eagle" <cre### [at] netscapenet> wrote:
> 2025 paper, if someone can access the full article.
>
>
https://www.researchgate.net/publication/3904359_Ray_tracing_surfaces_of_revolution_An_old_problem_with_a_new_perspec
ti
> ve

looks interesting. do you need it or do you already have it? I can mail it.

ingo

Post a reply to this message

"ingo" <nomail@nomail> wrote:

> looks interesting. do you need it or do you already have it? I can mail it.
>
> ingo

Thanks Ingo.
Let me check and see if I was able to get that one.

I'll look on my other computer at work tomorrow.
I was in the midst of a flurry of activity and can't recall if I managed to get
that one.

The original sor {} paper(s) for our source code would be very nice to have.  ;)

- BW

Post a reply to this message

On 10/10/25 16:03, William F Pokorny wrote:
> I'm loosely following your sor updates - good work. I'll try and carve 
> out some time to code up your suggested changes in my working yuqk code 
> & do some testing.

OK. Looking specifically at these two suggestions:

1. replace
D = ray.Direction;
with
D = <ray.Direction.x, 0, ray.Direction.z>;

and

2. replace
#if (r0 > Radius2)
with
#if (fabs(r0) > Radius2)

----

1) I cannot find 'D = ray.Direction;' in my yuqk code? Where and in what 
version of source code are you seeing this?

2) Your fabs() suggestion looks like a win and I plan to adopt it for 
release 20 of yuqk.

---

The benefits of (2) depend on:

a) how much void space there is between that outer cylinder and all the 
inner per segment ones.

b) the alignment of rays with the bounding box used for global bounding. 
I looked too at completely eliminating the initial cylinder bound test 
as we already do cylinder bounding for each segment / interval.

Summarizing the performance changes for a bunch of testing.

------ Worst case bounding box alignment (cyl bound helps most)

56.03 -> 54.79s ---> -2.21% Old to New

55.70 -> 54.79s ---> -1.63% No_InitialCylTest to New

------ Best case bounding box alignment (cyl bound helps least)

43.99 -> 43.89s ---> -0.23%   Old to New

44.47 -> 43.89s ---> -1.30%   No_InitialCylTest to New

------ Turning global bounding completely off in yuqk with '-b'

100.72 -> 98.01s ---> -2.69%  Old to New

  98.44 -> 98.01s ---> -0.44%  No_InitialCylTest to New (*)

(*) Where every ray sees the initial cylinder test (New) vs not, the 
performance is less different due the expensive sqrt() starting to eat 
the better in bounding benefit. The change is still a win because we're 
not running without global bounding except in rare cases.

Bill P.

Post a reply to this message

William F Pokorny <ano### [at] anonymousorg> wrote:

> 1) I cannot find 'D = ray.Direction;' in my yuqk code? Where and in what
> version of source code are you seeing this?

Right because that line is not explicitly there.
D is implicitly defined as the ray direction.

Then in line 256 we do:    len = D.length();

Which is why we need to get rid of the y component for reasons explained.

So really what we need is an extra line that REdefines the _projected_ ray
direction as D = <D.x, 0, D.y> before that length and normalization take place.

- BW

Post a reply to this message