Calculating the optimum resolution

Intermediate

Tuesday, July 6th 2010, 5:59pm

Calculating the optimum resolution

Hi there

I want to calculate the best resolutions for my photos. For this, it would be very useful to know when does krpano change from one resolution level to another. Has Klaus or someone else got a formula for this? I made my own attempts at calculating it, but failed miserably :P

Cheers
Manuel

PanoTag: Facebook tagging for krpano

KRPano development and more: http://www.digisfera.pt/en/development

Go to the top of the page

emediquei

Intermediate

Posts: 192

Saturday, July 10th 2010, 12:53pm

So, I got to this formula for multiresthreshold=0:

Fh = 2*arctan(w/r)

Fh - horizontal FOV
w - stage width (size of the embedded flash or, when in fullscreen, the screen resolution)
r - resolution level

After some tests, I can conclude that this formula isn't right :P However, it does fail "predictably". That is, the offset to the real value only seems to depends on the final Fh. So, I just did a function fitting and got values which match very well the tests I made.

The attached table shows the minimum horizontal fov at which krpano will keep a certain resolution level. This depends on the stage resolution, which is displayed on the left. The table contains values for the most commonly used screen resolutions (usage data from a website I have and the first couple of hits on google for "most common screen resolutions" ;).
Using this table, you can predict which resolution level will be seen, and hopefully it will help plan the multiresolution levels better :) I'm also attaching an Excel sheet, where you can add your own resolutions and levels.

By the way, the offset I got before compensating with the fitted function was this. Any thoughts on what this function may be? X is the predicted FOV and Y is the offset to the measured FOV.

I guess Klaus has more info on how multires levels are selected, as well as on how exactly the multiresthreshold parameter affects it. Could we have some insight on this? ;)
Also, is the manual setting of resolution levels in kmakemultires (like mentioned here ) planned for any time soon? It would be extremely helpful!

Cheers,
Manuel

emediquei has attached the following file:

tableXLS.zip (12.46 kB - 70 times downloaded - latest: Yesterday, 5:46pm)

PanoTag: Facebook tagging for krpano

KRPano development and more: http://www.digisfera.pt/en/development

Go to the top of the page

klaus.krpano

Administrator

Posts: 13,499

Tuesday, July 13th 2010, 3:17pm

Hi,

Quoted

I guess Klaus has more info on how multires levels are selected, as well as on how exactly the multiresthreshold parameter affects it. Could we have some insight on this? ;)

yes

I'm including the screen size/aspect and a correction for wide angles (to avoid using a too low resolution in that case),

this is the formula I'm using: (based on the vertical fov)

	Source code
1	optimal spherical vertical resolution = (1.0 + multiresthreshold) * 1/4 * PI * ((screenwidth / sin( atan( tan(vfov/2) * screenwidth/screenheight) )) + (screenheight / tan(vfov/2)))

to get the cubical resolution just:

	Source code
1	optimal cube pano resolution = spherical vertical resolution * 2 / PI

and then choosing the next level with a resolution higher than that,

Quoted

Also, is the manual setting of resolution levels in kmakemultires (like mentioned here ) planned for any time soon? It would be extremely helpful!

I will not forget it

best regards,
Klaus