Announcement

**stefan** · 19.03.2012, 10:43 PM

Hi Brian,

you can also use ImageMagick for that problem - via commandline you can remove all exif data:

Code:

mogrify -strip sample.jpg

What be great if you can provide us a testfile or something. So I can try it with the commandline command and later we can check the filesize!

So you can use IrfanView for that kind of "test" and I am using ImageMagick and later we can compare our results!

Stefan

**BJ1200** · 20.03.2012, 03:54 AM

Hi Stefan,

I'm not proficient with command line steps. I'm a GUI guy.
Here's a Folder that includes 43 images and Excel with comparison files sizes before and after IrfanView. Now that I see them side by side, it makes no sense what's going on. Take note of TerriGarr.jpg in the spreadsheet. It went from 47KB to 113KB after removing all metadata. The test had all "Options" boxes unchecked (no XMP, no EXIF, etc.)

Thanks.

**stefan** · 20.03.2012, 10:00 PM

Hi,

input files:

Code:

[stefan@politeia Invelos Update]$ du -skb * > old.txt
66633    allen bestwick.jpg
29759    Andrea Romano.jpg
28329    barney hall.jpg
35175    ben drew.jpg
105164    billoah greene.jpg
42672    BJ Hogg.jpg
152343    Bobby Labonte.jpg
204402    Bruce Timm.jpg
175466    Cary Alexander.jpg
37763    catalina rodriguez.jpg
71735    Cathy Cahin Ryan.jpg
25291    Christian Murphy.jpg
47163    Danny Jacobs.jpg
171149    david snell.jpg
200301    jack oconnell.jpg
5244    Jamie Downey.jpg
311628    Jeff Gordon.jpg
221497    Jimmie Johnson.jpg
14715    Joe Moore.jpg
56234    john fricker.jpg
36920    John Pickard.jpg
25908    Joseph Gilgun.jpg
124342    julianne grossman.jpg
28086    Keith Ferguson.jpg
83457    Ken Leung.jpg
189489    Keyonna Patterson.jpg
74552    Liz Dainiels.jpg
104490    Mary Alice Haney.jpg
221684    Matt Kenseth.jpg
19702    michael chiklis.jpg
27022    Micke Spreitz.jpg
28632    rasmus hardiker.jpg
33961    rebecca chaney.jpg
69179    Rick Gifford.jpg
137515    ryan newman.jpg
24359    Saratoga Ballentine.jpg
266524    simon farnaby.jpg
155083    steven christopher parker.jpg
125360    Susan Eisenberg.jpg
48147    Teri Garr.jpg
162943    Tony Stewart.jpg
81470    Wade Williams.jpg

(Size in kb)

Then remove all exif:

Code:

[stefan@politeia Invelos Update]$ mogrify -strip *.jpg

Gives:

Code:

[stefan@politeia Invelos Update]$ du -skb * > new.txt
63806    allen bestwick.jpg
27437    Andrea Romano.jpg
25036    barney hall.jpg
33120    ben drew.jpg
95501    billoah greene.jpg
42055    BJ Hogg.jpg
143919    Bobby Labonte.jpg
181433    Bruce Timm.jpg
113178    Cary Alexander.jpg
34161    catalina rodriguez.jpg
71689    Cathy Cahin Ryan.jpg
22670    Christian Murphy.jpg
43362    Danny Jacobs.jpg
154290    david snell.jpg
180480    jack oconnell.jpg
4579    Jamie Downey.jpg
284421    Jeff Gordon.jpg
228996    Jimmie Johnson.jpg
14712    Joe Moore.jpg
59687    john fricker.jpg
32846    John Pickard.jpg
26275    Joseph Gilgun.jpg
116324    julianne grossman.jpg
25936    Keith Ferguson.jpg
77889    Ken Leung.jpg
178470    Keyonna Patterson.jpg
68493    Liz Dainiels.jpg
97140    Mary Alice Haney.jpg
229877    Matt Kenseth.jpg
20175    michael chiklis.jpg
24876    Micke Spreitz.jpg
25053    rasmus hardiker.jpg
29629    rebecca chaney.jpg
66073    Rick Gifford.jpg
69591    ryan newman.jpg
22413    Saratoga Ballentine.jpg
249183    simon farnaby.jpg
147880    steven christopher parker.jpg
133561    Susan Eisenberg.jpg
47317    Teri Garr.jpg
94759    Tony Stewart.jpg
78263    Wade Williams.jpg

So I gonna write a little c++ program which stores filename + the file size for both lists in two different maps:

Code:

