File size / Metadata

[couman]

Also, consider what is meant by "image quality". I may be wrong, but I don't believe that the JPEG working group ever defined the term. That leaves it up to individual software developers to invent their own standards -- for some, 100% means "good enough", while for others it means "the best possible". You can see this by saving the same file in different editing packages. For example, PWP's 100% is not the same as Fastone's 100%, and both are different from IrfanView's 100%.

[jsachs]

Indeed the JPEG standard is very loose - many aspects are left to the developer.

[RScholl]

Dieter, you're right, the difference function in PWP shows very clear the differences between the pictures, easier than to compare it by eye. Thanks for the tip!

With exiftool I extracted a thumbnail image from a picture: It is definitely not the thumbnail which Metadatastripper leaves as a result, if a thumbnail is included in a picture.

I understand, that the jpg standard is very loose. That might be the reason, that a 1 MB-input-image without processing is saved again by PWP with 2.5 MB. Towards a - not objectively determined - "100 %" the file size increases rapidly.

Rainer

[doug]

Years ago, when I first learned about PWP's difference function, I used it as a way to "objectively" measure the deterioration caused by multiple saves of a jpg. There was (and still is) a lot written in other forums about how terrible it was to use jpg due to its "lossy" nature.

I saved the same jpg image over and over again, changing just the file name each time but with NO intervening edits. The difference function showed no significant changes until I got to, as I can best recall, about 10 or 12 successive saves. So unless I would be re-saving more than a handful of times, I concluded (rightly or wrongly) that the "lossy" aspect of jpg was a bit overstated.

More recently I tried the same technique to compare an original tif file with the initial save as a jpg. I was surprised to see some differences in just a single change in file format.

Do these observations seem right to you, Jonathan?

[jsachs]

It is reasonable to expect a TIFF and JPEG of the same image to have differences.

[Dieter Mayr]

The major difference (or maybe better, loss) happens at the first save as jpeg, so it is most likely to see a difference between a tiff and a jpeg.
When you resave a already "jpeg-ed" picture it has already the structures of the jpeg compression in it, so less differences would be visible.

My snaphot camera produce jpegs, at the high quality setting, of about 4.5 to 5 MB, when saving them in PWP at 100% Quality setting the file size is about 6.5 to 7 MB, the same file size as from the camera i get at a setting of 96% to 97%, and at a setting of 50% I can squeeze it down to under 1 MB. And pushing the slider to the far left at 0% the file has just 600kB anymore, so the file size of a jpeg depends highly on the setting of the quality slider. If you reopen the 600kB file and save it at 100% then again it will be in the 5 to 6 MB region again.
I think it is important to understand that as soon a jpeg file in opened in a image processing software there s nothing like a jpeg compression anymore, all images are the same in memory once they are opened, no matter from what file type they come from. They are simply a bunch of data 1 byte for each x - location of a pixel, for each y - location and 1 byte for each of the 3 primary colors (red, green and blue) for each pixel (assuming a 24 bit color image). So if the picture has for simplicity of calculation 1024 x 1024 pixel it occupies exactly 3 MB of your computers memory, no matter where it comes from, and you can save it in a jpeg file of few 100kB size, but when you reopen it in your software it has the 3MB of memory usage again.

It might help to understand why a 1MB jpeg file can result in a 6MB file after resaving it.

[Staylor2]

It should be noted that once a file has been saved as a jpeg the damage has been done and restoring to full size does not bring back the data that was lost. It does return the image to the full memory size but the data will be different. Saving it again at the same compression ratio will will produce much less compression data loss because the picture details that were lost in the first compression are no longer in the file and therefore will not be compressed again. As was stated earlier, it is the first compression where the most damage is done. However, if you save a jpeg at a higher compression in a subsequent save, then more data will be lost. If as in an example above, you compress a file down to some very small size and then restore it, you will still have the full image size in memory but the image data will now be way different. Then, saving that image as a jpeg at even 100% will result in a bigger file but the image data will no longer be the same as if you had never saved it as the highly compressed smaller file size.

[mjdl]

To learn more about the contents and encoding of your JPEG images than sensible people really need to know, download and use the freeware "JPEGsnoop" JPEG analysis utility.