hi,
i want to make a pano with really good settings for offline presentation.
the user should be able to zoom in quite a bit with still keeping much image quality.
the start FoV would be 70, but as said zooming in should give good results.
when i use higher cube face sizes as the optimal value would the image quality suffer with a FoV of 70 due to interpolation ?
what would be the optimal cube face size for this ?
thanks
michael
optimal cube face size
-
360Texas
- Moderator
- Posts: 3684
- Joined: Sat Sep 09, 2006 6:06 pm
- Location: Fort Worth, Texas USA
- Contact:
Optimum cube face size helps maintain the image quality.
If your panorama is 5000 pixels wide then cube face can be determined by dividing the image width dimension by Pi or 3.1416 . Cube face size would be 1592 pixels.
Then use cube face compression to bring the image file size down. Try using for L, b, r, f, all equal 70 %.
Then Top not less than 35% because blue sky gets pixelated if compressed too much.
Then Bottom can be 30% because there usually lots of detail
Result file size should be around 1.5mb
If your panorama is 5000 pixels wide then cube face can be determined by dividing the image width dimension by Pi or 3.1416 . Cube face size would be 1592 pixels.
Then use cube face compression to bring the image file size down. Try using for L, b, r, f, all equal 70 %.
Then Top not less than 35% because blue sky gets pixelated if compressed too much.
Then Bottom can be 30% because there usually lots of detail
Result file size should be around 1.5mb
thanks dave
but maybe it is not clear what my question is.
file size is not the problem here as i am doing offline presentations.
let´s assume i have a fixed window size of 1024x768.
for a start FoV of 70 the optimal cube face size would be 1096px.
the qestion is would a bigger cube face size give worse results when using the starting FoV of 70 (of course zooming in would result in better image quality)
as i am rendering the input images this is quite important to me, as i want to be able to zoom in as close as possible.
another question would be the performance.
is using bigger cube face sizes very bad for performance when the window size (view size) stays the same ?
but maybe it is not clear what my question is.
file size is not the problem here as i am doing offline presentations.
let´s assume i have a fixed window size of 1024x768.
for a start FoV of 70 the optimal cube face size would be 1096px.
the qestion is would a bigger cube face size give worse results when using the starting FoV of 70 (of course zooming in would result in better image quality)
as i am rendering the input images this is quite important to me, as i want to be able to zoom in as close as possible.
another question would be the performance.
is using bigger cube face sizes very bad for performance when the window size (view size) stays the same ?
-
360Texas
- Moderator
- Posts: 3684
- Joined: Sat Sep 09, 2006 6:06 pm
- Location: Fort Worth, Texas USA
- Contact:
I have changed this comment several times over the last hour. C4D. Is this Cinema 4D software ?
Ok OFF line. Meaning you are either displaying the panorama from a notebook hard drive or you have placed the panoramas on a DVD and are running them on the notebook.
Determine cube face size of 3443 pixels/ Pi 3.1416 = 1096 pixels
and the inverse is true too
Cube face size of 1096 px x Pi 3.1416 = 3443 pixels wide.
1024 x 768 is also a notebook screen native resolution. Also remember that unless you go full screen option the browser top IE tool bar takes up at least 120 pixels so screen height available vertical space would be about 648. And you cube face px changes to 925px
What would I use ? Sometimes I use 6000 x 3000 or 5000 x 2500 for the World Wide panorama Event. I still get excellent results displayed on our 24" or 12" notebook monitor
Cube face dimensions technique still apply regarding best image quality (regardless of file size). You are correct if you leave the image compression for all cube faces at 100%.
Other stuff that might be of interest.
I have been told that normal FOV for human eye sight to look normal is 50°. That is why 50mm lenses are typically stock lenses for new cameras. Pano2VR uses a default value of 70 which is good too. Only issue is when viewing the panorama.. the zoom in/ out limits must be manually set to prevent zooming out too far and seeing BARREL distortion and zooming in too far that the image gets pixelated. So
In Pano2VR main screen I would select 'Viewing Parameters' and click the [modify] button and
top 3 values are Pan, tilt, FoV.
