Multiresolution settings questions

[timmy2]

I've watched both tutorial videos about the multiresolution feature but stumbled over what seemed like inconsistencies. The narrator opens what appears to be an 80K px equirectangular image but then starts referring to a square image that's just a segment of the whole equi. At 1:36 she says the square image is 21,600 px wide. But then at 3:36 she refers to the same image and says it's 10,800 px wide.

A graphic then breaks the square image into 16 smaller squares and she says they're 675 px wide. This gives me a spin because the squares are only 4 across, which seems like they'd be 2700 px wide to total 10,800 wide.

If anyone can explain what they were saying or trying to say I'd greatly appreciate it.

I then jumped into trying a project. My original TIFF equirectangular image is 20K px wide. After opening a Flash project I enabled multiresolution and let it choose the level sizes. It created three: 5000, 2500, 1250. I was surprised because I figured it would start with 20K or at least 10K. What determines these levels and should I add any more?

It's my understanding that every level should be evenly divisible by the Level Tile Size, but I'm still not sure what the optimum Level Tile Size would be. 1250? (my original file size divided by 16)

I set Window Size to 1280 x 768.

Since the tutorial said that Cube Face Size is irrelevant when producing Multiresolution panos I ignored dealing with it even though I was looking forward to using pi after all these years. What I do konw is that I want the starting image to be as good as it can be at the specified window size. I read somewhere that Hans Nyberg recommends a 1650 px cube but if Cube Face Size is irrelevant here all I'm left with as variables are the Level widths and the Level Tile Size, and I cannot figure out how to use them to impact my starting image. I realize load times should also be my concern so I would like to know how best to set those values.

The resulting pano looks pretty good but I felt like after all my forum reading and tutorial watching that I'd be more confident about the settings I chose. Anyone willing to share some guidance out of my confused state?

[Hopki]

Hi Timmy,
An Equirec image is converted into cubes, the tutorial shows a large input image of 86,400 x 43,200px.
When using Multi-Resolution the Cube face size in the settings tab is ignored.

The Multi-Resolution adds the levels, so 86,400 divided by 4 = 21600px per cube face for the first level.

The 10,800 px cube face is taken from the project and is being used as an example to show sub tiles, the narrator does say this at 3:40 "using the numbers from our project".
This shows sub tiles 675px X 16 = cube face 10,800px. The narrator could have said 675 X 32 = 21,600, just making a point the the sub tile is equally divisible in all levels.

Your project, if your input Equirc is 20,000 then this divided by 4 will be 5000px, so this is correct.

Using Pi, you can set Pano2vr to use Pi, but this will make having nicely divisible tiles a bit more of a pain as it will generate odd numbers.
Hope this helps,
Hopki

[timmy2]

Thank you for the explanation, Hopki!

Given that my original equi. image is 20K px wide and the resulting multiresolution levels are 5000, 2500, and 1250, what Level Tile Size would you recommend and what criteria do you use to select it (other than that every level should be evenly divisible by it)? I chose 1250 but in older posts I've seen much smaller values used.

Also, given that I set Window Size to 1280 x 768 is the highest level of 1250 appropriate or should I use a different level value for the highest one ("high" as defined by the tutorial where one should think of the levels as an inverted pyramid, with the smallest level being the highest point on the pyramid)?