Page 1 of 1

Cube face compression issue

Posted: Tue Apr 22, 2008 8:01 pm
by Wim.Koornneef
Hello Thomas,

When I convert an equirectangular to cube faces with maximum JPEG quality and maximum size (for retouching purposes) and then create a flash9 pano the compression setting (image quality) of the cube faces in the Settings tab is ignored unless I change the size of the faces.
Perhaps this is a feature (if size in=size out then quality in=quality out), but it would be nice if there is a possibility to output with maximum face size and a lower quality to avoid a huge file size.

Best,

Wim.

Re: Cube face compression issue

Posted: Wed Apr 23, 2008 3:31 pm
by 360Texas
Not sure why you are using .jpg's. Using this file type generally produces less than best quality [read detail, color]

We take RAW, convert to 16bit tif's, stitch to 16bit equirectangle typically 5000 x 2500.

In Pano2VR create cube face image set. Photoshop edit the 16bit nadir and zenith. Convert back to 16bit equirectangle telling Pano2VR that you want a 5000 pixel wide panorama. (or go directly from cube faces to Flash or QTVR)

Then convert it to Flash + remapped thumbnail.

I have found that cube face pixel size controls the image detail quality. Individual cube face compression % control the end file size. Usually a combination of both results are clean, sharp and colorful and about 1.5mb suitable for broadband viewing.

Example:

5000 x 2500 equirectangle image with cube face size of 600 pixels and individual cube face compression of 70%, 70, 70, 70, 50 top and 50 bottom produces a poor quality image.

5000 x 2500 equirectangle image with cube face size of 1900 pixels and individual cube face compression of 70%, 70, 70, 70, 50 top and 50 bottom produces a more detail and color quality .swf or qtvr.

If the .swf or qtvr file size is too large say 2.5 - 3.0mb then keep the cube face size of 1900 and change the compressions to 50%, 50, 50, 50, 50 top and 30 bottom may change the file size to 1.5mb +/-

Usually a combination of both results are clean, sharp and colorful and about 1.5mb suitable for broadband viewing.

Re: Cube face compression issue

Posted: Wed Apr 23, 2008 4:14 pm
by Wim.Koornneef
360Texas wrote:Not sure why you are using .jpg's. Using this file type generally produces less than best quality [read detail, color]

We take RAW, convert to 16bit tif's, stitch to 16bit equirectangle typically 5000 x 2500.

In Pano2VR create cube face image set. Photoshop edit the 16bit nadir and zenith. Convert back to 16bit equirectangle telling Pano2VR that you want a 5000 pixel wide panorama. (or go directly from cube faces to Flash or QTVR)

Then convert it to Flash + remapped thumbnail.

I have found that cube face pixel size controls the image detail quality. Individual cube face compression % control the end file size. Usually a combination of both results are clean, sharp and colorful and about 1.5mb suitable for broadband viewing.

Example:

5000 x 2500 equirectangle image with cube face size of 600 pixels and individual cube face compression of 70%, 70, 70, 70, 50 top and 50 bottom produces a poor quality image.

5000 x 2500 equirectangle image with cube face size of 1900 pixels and individual cube face compression of 70%, 70, 70, 70, 50 top and 50 bottom produces a more detail and color quality .swf or qtvr.

If the .swf or qtvr file size is too large say 2.5 - 3.0mb then keep the cube face size of 1900 and change the compressions to 50%, 50, 50, 50, 50 top and 30 bottom may change the file size to 1.5mb +/-

Usually a combination of both results are clean, sharp and colorful and about 1.5mb suitable for broadband viewing.
Hello Dave,

Thanks for your reply but I think it is a little off-topic ;-)
I guess you are curious why I use 8bit JPEG instead of 16bit TIFF.
As usual there are many ways to get fine results and just like you I have my own workflow that "works" for me.
I prefer to work as much as possible in 8bit to enhance processing speed en to reduce file sizes.
When the pre processing of the RAWs is done properly and optimized to the max then the need of post processing of the stitched equi is very little.
By carefully pre processing the RAWs I will not get nasty 8bit issues like banding in blue skies afterwards.
When working with max. quality 100% JPEG cube faces instead of TIFFs the loss of quality after a single retouch / re-compress step is minimal and when the cube faces are used to create directly a Flash pano instead of remapping the set to an equi then the loss of details, sharpness, or reduced colorspace is absolutely not noticable.

Best,

Wim

Re: Cube face compression issue

Posted: Thu Apr 24, 2008 2:04 am
by 360Texas
hmm interesting 8bit work flow. Please show us your latest imaging.

Re: Cube face compression issue

Posted: Thu Apr 24, 2008 5:24 pm
by 360Texas

Re: Cube face compression issue

Posted: Mon Apr 28, 2008 2:12 am
by thomas
Wim.Koornneef wrote:Hello Thomas,

When I convert an equirectangular to cube faces with maximum JPEG quality and maximum size (for retouching purposes) and then create a flash9 pano the compression setting (image quality) of the cube faces in the Settings tab is ignored unless I change the size of the faces.
Perhaps this is a feature (if size in=size out then quality in=quality out), but it would be nice if there is a possibility to output with maximum face size and a lower quality to avoid a huge file size.
Have you seen that I added an option in the settings to save the TIFF files compressed? Of course I will also look into the JPEG issue to see what I can do.