export quality

[zhaopian]

I am new to Affinity.  I want to export my picture to jpeg file, and I want to keep the file size as less than 500k with the highest picture quality.  

I have two options:

one is at size 1800px by 1200 px, quality 79%;

the other is 1200 px by 800 px, with quality 91%.  They yield files with similar sizes: 490k and 494k.

When they are viewed on the screen after resizing to the same size, the 1800 px by 1200 px one likes better.  Can you tell me why?  Is this always the case.  What is the best combination of size and quality under the file size constraint?  I have attached both files.  Thank you for your help in advance.

 

 

[Oval]

Welcome here.

 

The 2.16 MPixel means much more details (informations) than your 0.96 MPixel.

 

The picture has big areas with low contrasts. 79% and 91% have no big negative effect on this picture.

 

There is no mathematical way to get the best combination because software does not know our needs. For example, for some people the second picture is too sharp or unnatural. For others it is ideal because they want to output on an old screen.

[R C-R]

In essence, JPEG compression reduces file size by reducing the number of different color & intensity values of the pixels in the image to a smaller number of values, which makes the compression algorithm more efficient than if it had to preserve all those values. (For a more technical explanation of this process see for example this web page.)

 

A consequence of this is that the quality/file size trade off depends greatly on how much variation of these values there is in the image to begin with -- images with fewer variations can be compressed more without appreciable loss in quality than those with more variations.

 

This means the same quality setting will produce different file sizes for different images & conversely that for a desired file size the quality will be different for different images. In other words, some images are just naturally more compressible than others.

 

So for example, if you resize an image in some way that reduces its variation in pixel values, even just in some visually unimportant parts of the image, you will make it more compressible. How best to do this is not always obvious & may even be counter-intuitive. For instance, blurring the background may actually increase the number of different colors in it or not reduce them appreciably, but doing that plus posterizing the background using a relatively small number of levels may work. Likewise, the resampling algorithm used can make a difference, but the important thing to understand is that what works best will be different for different images.

[zhaopian]

Thank you very much for your help.  Now I understand that the size control number of px, and the quality control the intensity of the signal at each point.  Is this correct?  Currently, I keep quality at 80%, then use the maximum pixel size allowed to make file size under 500k (some websites only accept files under 500k).  Or for a 4k monitor (4096 X 2160 px), Should I keep the shorter side of the picture at 2160 px, and use the maximum quality allowed to make file size under 500k?  Or for each picture, I have to try different combinations to find the best balance between pixel size and quality?

[R C-R]

For each picture, you will have to try different combinations to find the best balance.

 

To better understand why, consider two photos, each with for example 2.5 million pixels. Photo A might have only 100 thousand different colors in its 2.5 million pixels, while Photo B might have five or ten times as many different colors. But Photo B might have a lot of colors that are almost the same, so reducing that to say 10 thousand different colors may be much less noticeable than for Photo A if a lot of its colors are distinctly different.

 

It is actually more complicated than that because it also depends on how the colors are distributed throughout the image, but the principle is the same.