4 edited tags
| link
3 edited title
| link

Comparing two sets of different files with pairs of the same image each in raw format Convert from DDS (b8g8r8a8)/uexp (custom bgra) to uexp (custom abgr +)

2 added 402 characters in body
source | link

I have two datasets. Files with the same name contain exactly the same image. But first set uses different alghoritm than second one.

For second data set I can convert files to f.e. DDS and reverse. For first one I can't. This two data sets are similar, but not enough to make packer for first data set files (even sizes are different, but in binary they look very similar).

What tool would be the best to use with that amount of data (>100 000, but needs to be sorted first by type) to find pattern that will conclude in making script converting one file to another?

Edit: tl;dr i want to make converter with machine learning

From: DDS (b8g8r8a8)/uexp (custom bgra)

To: uexp (custom abgr + 2 additional bytes for each pixel and they are chunked/swizzled + some additional bytes) uexp are uncompressed unencrypted archives for data, where one file = one texture.

It needs to be cleaned from additional informations at the beginning and end of file before analyzing to simplify process.

I have two datasets. Files with the same name contain exactly the same image. But first set uses different alghoritm than second one.

For second data set I can convert files to f.e. DDS and reverse. For first one I can't. This two data sets are similar, but not enough to make packer for first data set files (even sizes are different, but in binary they look very similar).

What tool would be the best to use with that amount of data (>100 000, but needs to be sorted first by type) to find pattern that will conclude in making script converting one file to another?

I have two datasets. Files with the same name contain exactly the same image. But first set uses different alghoritm than second one.

For second data set I can convert files to f.e. DDS and reverse. For first one I can't. This two data sets are similar, but not enough to make packer for first data set files (even sizes are different, but in binary they look very similar).

What tool would be the best to use with that amount of data (>100 000, but needs to be sorted first by type) to find pattern that will conclude in making script converting one file to another?

Edit: tl;dr i want to make converter with machine learning

From: DDS (b8g8r8a8)/uexp (custom bgra)

To: uexp (custom abgr + 2 additional bytes for each pixel and they are chunked/swizzled + some additional bytes) uexp are uncompressed unencrypted archives for data, where one file = one texture.

It needs to be cleaned from additional informations at the beginning and end of file before analyzing to simplify process.

1
source | link