Skip navigation
Currently Being Moderated

Why do my JPEG file sizes shrink so dramatically?

Oct 3, 2013 5:02 AM

Tags: #cmyk_size

This has got me completely and utterly baffled.

 

I have an image that I saved using Photoshop as JPEG at max quality and on disk it is 12.3mb.

 

After placing it at 100% in InDesign that spread when exported as a PDF with images set to max quality is a mere 2.9mb. (The spread is two pages, the image on one and the month of a calendar on the other)

 

The image in Photoshop is 3900px by 2650px and at 300ppi it's 33.02cm by 22.44. It is converted to CMYK and saved as JPEG at max 12

 

In InDesign it's frame is 330.2mm x 224.367mm on a document of 318mm x 212mm with 6mm bleed all round.

 

When InDesign is exporting as a PDF it must be recompressing the JPEG? Yet it seems to be throwing away 10mb even at max quality (see below) Or is it that Photoshop is saving CMYK JPEGs wrongly and producing files way too big - I've always thought PS produces huge JPEGs once you convert them to CMYK

 

What tests can I do? Have tried various output settings, these are current

 

screenshot.jpg

 
Replies 1 2 Previous Next
  • Currently Being Moderated
    Oct 4, 2013 1:21 AM   in reply to Amy2014

    Have you verified that the exported PDF contains CMYK images?

     
    |
    Mark as:
  • Currently Being Moderated
    Oct 4, 2013 2:53 AM   in reply to Amy2014

    http://forums.adobe.com/thread/1068393

     

    There's a very real difference between "jpeg" and "automatic jpeg".

     

    It's most likely with Automatic JPEG that the information is being zipped rather than lossy.

     
    |
    Mark as:
  • Currently Being Moderated
    Oct 4, 2013 3:11 AM   in reply to Amy2014

    Choosing JPEG again might be adding another compression level again.

     

    Choosing Automatic JPEG the application will decide what's best, lossy compression (losing actual data), and zipping.

     

    And it's probably zipping the JPEG which would make it smaller, not bigger.

     

     

    If you think you're losing data then choose no compression at all. You'll see a much larger file than what you actually started with.

     

    You can always set your compression setting to compress images of 301 (instead of 450) - and then the image shouldn't be hit by the compression settings at all.

     
    |
    Mark as:
  • Currently Being Moderated
    Oct 4, 2013 3:15 AM   in reply to Amy2014

    If you had Acrobat Pro you would be able to tell if the image was compressed again.

     

    If you're not sure - feel free to send me a private message and you can email me the PDF and I will check out the image to it.

     

    If it's not sensitive, and can be shared openly - feel free to upload to file sharing site, like dropbox, google drive etc. and share a public link.

     
    |
    Mark as:
  • Currently Being Moderated
    Oct 4, 2013 3:34 AM   in reply to Amy2014

    Amy2014 wrote:

     

    I'm pretty sure that for that page InDesign chose max quality JPEG in either case rather than using zip or any other compression.

     

    Well you're not really "pretty sure" - that's a guess.

     

    Get a trial of Acrobat Pro.

     

    You will be able to use Edit to open the image in photoshop to see the real result.

    http://wwwimages.adobe.com/www.adobe.com/content/dam/Adobe/en/products /acrobat/pdfs/adobe-acrobat-xi-edit-text-and-images-in-a-pdf-file-tuto rial-ie.pdf

     

    Plus you could be converting to a different colour profile, like RGB which would add to the data loss (1 less channel than CMYK)

     
    |
    Mark as:
  • Currently Being Moderated
    Oct 4, 2013 3:35 AM   in reply to Eugene Tyson

    Or converting to a colour profile of sRGB -

     
    |
    Mark as:
  • Currently Being Moderated
    Oct 4, 2013 3:58 AM   in reply to Amy2014

    Well you need to go to the Colour portion of the PDF settings and choose "No Colour Conversion"

     

    Don't select anything for compression in the compression portion of the PDF settings.

     
    |
    Mark as:
  • Currently Being Moderated
    Oct 4, 2013 4:11 AM   in reply to Amy2014

    Is your JPEG image set to Coated Fogra 39 in the Colour Profiles in Photoshop?