#include <map>
#include <iostream>
#include <fstream>
#include <string>
#include <boost/regex.hpp>
#include <boost/lexical_cast.hpp>

using namespace std;
using namespace boost;

typedef map<string, long> dataSheet;


void readFileIntoMap(string &fileName, dataSheet &map) {
	
	ifstream inputFileStream{fileName.c_str()};
	
	string currentLine;
	
	while(getline(inputFileStream, currentLine)) {
	
		// Boost Matching
		regex pattern{"(\\d*)\\s*(.+)"};
		
		smatch result;
		
		if (regex_match(currentLine, result, pattern)) {
			
			string currentFileName{(string)result[2]};
			
			long currentFileSize = boost::lexical_cast<long>(result[1]);
			
			map.insert(dataSheet::value_type(currentFileName, currentFileSize));
		}
	}
}

int main()  {
	
	string beforeStrippingExif{"./old.txt"};
	string afterStrippingExif{"./new.txt"};
	
	// Map contains filename and filesize - multiple filenames are not allowed!!!!
	// We are a bit restricted: no unicode filenames, please ;)
	dataSheet exifStrippingSheetBefore;
	dataSheet exifStrippingSheetAfter;
		
	readFileIntoMap(beforeStrippingExif, exifStrippingSheetBefore);
	readFileIntoMap(afterStrippingExif, exifStrippingSheetAfter);
	
	cout << "FileSize - size before stripping / size after stripping - diff" << endl;
	
	for (auto it = exifStrippingSheetBefore.cbegin(); it != exifStrippingSheetBefore.cend(); ++it) {
		// print some statistics
		
		if (exifStrippingSheetAfter.count(it->first)) {
			cout << it->first << " - " << it->second;
			cout << " / " << exifStrippingSheetAfter[it->first] << " - " << (long)it->second - (long)exifStrippingSheetAfter[it->first] << endl;
		}
	}
}

Compiled via:

Code:

[stefan@politeia Invelos Update]$ g++ -std=c++0x -lboost_regex exifSheet.cpp

Results:

Code:

FileSize - size before stripping / size after stripping - diff
Andrea Romano.jpg - 29759 / 27437 - 2322
BJ Hogg.jpg - 42672 / 42055 - 617
Bobby Labonte.jpg - 152343 / 143919 - 8424
Bruce Timm.jpg - 204402 / 181433 - 22969
Cary Alexander.jpg - 175466 / 113178 - 62288
Cathy Cahin Ryan.jpg - 71735 / 71689 - 46
Christian Murphy.jpg - 25291 / 22670 - 2621
Danny Jacobs.jpg - 47163 / 43362 - 3801
Jamie Downey.jpg - 5244 / 4579 - 665
Jeff Gordon.jpg - 311628 / 284421 - 27207
Jimmie Johnson.jpg - 221497 / 228996 - -7499
Joe Moore.jpg - 14715 / 14712 - 3
John Pickard.jpg - 36920 / 32846 - 4074
Joseph Gilgun.jpg - 25908 / 26275 - -367
Keith Ferguson.jpg - 28086 / 25936 - 2150
Ken Leung.jpg - 83457 / 77889 - 5568
Keyonna Patterson.jpg - 189489 / 178470 - 11019
Liz Dainiels.jpg - 74552 / 68493 - 6059
Mary Alice Haney.jpg - 104490 / 97140 - 7350
Matt Kenseth.jpg - 221684 / 229877 - -8193
Micke Spreitz.jpg - 27022 / 24876 - 2146
Rick Gifford.jpg - 69179 / 66073 - 3106
Saratoga Ballentine.jpg - 24359 / 22413 - 1946
Susan Eisenberg.jpg - 125360 / 133561 - -8201
Teri Garr.jpg - 48147 / 47317 - 830
Tony Stewart.jpg - 162943 / 94759 - 68184
Wade Williams.jpg - 81470 / 78263 - 3207
allen bestwick.jpg - 66633 / 63806 - 2827
barney hall.jpg - 28329 / 25036 - 3293
ben drew.jpg - 35175 / 33120 - 2055
billoah greene.jpg - 105164 / 95501 - 9663
catalina rodriguez.jpg - 37763 / 34161 - 3602
david snell.jpg - 171149 / 154290 - 16859
jack oconnell.jpg - 200301 / 180480 - 19821
john fricker.jpg - 56234 / 59687 - -3453
julianne grossman.jpg - 124342 / 116324 - 8018
michael chiklis.jpg - 19702 / 20175 - -473
rasmus hardiker.jpg - 28632 / 25053 - 3579
rebecca chaney.jpg - 33961 / 29629 - 4332
ryan newman.jpg - 137515 / 69591 - 67924
simon farnaby.jpg - 266524 / 249183 - 17341
steven christopher parker.jpg - 155083 / 147880 - 7203

Stefan

**stefan** · 20.03.2012, 10:04 PM

Hi,

what happened with:

Jimmie Johnson.jpg - 221497 / 228996 - -7499
Joseph Gilgun.jpg - 25908 / 26275 - -367
Matt Kenseth.jpg - 221684 / 229877 - -8193
Susan Eisenberg.jpg - 125360 / 133561 - -8201
john fricker.jpg - 56234 / 59687 - -3453
michael chiklis.jpg - 19702 / 20175 - -473

