Re: [Pipmak-Users] Patch rotation proposal

[Christian W. <cwa...@gm...>]

Andrea Viarengoa wrote:
> I'm a little confused about set anchorx=anchory=0.5 to make a patch
> to coincide with a cube face, can you explain better?

This is related to the following two facts:

- The cube face images don't extend exactly to the cube edge but half a 
pixel beyond. This is explained under "Making Cubic Panoramas" in 
section 2.8 of the reference.

- For a similar reason (seamless bilinear interpolation), a margin of 
half a pixel is cut off each edge of a patch, so that the visible part 
of the patch ends at the centers of the outermost pixels. This is hinted 
at (though not properly explained) in the last paragraph of section 2.9 
"Patches".

By setting the anchor to (0.5, 0.5), i.e. the center of the corner 
pixel, we place that pixel at the same position (centered on the cube 
corner) as the corner pixel of the cube face.

> When you speak about putting anchorx,anchory at (x,y),
> I understand that you refer to the relative coordinates of the image,
> is it correct?

Exactly.

> If I use the notation (nx,ny,nz), I could set an anchor that doesn't 
> lie on the same plane on which the patch image lie.

I don't understand what you mean by that. The anchor by definition 
always lies in the plane of the patch. It need not lie in the plane of a 
cube face, but that's what we want.

> About the performance, the behaviour is strange, because I have notice 
> that also with a traditional pipmak project (like MyHouse Tour), 
> sometime on my notebook the rotation along the azimuth is jerk, sometime no.

I sometimes see this behavior as well, but so far I never bothered to 
find out what causes it and if it can be fixed in Pipmak.

> Another strange things: if I put the pipmak window in the background, 
> pipmak eat over 80% of cpu time and when I put in the foreground,
> the cpu amount rise to 5%, if I leave pipmak in the background for a lot
> of time without do nothing, I have to kill because take 100% of cpu time.
> Do you think this behaviour depend by my graphic card? 

Just to get the language straight - does the CPU usage rise *by* 5% (to 
85%) or *fall* to 5%?

Pipmak is supposed to render (at most) one frame per screen refresh 
(i.e. 60 frames per second on usual LCDs). If that limitation works, it 
should not use excessive amounts of CPU time. If for some reason it 
doesn't, Pipmak renders as fast as it can and takes all available CPU 
time. Could be that the limitation works when the Pipmak window is in 
the foreground and OpenGL can render directly to the frame buffer, but 
not when the window is partially obscured and needs to be composited 
with the overlapping windows. On what platform is this?

  -Christian