     

    What are your colour management policies in InDesign?

     
    |
    Mark as:
  • Currently Being Moderated
    Oct 4, 2013 4:14 AM   in reply to Amy2014

    Is it possible to have one page of the exported PDF for analysis? I got Acrobat Pro here.

     

    Just attach it to a reply.

     
    |
    Mark as:
  • Currently Being Moderated
    Oct 4, 2013 4:18 AM   in reply to Doc Maik

    Doc Maik wrote:

     

    Just attach it to a reply.

    Sorry, attachemnts are not enabled on the forums. You would need to link to a server someplace....

     
    |
    Mark as:
  • Currently Being Moderated
    Oct 4, 2013 5:26 AM   in reply to Amy2014

     

     

    What tests can I do?

     

     

    Print it.

     
    |
    Mark as:
  • Currently Being Moderated
    Oct 4, 2013 9:07 AM   in reply to Amy2014

    Hey Gang, just in case you ever wondered about JPG compression:

     

    The JPEG algorithms performs its compression in four phases:

    1. The JPEG algorithms first cuts up an image in blocks of 8x8 pixels. It converts image data to a luminance/chrominance color space, such as YUV. The algorithm retains more of the luminance in the compressed file.
    2. Next, apply a Discrete Cosine Transform (DCT) for the block. It replaces actual pixel color data for values that are relative to the average of the entire matrix. This replaces 8x8 pixel values by an 8x8 matrix of DCT coefficients.
    3. Two tables of quantization constants are calculated; luminance and chrominance. The constants from the two tables are used to quantize the DCT coefficients. Each DCT coefficient is divided by its corresponding constant in the quantization table and rounded off. The result of quantizing the DCT coefficients is that smaller coefficients will be replaced by zeros and larger coefficients will lose precision. This rounding-off causes a loss in quality. The resulting data are a list of streamlined DCT coefficients.
    4. Last, compress these coefficients using either a Huffman or arithmetic encoding scheme. Usually Huffman encoding is used. This is a second (lossless) compression that is applied.

     

    By putting 2 compression algorithms on top of each other, JPEG achieves remarkable compression ratios, up to 20-to-1 can be achieved. JPEG decompression is supported in PostScript level 2 and 3 RIPs. This means that smaller files can be sent across the network to the RIP which frees the sending station faster, minimizes overhead on the print server and speeds up the RIP. The downside of JPEG compression is that the algorithm is only designed for continuous tone images. JPEG not does not lend itself for images with sharp changes in tone.

     
    |
    Mark as:
  • Currently Being Moderated
    Oct 4, 2013 9:11 AM   in reply to Michael Witherell

    What I've always wondered about is why so many web sites default to jpg when gif would be much better for flat art.

     
    |
    Mark as:
  • Currently Being Moderated
    Oct 4, 2013 9:13 AM   in reply to Michael Witherell

    Makes sense.

     

    To put the OP at rest - I opened the image from the PDF in Photoshop - by using the Edit options in Acrobat.

     

    The image is opening at 49mb at the original specs noted above.

     
    |
    Mark as:
  • Currently Being Moderated
    Oct 4, 2013 9:54 AM   in reply to Eugene Tyson

    Thanks, Eugene.

     

    The ironic thing is most people complain that their PDFs are too large.

     
    |
    Mark as:
  • Currently Being Moderated
    Oct 4, 2013 10:47 AM   in reply to Amy2014

    Using PDF/X-4 the file is 2.58 mb. Using Press setting it's 2.1.

     

    I still say you're worrying about nothing.

     
    |
    Mark as:
  • Currently Being Moderated
    Oct 4, 2013 12:13 PM   in reply to Amy2014

    There is nothing lossy going on.

     

    You've been told this so many times.

     

    Get on with the job.

     

    If you're worried about your image quality then saving to a jpeg firstly was a huge mistake.