I would use the
pan to rotate around to where you wish to see the initial view. I am looking at mine it says 345.5 pan angle.
Tilt is tilted down -1.5 degrees from the horizon because I have a interest item I want the visitor to see first.
Fov is set to 71.0 which is very close to your FoV value. You can use the Shift key to zoom IN and the CTRL key to zoom out. Notice the values change on the left side "Current". When you feel visually comfortable that this viewer is showing the INITIAL VIEW zoom in/out .. then click on the [SET] button.
Below is Field of View (Zoom) limits. Think of a water well with a limit switch... water drains down too far and the limit switch turns on to fill .. when the water rises up.. the limit switch float then clicks and turns off the pump. These MIN MAX values are similar to the float limit switch functions.
MIN means zoom IN value.. this the angle of view in horizon degrees the viewer will be able to see when fully zoomed in.
MAX means zoom OUT value. This is the widest angle of view in degrees the viewer will be ble to see when fully zoomed out. Too far out generally produces a very barrel distorted view of the scene.
So with an angle of view of 70 degrees viewing angle from left to right, I would probably set the zoomed in limit angle of view to say around a narrower 30 degrees. And 100 degrees viewing angle on the horizon.. would be far enough zoom out.
LIMITS MIN zoom in 30 <<< initial view 70 >>>>>> 100 MAX zoom out.
By tweaking the Viewing Parameter/ limits FoV up/down increment arrows you can visually see the different viewing angle values to use in the MIN/MAX value boxes.
What values to use? Really its your judgement call. But with each Max /Min setting I would return to the main menu.. and generate and view the results until you are satisified with the value results.
Your last question
BUT you probably already knew all this.
Ok OFF line. Meaning you are either displaying the panorama from a notebook hard drive or you have placed the panoramas on a DVD and are running them on the notebook.
I do see where you are getting this information... in the Flash/QTVR OUTput dialog box. Insert the Window size 1024 x 768 and the recommended Cube face Optimal adjusts to 1096 pixels. So it follows, that for Optimal cube face size is 1096.. then what would be panorama image width ?let´s assume i have a fixed window size of 1024x768.
for a start FoV of 70 the optimal cube face size would be 1096px.
Determine cube face size of 3443 pixels/ Pi 3.1416 = 1096 pixels
and the inverse is true too
Cube face size of 1096 px x Pi 3.1416 = 3443 pixels wide.
1024 x 768 is also a notebook screen native resolution. Also remember that unless you go full screen option the browser top IE tool bar takes up at least 120 pixels so screen height available vertical space would be about 648. And you cube face px changes to 925px
What would I use ? Sometimes I use 6000 x 3000 or 5000 x 2500 for the World Wide panorama Event. I still get excellent results displayed on our 24" or 12" notebook monitor
Cube face dimensions technique still apply regarding best image quality (regardless of file size). You are correct if you leave the image compression for all cube faces at 100%.
Other stuff that might be of interest.
I have been told that normal FOV for human eye sight to look normal is 50°. That is why 50mm lenses are typically stock lenses for new cameras. Pano2VR uses a default value of 70 which is good too. Only issue is when viewing the panorama.. the zoom in/ out limits must be manually set to prevent zooming out too far and seeing BARREL distortion and zooming in too far that the image gets pixelated. So
In Pano2VR main screen I would select 'Viewing Parameters' and click the [modify] button and
top 3 values are Pan, tilt, FoV.
I would use the
pan to rotate around to where you wish to see the initial view. I am looking at mine it says 345.5 pan angle.
Tilt is tilted down -1.5 degrees from the horizon because I have a interest item I want the visitor to see first.
Fov is set to 71.0 which is very close to your FoV value. You can use the Shift key to zoom IN and the CTRL key to zoom out. Notice the values change on the left side "Current". When you feel visually comfortable that this viewer is showing the INITIAL VIEW zoom in/out .. then click on the [SET] button.
Below is Field of View (Zoom) limits. Think of a water well with a limit switch... water drains down too far and the limit switch turns on to fill .. when the water rises up.. the limit switch float then clicks and turns off the pump. These MIN MAX values are similar to the float limit switch functions.
