TIFF file size reduces by more than half after adding EXIF data

[Richard D.]

I've been scanning some of my old prints using my Canon Pixma MP760 printer/scanner with ScanGear MP software. After saving a file, Windows tells me the file size is 197 MB.

I right click on the file and select Properties, then select the Details tab so that I can add details to the file such as author, copyright, camera type, etc. (Is this known as EXIF data?)

After adding these details to the file, when I check the file size again Windows tells me it is now 88.6 MB, i.e. less than half the size of the file before I entered the copyright details, etc. Does anyone have any idea what is going on?

I'm using a fully patched version of Windows Vista Ultimate (64-bit version). According to the Windows file details, the horizontal and vertical resolution are both 1600 dpi, bit depth is 24, the compression algorithm is LZW, and the Resolution Unit is 2 (whatever that means). These details do not change.

[pegelli]

Did the bit depth drop from 16 to 8 bit?

[Richard D.]

@pegelli

No, the Windows file details show the bit depth as remaining at 24. There is no change.

[kaelaria]

Windows always saves in LZW compression.  Your original was not.

[Ellis Vener]

@pegelli

No, the Windows file details show the bit depth as remaining at 24. There is no change.

So you always were in 8 bit per channel mode .

[Richard D.]

@kaelaria: The file details show that the originally saved file uses LZW compression.

@ Ellis Vener: Yes, I didn't change the bits per channel mode at any point and I cannot even find a way to change it.

A couple of people on the dpreview forums suggested I should convert both the original TIFF file and the TIFF file with EXIF data added, to a noncompressed, lossless format (BMP) and then perform a byte to byte comparison to see if any image data has been lost.

So I did, using Microsoft's File Compare command, and no differences were encountered.

So I think my Canon ScanGear MP software is probably using an older, less efficient version of the LZW compression algorithm than Windows does. When I add EXIF data to the file in Windows, Windows appears to automatically re-run the compression process to reduce the file size. That's my best guess anyway.

[kaelaria]

So do a test, make some scans without compression and with, compare them then do the windows edits on both and compare them.  I bet you'll find it just says it was compressing but really wasn't.  I'm not aware of any such version problems that would do such a thing.  More likely it's just a bug in the scan software not actually compressing.

[Richard D.]

@kaelaria: Thanks for your suggestions.

You're probably right that it wasn't compressing at all, despite saying that it was. My scanning software is old and there is no option to change the compression level. I'm trying to get some newer software and explanations from Canon's technical support people.

[Bart_van_der_Wolf]

@kaelaria: Thanks for your suggestions.

You're probably right that it wasn't compressing at all, despite saying that it was. My scanning software is old and there is no option to change the compression level. I'm trying to get some newer software and explanations from Canon's technical support people.

Hi Richard,

From the descriptions, it still looks more like a conversion from 16-bits per channel to 8-b/ch. Trying to compress a 16-b/ch scan file may effectively increase the file size (!) and each file will have a different size (because potential compression is image content dependent). You need a 16-b/ch application to really see what's going on.

Cheers,
Bart

[Richard D.]

kaelaria was right. Canon's scanning software is NOT compressing the TIFFs when saving them.

According to the Windows file properties of several files that I scanned today, they are actually "uncompressed" when saved straight from the scanning software. As soon as I apply EXIF data to a file there is a slight pause as Windows does something. Then the file properties show it has been compressed using the LZW algorithm.

It still doesn't explain why the scanner outputs my scans as 8-bit per channel files, but that's something for Canon's technical support people to answer. It looks like Canon haven't released an update for my scanner since January 2007 so I'm not hopeful.

Thank you for your help. I guess I will have to buy something more professional and reliable than an all-in-one printer/copier/scanner next time.

[kaelaria]

Anytime!

[Richard D.]

For the sake of completeness, here is the response from Canon Consumer Support as to why Canon's scanning software has been saving files with a bit depth of 24 instead of 48:

"The scanner unit in the printer is able to produce 48-bit scans. Our software, MP Navigator, on the other hand does not support 48-bit images. To be able to scan images in 48-bit you will have to use a third hand software such as Photoshop or Silverfast. MP Navigator is a basic software that is a bit limited when coming to more advanced functions."

Like most consumer support services these days, they made me uninstall and reinstall all my Canon software before escalating my enquiry to someone who knew the answer.