I tryed converting this 128*128 RGB TGA in PVR OGLPVRTC4:
http://bruno.deligny.googlepages.com/src.tga
And i get this result with the cmd line version:
http://bruno.deligny.googlepages.com/cmd_line_out.pvr
and this result with the GUI version:
http://bruno.deligny.googlepages.com/gui_out.pvr
There are little corruptions with or without borders. I dont think it's a compression artifact because it manage to compress very well the cloud gradiant on the rest of the image.
It's not a serious problem, i dont see it on my PC application or on my IPhone, maybe it's just the displaying in PVRTexTool.
bruno.deligny2009-01-08 15:19:13
Apologies for not getting back to you sooner - I have ben investigating what may have happened, but I must admit I looked at these textures and I wasn’t sure what you were referring to as corruptions. The PVR files you have suffer from compression artefacts, but not more than I might expect from PVRTC. Can you describe the corruption you’re seeing in more detail?
PVRTC is a lossy compression scheme so there will always be differences between a compressed image and and the original. On some images or sections of images the difference is more noticeable and it’s sometimes very difficult to predict how well the scheme will deal with a particular set of data. Work on the PVRTC compressor is ongoing so check the releases of PVRTexTool for improvements and changes to this.
PVRTexTool is designed to let you examine textures very closely so differences can be much more noticeable here than when the texture is used in an application.
I’m sorry I can’t explain more.
I drew a mark that surround the corruptions.
http://bruno.deligny.googlepages.com/src2.tga
I think this is, unfortunately, just an artefact of the PVRTC compression. In this area of the image it looks to me like fairly contrasting colours have been chosen for the PVRTC blocks in that area and so a much more coarse image is produced.
As I said above, hopefully these effects can be improved upon in future.
