jpgQ - JPEG Quality Estimator
I often receive this or similar question: When I save JPG, with quality 75% then load it, adjust it and save again; what quality I should choose now?
We all know that JPG use lose compression, so anything you save is always in exact terms worse than original. It may not look like because of the way it is compressed but it is. And it is logical that if we later load JPG and then save again the degradation goes of course further.
Right now I can tell you the most degradation will be probably in the first round - from the original to the first JPG. If we load the JPG and save again this time the data are already somehow processed by first round so the degradation won't be as strong as first time. Of course there is quality loss on the second time and if we load and save it again and again we will soon realize the quality loss is adding up.
Loss is adding up but not as many people think - if we use twice 75% first time then 75% second time it doesn't mean that the second round is equal to 55% of original quality (75% of first 75%). There is some loss but far less. So it doesn't work like your taxes.
If we consider all this then it start to be obvious that if we ever need re-save a JPG we should use the same quality settings as a first time. Why not bigger? Because by bigger Q second time we won't gain anything at all! The file is already degraded by first round and these data will not be returned by simply setting higher quality next round. Setting the Q second time higher than first time will just make the JPG bigger, but still worse than the first.
The simple answer is: If you need to load and save a JPEG you should use the same Quality settings as was used on the JPG when it was first created.
We of course assume that we work on the same pixels. If you for example load an image of dog completely repaint it into image of sunflower then of course you can also choose any quality setting you like (and also a doctor). But this isn't the case in 99.9%
Talking now about digital cameras - our main topic: After we make a change on the image from digital camera we should save it with the same Quality settings as it was produced from the raw data inside the camera.
Now I got you!
The theory above is great but what is the quality settings your camera originally used? Or what is the quality used on image which you got from some CD, Web, newsgroup etc....? Nobody will tell you so...
..use this unique JPEG Quality Estimator!
As another gift to the large digital camera user base I decided to create a small free software applet which after you load the JPEG will estimate the quality the image was last saved. This is also the quality you should use if you load this image again in the photo editor (such PhotoBrush) and then save as JPG once again.
|Stand-Alone version ( 530 kB)
This file is without installation, just unzip and run.
you like our tools, please put a link or banner to your site. Get
the graphics here.
see even more cool tools go to our home
How this all works?
Very simple - Load any JPG and press RUN! button. The software will do some calculating and then display the values it estimates.
Value here apply to most JPEG images. This is adjusted to the standard 2-1-1 subsampling.
At this point you may stop reading if you don't need to know any more details.
This is mostly for PhotoBrush. If the JPG you are testing was saved in PhotoBrush with Enhanced Sampling (set on by default) then you should look at this number. Also some cameras use this sampling type for their Super Fine images.
The bottom quality number is always lower than the Standard 2-1-1 one. That's simple - because Enhanced Sampling produce images with better quality, the 86% Enhanced Sampling has as much information as Standard JPEG in 88%.
Ok, what number to use?
If you are sure that the image you checking was previously saved with Enhanced Sampling then use the bottom number and also set the Enhanced Sampling in Photo-Brush next time you saving.
If you don't know this (image from camera) or don't use Photo-Brush then simply consider the Standard JPEG number as the one. It is always safer to choose the higher number anyway.
If the Standard JPEG number jump over 90, this is a good indication that some kind of Enhanced Sampling was probably used.
You don't have to worry too much about this number. The value shows how close the estimated number is to the integer. For example on 86% and 0.45 tells that it is estimated 86.47 % and 86% and -0.20 tells the estimated value was 85.80%. The number is almost always other than 0 because of the recompression loss.
The software is fairy accurate in most cases I tested. However I tried only images from digital camera and files from my other software PhotoBrush and Real-Draw. I assume that your software has JPEG encoding based on code from IJG (independent jpeg group) or similar.
It never change or save anything so you can run any file through it.
You don't have to test every file from your digital camera - just few for each settings (fine, super fine etc.) to see if the quality is consistent for each setting and then simply remember this number.
As always if you remember not to rewrite the original image then if something doesn't look good you can always load the original in your editor again and try higher quality.
For photography the reasonable quality is never below 72%, more likely around 86%.
On the web the images often uses lower quality, it could get even close to 50-60% - just to cut the file size down.