Marcin Romaszewicz (marcin++at++asmodean.engr.sgi.com)
Thu, 28 Oct 1999 13:34:15 -0700 (PDT)
Performer does indeed use the stencil planes to calculate the number of
times each pixel is drawn. In fill-stat gathering mode, performer
increments the stencil value by 1 each time a pixel is drawn. If you do
not have enough stencil planes, the draw count will be clamped at the
maximum representable unsigned value (15 for 4 bit, 255 for 8 bit, etc).
So, when the depth complexity is high and counts are clamped, the reported
depth complexity will be less than the actual depth complexity.
The visual which gives you the correct depth complexity is the one with 8
stencil planes (assuming DC < 256 for most pixels).
-- Marcin
On Thu, 28 Oct 1999, Rudi Schmidt wrote:
> Hi pfFolks!
>
> I'm having trouble understanding some phenomenon in Performer's
> pixelfill-statistics.
> I am running a driving-database (town) in perfly on an Onyx2-IR,
> single-channel
> For specific positions in the database, I need to measure the number of
> pixel-draws, or -equivalently- the average depth complexity (=DC) of the
> scene.
> My phenomenon now is, that I get different values for DC depending on
> the visual chosen -at exactly the same database position !
> The visuals I am testing are all double-buffered, 10bit RGB, 8-fold
> multisampled and Z-buffered. The only difference is the number of
> stencil-planes, and the depth of the z-buffer.
> To give an example of the degree of deviations: the DC-value might be
> of value 2.5 with one visual, and DC=2.85 (at the same position) with
> another visual.
> I am certainly aware, that Performer could use stencil-planes for
> accumulating the number of pixel-visits, but the difference I am seeing
> must have some reason, or at least an explanation !
>
> Now my questions:
> 1. CAN SOMEBODY MAKE SENSE OUT OF THAT, AND EXPLAIN WHAT IS GOING ON
> THERE ?
> 2. WHICH VISUAL GIVES ME THE CORRECT, ACTUAL DC-VALUE ? IS IT THE ONE
> WITH 8 BITS STENCIL (which is my suspision ! )?
>
> -- Rudi
> --------------------
> Rudolf Schmidt
> Geltendorf (Munich)
> mailto:Rudolf++at++Schmidt.net (home)
> mailto:Rudolf.Schmidt++at++KMWeg.de (office)
> -----------------------------
>
>
> -----------------------------------------------------------------------
> List Archives, FAQ, FTP: http://www.sgi.com/software/performer/
> Submissions: info-performer++at++sgi.com
> Admin. requests: info-performer-request++at++sgi.com
>
This archive was generated by hypermail 2.0b2 on Thu Oct 28 1999 - 13:34:32 PDT