| Articles |

All 5D JPGs Are Beautiful

This is a subject of great interest to me, because I shoot JPGs exclusively. My feeling is that the JPG format of the Konica Minolta 5D (and 7D) is much more capable than many people realize - even some advanced and otherwise well-informed shooters.

Understand that I'm not talking RAW vs. JPG at this time. There are many significant differences between RAW and JPG capabilities and workflows, and that's a different discussion. I'm also not discussing post-processing or archiving JPGs. Again, that's a different topic.


It's my observation that the 5D options of Standard, Fine, and Extra Fine correspond to three of the highest-quality settings available in the JPG format. The same settings in Photoshop are labeled as 10, 11, and 12. It's easy to check this by comparing the sizes of the original camera files against re-saved versions of those files at various quality levels. Although I actually have standardized on Fine (11) for my own work, my tests involve only Standard (10) and Extra Fine (12) settings in order to show the widest range of variation.

Here's a test shot of my SpongeBob SquarePants lunchbox. Below are Standard and Extra Fine cropped versions at 100% magnification (each pixel from the image equals one screen pixel).
I'm not going to tell you which is which, and I'm sure you'll see no visible difference at this level:

 Now here are smaller sections of the same two images at 400%. The one on the left is Extra Fine, and the one on the right is Standard:

These examples show some discernable differences, but they are very subtle. I've looked all over these images (and many others) and I almost always find the difference between Extra Fine and Standard to be inconsequential. I would feel comfortable enough shooting everything at Standard; but just to be on the safe side I shoot in the middle, at Fine quality. Since I use 1MB memory cards, I can fit about 330 Fine images on each, as opposed to half as many Extra Fine ones. To me it's the perfect 'sweet spot' of quality versus file size.

How to Make a Not-So-Beautiful JPG

As long as we're looking at JPGs, just for fun let's try a different kind of side-by-side comparison at 100%. What I've done this time is to place another section of that same original Extra Fine image next to a rather 'abused' version of itself. I did a no-no. I loaded the original in Photoshop and re-saved it as another JPG. Not an Extra Fine (12), but only a Standard (10). Not once, not five times, but nine times. That's right. Saved, closed, re-loaded and re-saved as a Standard JPG nine times. Have a look. Original on the left, re-saved as Standard nine times on the right:

Can you see a difference? I think I can, in a couple of places, but I'm not sure. So here's a closer view of SpongeBob's cap:

These are 500% screen magnifications where individual pixels are clearly visible. Now there is something to see. If you look carefully at that 10th generation image on the right, you can just make out where some odd squares are beginning to form. These 8x8 squares are the building blocks of JPG images, and you'll only encounter them with very high compression levels and/or repeated generation loss. To me, it looks like the squares are taking shape primarily in areas of smooth tonality. The boundaries between light and dark areas are not as noticeably affected.

That's why I think my test scene is a good one. It has some solid blacks, some blown-out whites, sharp edges and details, and many colors and tonal ranges. I've looked for nastiness all over this image at high magnification, and if it's visible at all, most of it is showing up in those smooth tonal areas.

So this shows the worst-case degradation moving from a nice 5D Extra Fine JPG to a repeatedly bludgeoned Standard version. I don't think it's so bad, considering what it's been through!

For further reading, see this detailed analysis of JPG quality, where the author states that the 5D/7D files are among the highest quality JPGs he has tested from any source. And while you're there you should look around the rest of the site as well. It's very informative:

A few additional comments are in order. First, my observations are based on my own 5D images and careful use of Photoshop 7. I can't predict how some other camera, or software, or bad user techniques might mangle JPGs.

Second, obviously, you don't want to make it a habit to save, re-load, and re-save JPGs unless you're exploring the limits as I've been doing. It's easy to wreak havoc on images this way if you make bad choices. On the other hand, don't be terrified to experiment as long as you keep the originals safe - you might be surprised. I have post-processed JPGs in Photoshop and re-saved them more than once, and lived to tell the tale.

Third, there's still more to explore. Rotating or cropping an original or changing its aspect ratio, and then re-saving it as a JPG, can sometimes introduce additional problems as well. This is due to that 8x8 building block thing mentioned above. I'll probably end up testing that too.

 Text and images 2007 Ray Lemieux