No announcement yet.

Batch resize behaviour

  • Filter
  • Time
  • Show
Clear All
new posts

  • Batch resize behaviour

    I am working on an image from a tiff file and have saved it as a 100% jpeg file; it's 29.8 mb. When I resize it with Batch Resize, to 99%, it's 3.9 mb. Why?

  • #2
    99% is not 100% - as soon as you reduce from 100%, you see a reduction in file size. It's optimised.


    • #3
      Originally posted by Bruce Whittington View Post
      I am working on an image from a tiff file and have saved it as a 100% jpeg file; it's 29.8 mb. When I resize it with Batch Resize, to 99%, it's 3.9 mb. Why?
      The key to this (at least in the Ultimate 2023 version I'm using) is likely in the Batch resize options.

      The default quality setting in Batch Resize in Ultimate 2023 is 90%. I don't know what it is in the Mac version, but since you didn't say what quality setting you used, I will assume it is 90% in the Mac version also.
      If I start with a TIFF image that is 7952 x 5304 pixels, and export it without changing its size as a JPG with 100% quality, that image is 7952 x 5304 pixels 24 bit color depth and the file size is 23.4 MB

      If I then take that JPG image and use Batch Resize to resize it to 99% in both width and height, and the default qualify of 90%, the resultant image is 7872 x 5251 pixels, and the file size for the particular image I used is 7.4MB. (it will vary depending on the content of the image) A reduction of 10% in quality adds significant compression, and possibly artifacts. There are some other settings that affect the file size, the type of compression algorithm selected, and whether or not you opt to maintain the metadata but the changes they make are not as significant.


      • #4
        Thanks for the replies. First, in the Mac version I can't find a way in Batch Resize to specify quality - I'd be happy if someone would correct me. I've learned that jpegs remove a lot of data even in a modest reduction. I tried some things, using the same 41 mb .nef file:
        "Save as" jpeg at max on slider (no percentage is given) = 27.4 mb.
        "Batch resize" that jpeg at 99% = 3.5 mb.
        Using the original .nef file, "save as" jpeg at 18/20 on the scale (90%?) = 5.2 mb (so the 10% difference loses a further 22 mb - that's remarkable to me)
        "Batch resize" that jpeg at 99% = 3.4 mb.
        It appears that the jpeg function saves the most in the first saves, and less in subsequent saves, but the difference in the two examples baffles me:
        In the first case, a 100% jpeg is followed by a 90% jpeg, resulting in a 3.5 mb file.
        In the second case the same file, with a 90% jpeg followed by a 90% jpeg, the result is a 3.4 mb file.
        And just to confuse things, I tried the same .nef file in Affinity Photo. The first "export as" jpeg 100% resulted in 14.9 mb, and the second export at 90% resulted in 2.8 mb. I am not sure what to make of it! I'll experiment some more.


        • #5
          Bruce Whittington


          First the basics. A RGB 24 bit color depth image has three bytes per pixel. If you multiply the width by the height in pixels, and then by 3, you have the total number of bytes for the uncompressed size..

          So the uncompressed size of a 4032 x 3024 pixel image with a color depth of 24 bits is 36,578,304 bytes, or approx 34.9MB.

          Now by using one of a number of possible algorithms, the information in the uncompressed image can be compressed when saving the file. Note that the amount of compression will vary from image to image depending on the content of the image.

          Some of those compression methods are said to be non lossy, but the the common JPG format uses lossy compression. That means that when the file is decompressed, some of the information is no longer present.

          At the time of the file save, the amount of compression can be increased by reducing the quality, but it is important to note that the max 100 quality scale does not mean there is no compression. For example using a randomly chosen 4032 x 3024 pixel 24 bit image which has an uncompressed size of 34.9MB, when it was exported as a JPG from ACDSee Ultimate 2023 with quality set to 100, the file size was 9.6MB. When the same (original) image was exported as a JPG with a quality setting of 95, the file size was 7.0MB, and when exported as a JPG at 90% the file size was 5.9MB

          There are several points to note.

          (a) When a JPG is re-saved as a JPG, the loss increases. That loss is known as generation loss.

          (b) The same setting in the quality scale in different applications should not be expected to produce exactly the same result. The way I look at it is that a qualify setting of 100 is simply the best quality a particular application is going to produce from the algorithm it is using for compression.

          For example, when the image above was saved from Affinity Photo 2 as a JPG with Quality setting of 100 and using Bicubic, the file size was 10.6MB.
          When exported at 90, using Bicubic, the file size was 4,6MB

          (c) Embedding metadata and embedding ICC color profiles can increase the size of the file.

          Hope that helps.
          Last edited by Greyfox; 01-20-2023, 08:10 PM.


          • #6
            Thanks again for your time on this. I do understand how jpeg compression works, but you have clarified some points for me. It's why I have shot RAW for years, and often save a modified image as tiff until I am ready to jpeg the final version. I remain confused about why the compression seems so inconsistent in the two examples I gave (in ACDSee), but they both involve two jpeg steps, which I should be avoiding anyway! I am interested to look more closely at whether there is any visible difference in the quality of jpegs from ACDSee and Affinity, given the resulting file size differences. It may well be worth using different software for converting to jpeg in different situations. A good day when you learn something . . .