     

    InDesign is not creating a lossy version of your image - it's compressing it using all the techniques that Michael has clearly explained.

     
    |
    Mark as:
  • Currently Being Moderated
    Oct 4, 2013 4:19 PM   in reply to Amy2014

    Don't mix "rude" with "exasparation".

     

    I have taken my own time to look at your file.

     

    No further compression has happened with your file.

     

    If so worried about quality then saving as a jpeg from the start was your mistake.

     

     

    If you don't like InDesigns compression of the images then talk to your printers about preparing an FPO (For Postion Only) and send them the original JPEG.

     

    They can sway the image in the layout for your JPEG at the RIP.

     

    That will quall your worries.

     

    And you have nothing to worry about.

     

    Press on with the job.

     

    Get a chromalain proof get what you need to make the job happen the way you want.

     

     

    But don't spend any more time on this. You're wasting your time.

     

    Nothing to worry about.

     
    |
    Mark as:
  • Rob Day
    3,142 posts
    Oct 16, 2007
    Currently Being Moderated
    Oct 5, 2013 6:19 AM   in reply to Amy2014

    There's a trick you can use in Photoshop that shows the changed pixels in a compessed jpeg file.

     

    I replicated your scenario by saving a 16"x25", CMYK image as 12 JPEG out of Photoshop. Then placed it in a same sized ID page and exported to PDF/X-1a with Maximum quality compression. I got a similar change in file size 23MB vs. 7MB.

     

    If I place the exported PDF on the original PS JPEG I can then use the Difference blend mode and see any changed pixels. Like this:

     

    Screen Shot 2013-10-05 at 8.55.23 AM.png

     

    Screen Shot 2013-10-05 at 8.56.09 AM.png

     

    Screen Shot 2013-10-05 at 8.57.54 AM.png

     

    If the pixels were identical the Difference mode would show them as all black (0) and the Std. Dev. would be 0. The more compression the greater the number of non-black pixels, and the higher the Std. Dev. gray level showing in the Histogram.

     

    You can see in my test the Std. Dev. is less than half a gray level, which would be impossible to detect in print.

     
    |
    Mark as:
  • Currently Being Moderated
    Oct 5, 2013 6:44 AM   in reply to Amy2014

    Don't use JPEG in the export.

     
    |
    Mark as:
  • Rob Day
    3,142 posts
    Oct 16, 2007
    Currently Being Moderated
    Oct 5, 2013 7:44 AM   in reply to Amy2014

    is concerned I'm over-compressing when exporting

     

    It looks to me like PDF simply has a better compression algorithm and you want to take advantage of it. You could ask the photographer to provide .psd so there's only one round of compression, but there's no press condition where half a gray level would be visible—I'm sure you would have variations larger than that within the press run.

     
    |
    Mark as:
  • Currently Being Moderated
    Oct 5, 2013 8:20 AM   in reply to Amy2014

    Then why is he sending you JPEGs? Seems a bit crazy to be sending those

    and then worrying about what JPG does to a photo.

     
    |
    Mark as:
  • Currently Being Moderated
    Oct 6, 2013 5:26 AM   in reply to Michael Witherell

    (Referring to Mike Witherell., #19)

     

    Mike, your description is correct for RGB images. Here are some additional illustrations:

    http://docs-hoffmann.de/jpeg131200.pdf

     

    For CMYK images a decomposition Luminance-Chrominance (Y-CbCr) would destroy

    the black generation. The problem can be overcome by compressing each channel

    C,M,Y,K  independently as a grayscale image.

     

    Edited:

    Very little reliable information can be found. It seems, for this application the color space

    YCCK had been invented, which is a combination of Y-CbCr plus a single channel for K.

    Compressing the first part as usual and the second as a single grayscale should work,

    preserving the black generation.

    http://software.intel.com/sites/products/documentation/hpc/ipp/ippi/ip pi_ch6/ch6_color_models.html

     

    Of course these issues are not really relevant for the actual thread.