That means the file is bigger after exif stripping

Stefan

EDIT: I'm using a Fedora (16) system, gcc 4.6.2, ImageMagick 6.7.0-10 2011-11-03 Q16

**stefan** · 20.03.2012, 11:22 PM

Hi,

I just found out that the strip command does some kind of "compressing" to the images:

Code:

[stefan@politeia Pictures]$ perceptualdiff org.jpg mod.jpg 
FAIL: Images are visibly different
50744 pixels are different

[stefan@politeia Pictures]$ perceptualdiff andrea_org.jpg andrea_mod.jpg 
FAIL: Images are visibly different
1790 pixels are different

Very bad. So i tried the perl image tool "exiftool" for removing exif metadata:

Code:

FileSize - size before stripping / size after stripping - diff
Andrea Romano.jpg - 29759 / 29481 - 278
BJ Hogg.jpg - 42672 / 42394 - 278
Bobby Labonte.jpg - 152343 / 152065 - 278
Bruce Timm.jpg - 204402 / 204124 - 278
Cary Alexander.jpg - 175466 / 117385 - 58081
Cathy Cahin Ryan.jpg - 71735 / 71717 - 18
Christian Murphy.jpg - 25291 / 25013 - 278
Danny Jacobs.jpg - 47163 / 46885 - 278
Jamie Downey.jpg - 5244 / 5226 - 18
Jeff Gordon.jpg - 311628 / 311350 - 278
Jimmie Johnson.jpg - 221497 / 221219 - 278
Joe Moore.jpg - 14715 / 14697 - 18
John Pickard.jpg - 36920 / 36642 - 278
Joseph Gilgun.jpg - 25908 / 25855 - 53
Keith Ferguson.jpg - 28086 / 27808 - 278
Ken Leung.jpg - 83457 / 83179 - 278
Keyonna Patterson.jpg - 189489 / 189211 - 278
Liz Dainiels.jpg - 74552 / 74274 - 278
Mary Alice Haney.jpg - 104490 / 104212 - 278
Matt Kenseth.jpg - 221684 / 221406 - 278
Micke Spreitz.jpg - 27022 / 26744 - 278
Rick Gifford.jpg - 69179 / 68901 - 278
Saratoga Ballentine.jpg - 24359 / 24081 - 278
Susan Eisenberg.jpg - 125360 / 125082 - 278
Teri Garr.jpg - 48147 / 48068 - 79
Tony Stewart.jpg - 162943 / 103802 - 59141
Wade Williams.jpg - 81470 / 81192 - 278
allen bestwick.jpg - 66633 / 66355 - 278
barney hall.jpg - 28329 / 28051 - 278
ben drew.jpg - 35175 / 34897 - 278
billoah greene.jpg - 105164 / 104886 - 278
catalina rodriguez.jpg - 37763 / 37485 - 278
david snell.jpg - 171149 / 170871 - 278
jack oconnell.jpg - 200301 / 179632 - 20669
john fricker.jpg - 56234 / 55956 - 278
julianne grossman.jpg - 124342 / 124064 - 278
michael chiklis.jpg - 19702 / 16534 - 3168
rasmus hardiker.jpg - 28632 / 28354 - 278
rebecca chaney.jpg - 33961 / 33683 - 278
ryan newman.jpg - 137515 / 78625 - 58890
simon farnaby.jpg - 266524 / 266246 - 278
steven christopher parker.jpg - 155083 / 154805 - 278

Diff an original and a exiftool "edited" file:

Code:

[stefan@politeia Pictures]$ perceptualdiff Jimmie\ Johnson.jpg mod.jpg

No difference! That means exiftool does a great job in removing exif metadata without touching the image (just filesize will be smaller!).

Hope this could help someone

Stefan

**BJ1200** · 21.03.2012, 07:48 AM

Thanks Stefan, you just put me through Fortran flashbacks from college. That's good to know it's messing with the images. Fortunately, I already have Exiftool....wait for it.......their GUI executable. I'm not sure their GUI can clean. I will look into it.

**BJ1200** · 21.03.2012, 11:35 AM

Hey Stefan,

ExifTool GUI does a great job for metadata cleaning. It creates a report when done of any files that had errors and couldn't be processed, including explanations of the errors. However, on a set of 16,000 images it stalled and the end and couldn't generate the report (Not Responding). I used Irfan and reconverted the error files to JPG and it fixed all the errors.

**stefan** · 21.03.2012, 07:35 PM

Hi,

you should try to use normal batch script (call the exiftool - the exif tool gui also calls it) - maybe it works

Stefan

Announcement

Batch Removing Metadata (EXIF, XMP...) Increases File Size - Why?

Batch Removing Metadata (EXIF, XMP...) Increases File Size - Why?

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment