Here is a JPG image (unprocessed) followed by the corresponding DNG that I post-processed in Photivo/GIMP. (Canon Powershot SD750)

Flip between the two images in Chrome. Why is the raw image slightly larger as well as softer?

  • \$\begingroup\$ Note: I did denoise the latter, but when I applied the same filters to the JPG, it was still sharper. \$\endgroup\$ Commented Jul 15, 2014 at 4:32
  • \$\begingroup\$ Are you asking if RAW is smaller or larger? BTW, after you've edited/post-processed a RAW file, it then compresses it into a jpg afterwards anyway. \$\endgroup\$
    – BBking
    Commented Jul 15, 2014 at 5:10
  • \$\begingroup\$ @BBking oops. I actually meant larger. I'll go fix the title. Also, I did compress it heartbreakingly into a humble JPG, but that alone couldn't explain either phenomenon. \$\endgroup\$ Commented Jul 15, 2014 at 5:18
  • \$\begingroup\$ possible duplicate of Good examples of RAW's advantages over JPEG? \$\endgroup\$
    – Jez'r 570
    Commented Jul 15, 2014 at 6:37
  • \$\begingroup\$ possible duplicate of What is RAW, technically? \$\endgroup\$
    – mattdm
    Commented Jul 15, 2014 at 8:08

1 Answer 1


Of course I don't know. :)

However the most common causes for these things are:

  • RAW is "softer"
    1. The RAW image isn't 'softer' - JPEG image is doctored in-camera usually with quite aggressive sharpening. Especially on entry-level cameras this sharpening is 'yelling' sometimes and can cause artifacts
    2. Besides sharpening, JPEGs usually have in camera some curves applied to increase contrast and 'push' having as a side effect a perception of increased sharpening.
    3. Noise reduction applied in post-processing can make the image look softer because (usually) removing noise means removing detail.
    4. Psychological effect of more light. A lighter image feels/appear sharper than a darker one. You can play with a properly exposed photo by moving the 'Exposure' slider to make it a little darker and after this a little lighter. The first one is more "flat", the latter one has "more" detail.
    5. Also, depending on current camera program ('Vivid' etc.) there are also other Color manipulations like Saturation and/or Vibrance boost (and others) which also give the impression of 'more sharper/pop'. (HDR I'm looking at you...)

In your concrete case, I think that at least the first 4 points above apply.

  • JPEG has different dimensions than RAW

Yes, known fact. And more: it differs depending on what RAW conversion engine you use. See here:


It has to do mostly with how the demosaicing engine treats the near-edge R, G or B pixels in order to compose the actual picture.

For reference, Photivo uses DCRaw.

Also, some JPEG engines truncate the dimensions of the image at multiples of 8, other ones do not.

So it isn't a surprise that the dimensions are different. I know, it doesn't feel natural, but this is the way it is. Perhaps when we'll get rid of the classical single color (Bayer/TransX etc) sensors we'll get rid of these oddities as well.

  • \$\begingroup\$ I'm not sure I agree with the last sentence. AFAIK, the reason why JPEGS are cropped to multiples of 8 is because of the compression algorithm which works on square portions of the image. \$\endgroup\$
    – clabacchio
    Commented Jul 15, 2014 at 11:48
  • \$\begingroup\$ @clabacchio The JPEG standard allows for incomplete blocks along the bottom and/or right edges of the image, meaning image dimensions do not have to be multiples of 8. Additionally I know of no image editor that crops images to a multiple of 8 before saving. RAW converters may aim for multiples of 8 in order to produce JPEG images which can be rotated without re-encoding (a desirable property of camera generated JPEGs). \$\endgroup\$
    – Matt Grum
    Commented Jul 15, 2014 at 13:23
  • \$\begingroup\$ @MattGrum Sure, Jpeg supports other sizes than just multiples of 8, but it seems a better reason for rounding than the Bayermatrix, as (AFAIK again) it's a sub-pixel division. \$\endgroup\$
    – clabacchio
    Commented Jul 15, 2014 at 14:41
  • 1
    \$\begingroup\$ @clabacchio I think the real reason is, as stated in the dcraw documentation: "Any algorithm that combines each pixel with its neighbors is going to have problems near the edges. C code is cheap, so dcraw applies a different algorithm to edge pixels. Hardware logic is expensive, so cameras crop off the edge pixels after processing" \$\endgroup\$
    – Matt Grum
    Commented Jul 15, 2014 at 15:06

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.