DAT (Ever17): Difference between revisions
No edit summary |
|||
| Line 8: | Line 8: | ||
!colspan="15"|Header | !colspan="15"|Header | ||
|- | |- | ||
! Size !! Content | ! Size !! Content !! Description | ||
|- | |- | ||
| 4Bytes || Magic/ID | | 4Bytes || Magic/ID || | ||
|- | |- | ||
| 4Bytes|| File Count | | 4Bytes|| File Count || | ||
|- | |- | ||
| 8Bytes|| 0x00 Filler | | 8Bytes|| 0x00 Filler || | ||
| | |- | ||
| class="wikitable" | |||
!colspan="15"|Index | !colspan="15"|Index | ||
|- | |- | ||
! Size !! Content | ! Size !! Content !! Description | ||
|- | |- | ||
| 4Bytes || Offset | | 4Bytes || Offset || | ||
|- | |- | ||
| 4Bytes|| File Size | | 4Bytes|| File Size || | ||
|- | |- | ||
| 24Bytes|| File name | | 24Bytes|| File name || | ||
|} | |} | ||
| Line 33: | Line 33: | ||
Internal formats, specifically *.JPG's from the wallpaper.dat have a strange thing about them. While I have only tested a few files, it appears about 4,000 bytes into the file, there is a 256 byte chuck that differs which breaks the image. When using the in-game option to save wallpapers, the image is proper and correct. Will most likely require a disassembler to figure out what exactly is going on. Note: This is only about the *.JPG's inside wallpaper.dat. Other files don't show this pattern. | Internal formats, specifically *.JPG's from the wallpaper.dat have a strange thing about them. While I have only tested a few files, it appears about 4,000 bytes into the file, there is a 256 byte chuck that differs which breaks the image. When using the in-game option to save wallpapers, the image is proper and correct. Will most likely require a disassembler to figure out what exactly is going on. Note: This is only about the *.JPG's inside wallpaper.dat. Other files don't show this pattern. | ||
2014/10/07 Research | |||
Program only reads from 'wallpaper.dat' when saving files. In-engine view of picture uses the game asset version loaded from 'bg.dat'. Currently using dissembler/debugger to see what it does with the file after it has loaded it into memory. External monitor shows it first loads the first 16 bytes of the file (wallpaper.da), which is the header, and then loads 22,778 bytes which is probably the entire index since when it reads the JPG file, it seeks to position and reads size of JPG without checking index. Need to compare more than one jpg to see if part that is different is predictable. This may only apply to wallpaper.dat since no other files extracted show this issue. the JPGs could have been | |||
processed before building the wallpaper archive and the engine, when saving, will reverse/repair the broken part when saving. | |||
-Possible test: | |||
Extract all data and replace a specific image and try to extract it and see what the results are. | |||
Revision as of 21:15, 10 July 2014
Seen/used in the follow game(s):
- Ever17
Structure
| Header | |||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Size | Content | Description | |||||||||||||
| 4Bytes | Magic/ID | ||||||||||||||
| 4Bytes | File Count | ||||||||||||||
| 8Bytes | 0x00 Filler | ||||||||||||||
| class="wikitable" | Index | ||||||||||||||
| Size | Content | Description | |||||||||||||
| 4Bytes | Offset | ||||||||||||||
| 4Bytes | File Size | ||||||||||||||
| 24Bytes | File name | ||||||||||||||
Stored offset starts at zero. Actual offset = (16+(file count * 32)).
Stored file size is doubled. Actual file size = (file size /2).
Internal formats, specifically *.JPG's from the wallpaper.dat have a strange thing about them. While I have only tested a few files, it appears about 4,000 bytes into the file, there is a 256 byte chuck that differs which breaks the image. When using the in-game option to save wallpapers, the image is proper and correct. Will most likely require a disassembler to figure out what exactly is going on. Note: This is only about the *.JPG's inside wallpaper.dat. Other files don't show this pattern.
2014/10/07 Research Program only reads from 'wallpaper.dat' when saving files. In-engine view of picture uses the game asset version loaded from 'bg.dat'. Currently using dissembler/debugger to see what it does with the file after it has loaded it into memory. External monitor shows it first loads the first 16 bytes of the file (wallpaper.da), which is the header, and then loads 22,778 bytes which is probably the entire index since when it reads the JPG file, it seeks to position and reads size of JPG without checking index. Need to compare more than one jpg to see if part that is different is predictable. This may only apply to wallpaper.dat since no other files extracted show this issue. the JPGs could have been processed before building the wallpaper archive and the engine, when saving, will reverse/repair the broken part when saving. -Possible test: Extract all data and replace a specific image and try to extract it and see what the results are.