     

    Best regards --Gernot Hoffmann

     

    There is a discussion, initiated by Amy, in the Photoshop Forum as well:

    http://forums.adobe.com/message/5738933

     
    |
    Mark as:
  • Currently Being Moderated
    Oct 6, 2013 2:06 PM   in reply to G.Hoffmann

    Gernot,

     

    I had this summary in my Photoshop ppt of notes now already for many years. Probably originally I was quoting from your writings, but I don't remember. I should acknowledge that I am not the originator of these thoughts. I just share them in the classroom.

     
    |
    Mark as:
  • Currently Being Moderated
    Oct 6, 2013 4:10 PM   in reply to Amy2014

    Amy2014 wrote:

     

    This has got me completely and utterly baffled.

     

    I have an image that I saved using Photoshop as JPEG at max quality and on disk it is 12.3mb.

     

    Your disk image is not JPEG-compressed; it's a TIFF file with an EXIF wrapper. The large difference is because (1) the original image is *not* compressed, and (2) in the PDF it is *highly* compressed (with a slight loss of data). And (3) there is a slight size reduction -- 3900 x 2650 in the original, 3765 x 2513 in the PDF, so it's also 8.5% smaller (admittedly, not as significant as pts 1 and 2 combined).

     

    Inside the PDF it is a JPEG, as confirmed by Acrobat Preflight -> Inventory. This shows it uses "JPEG compression (DCTDecode)".

     

    For the original image, Apple's Preview (of all!) shows a 'TIFF information' field and says it's an uncompressed (!!) TIFF image. Examining the original file with a hex editor shows that this appears to be correct. (*)

     

    You or your photographer presumably saved this from within Photoshop as a JPEG, but set the dialog options in such a way PS decided to save it in this format. Hold on while I check ... Hmm. By default, saving as TIFF preserves the original compression, and when saving it indeed says "None" in the Image Compression field. I wonder if the presence of EXIF information made PS work this way. Maybe someone on the PS forum can tell you that.

     


     

    (*) Maybe not entirely true. For CMYK, the raw size would be 4 x 3900 x 2650 ~ 40 MB. If pressed, I could probably find out why this image is still only a quarter of that, but I think the main point is clear enough.

     
    |
    Mark as:
  • Currently Being Moderated
    Oct 7, 2013 12:11 AM   in reply to [Jongware]

    Jongware,

     

    I had confirmed Amy's workflow with my own image. Data are here in #3:

    http://forums.adobe.com/message/5740039

     

    In my opinion, the JPEG compression settings in InDesign and Photoshop

    are different.

    It can be assumed that PhS qual.12 doesn't use color subsampling

    but ID does. Color subsampling*) shrinks the file size for an RGB image

    already by factor 1/2. For CMYK, which contains RGB=3/4 plus K=1/4

    size the compressed size will be (1/2)·(3/4) + 1/4 = 5/8.

     

    All the other compression effects are additional.

     

    Furtheron it's very important to know, whether the CMYK profile (ca.2MB)

    is embedded or not. In my example this is clearly mentioned.

     

    I don't think that InDesign has an "improved" JPEG compression which

    could explain  the observed effects. 

     

    Best regards --Gernot Hoffmann

     

    *) the most effective method uses only one of four value pairs Cb,Cr.

     
    |
    Mark as:
  • Currently Being Moderated
    Oct 7, 2013 3:41 AM   in reply to G.Hoffmann

    Gernot, sorry, I didn't follow the discussion all the way to the PS forum.

     

    Nevertheless, my main point:

     

    I don't think that InDesign has an "improved" JPEG compression which

    could explain  the observed effects.

     

    .. the original file was not JPEG compressed (even though the file name seemed to indicate that), and so that fact in itself does explain the huge difference.

     
    |
    Mark as:
1 2 Previous Next

More Like This

  • Retrieving data ...

Bookmarked By (0)

Answers + Points = Status

  • 10 points awarded for Correct Answers
  • 5 points awarded for Helpful Answers
  • 10,000+ points
  • 1,001-10,000 points
  • 501-1,000 points
  • 5-500 points