MIN means zoom IN value.. this the angle of view in horizon degrees the viewer will be able to see when fully zoomed in.
MAX means zoom OUT value. This is the widest angle of view in degrees the viewer will be ble to see when fully zoomed out. Too far out generally produces a very barrel distorted view of the scene.
So with an angle of view of 70 degrees viewing angle from left to right, I would probably set the zoomed in limit angle of view to say around a narrower 30 degrees. And 100 degrees viewing angle on the horizon.. would be far enough zoom out.
LIMITS MIN zoom in 30 <<< initial view 70 >>>>>> 100 MAX zoom out.
By tweaking the Viewing Parameter/ limits FoV up/down increment arrows you can visually see the different viewing angle values to use in the MIN/MAX value boxes.
What values to use? Really its your judgement call. But with each Max /Min setting I would return to the main menu.. and generate and view the results until you are satisified with the value results.
Your last question
The answer is that it probably has little affect as long as the cube face size is proportioned to the panorama image width. You achieve little in the way of quality or computer video card display performance by making the cube face sizes larger. Bigger cube face tile sizes? Needs BIGGER stitched panorama width.another question would be the performance.
is using bigger cube face sizes very bad for performance when the window size (view size) stays the same ?
BUT you probably already knew all this.
wow, dave thanks for the long reply,
unfortunatly yes, i knew almost all of this, but thanks anyway.
what i need to know about is what size i have to render my image (on a 3d app - c4d in my case) to have best image quality (on starting FoV and when zooming in) in my pano.
now when i use Pano2VR and set my starting FoV and the windows resolution i get a optimal dimension of the cube face size (in my case this was 1096px - this is under the QTVR output) so if i would change the window size or the FoV there would be another optimal resolution (lower FoV leads to bigger resolution)
i am rendering a cross image out of my 3dapp to be converted to a pano with Pano2VR. i can caculate the image size of this rendering to be 3 times the optimal resolution in one dimesnion and 4 times the optimal resolution in the other dimension (as the cross has all six cube faces on a 3x4 image format) resulting in 3288 x 4386px
now what Pano2VR gives as the optimal cube size resolution is based on the starting FoV and the window size (we forget about the window size as this won´t change).
the question is if i render in much better resolution (lets say 6000x8000) will the pano quality for the starting FoV will look the same or even better as with the lower resolution or will it be worse because the image has to be scaled down (from the player)
when i zoom in the quality will of course be much better with the bigger image - but i am afraid that the normal view will suffer a bit.
unfortunatly yes, i knew almost all of this, but thanks anyway.
what i need to know about is what size i have to render my image (on a 3d app - c4d in my case) to have best image quality (on starting FoV and when zooming in) in my pano.
now when i use Pano2VR and set my starting FoV and the windows resolution i get a optimal dimension of the cube face size (in my case this was 1096px - this is under the QTVR output) so if i would change the window size or the FoV there would be another optimal resolution (lower FoV leads to bigger resolution)
i am rendering a cross image out of my 3dapp to be converted to a pano with Pano2VR. i can caculate the image size of this rendering to be 3 times the optimal resolution in one dimesnion and 4 times the optimal resolution in the other dimension (as the cross has all six cube faces on a 3x4 image format) resulting in 3288 x 4386px
now what Pano2VR gives as the optimal cube size resolution is based on the starting FoV and the window size (we forget about the window size as this won´t change).
the question is if i render in much better resolution (lets say 6000x8000) will the pano quality for the starting FoV will look the same or even better as with the lower resolution or will it be worse because the image has to be scaled down (from the player)
when i zoom in the quality will of course be much better with the bigger image - but i am afraid that the normal view will suffer a bit.
The quality will suffer a bit because the Flash scaler tends produce aliasing. You can avoid this (and trade it in for speed) a bit if you disable "change stage quality" and "change bitmap filtering".
MfG, Thomas
I guess it is almost the same. The best way is to test it and to see if you can live with it.
MfG, Thomas