300dpi

Hi

Let me see if I can explain my previous post better.

1. In way of explanation for those people from countries that don't uses inches, 1 inch = 2.54 cm.

2. If you are making images purely for viewing on screens then ppi (which is metadata i.e. a value stored alongside the image) is totally irrelevant. All that matters is the size of the image in pixels. So let's say we have a 6000 x 4000 pixel image. Different screens will display that image at different sizes based on the physical size of the screen and how many physical screen pixels make up those screens. A 24 inch 4K screen will display the same image using the same number of pixels at a smaller physical size than a 48 inch 4K screen. The ppi data is ignored so you could make that value 1 ppi or 1000 ppi it will make no difference on the screen.

3. For printing, as stated by D Fosse, dpi and ppi are not necessarily the same value. Think of it simply -  pixels are the smallest unit of picture information that make up a rasterised image. So our image above has 6000 x 4000 pixels. The printer however can often print those pixels using even smaller dots. So a group of dots is put on the paper for each image pixel. Therefore dots per inch (dpi) can be higher than pixels per inch (ppi). You will not see a control to enter dots per inch (dpi) in an image editor such as Photoshop, at least not one that makes proper use of the terms. You may see a way to change dpi inside the printer driver but it is often labelled differently anyway, e.g. an Epson inkjet driver has a quality control which is actually setting the dots per inch value, it is just not labelled as that.
What you do have in Photoshop is a value for how many pixels should be used to make 1 inch (or 2.54 cm) of printed output. This is the pixels per inch (ppi) value. If such a value was not sent to the printer then the printer would have no way of calculating how big the physical print should be.

An example :

Our 6000 x 4000 image could be printed at 300 pixels per inch. That will print 20 inches x 13.3 inches ( 50.8 cm x 33.9cm).

We could also print the same image at 100 pixels per inch, without any resampling, so it still contains 6000 x 4000 pixels. This time it would print 60 inches x 40 inches which is much larger. But the larger print has no more picture information than the smaller print. 

In addition, a bit of trigonometry tells us that the first, smaller, print viewed from 22.9 inches will look identical in every way to the second, larger, print viewed from 68.7 inches away. That is why we do not need massive pixel counts in large images viewed from a distance.

Of course we can resample an image so that the larger print could be printed at 300ppi. That would then require 18,000 x 12,000 pixels and resampling has to 'invent' those extra 192 million pixels based on the known image pixels of which there were 24 million. It cannot add extra image detail, it can only make the additional pixels up based on the existing picture information. There are various algorithms to do this and all introduce artifacts. The better AI based algorithms can do a good job, but even they are still only making up the missing information. So why resample at all? In some circumstances it may be better to make up those intermediate pixels than to see the pixel structure when viewing the larger image from a close distance. That balance between introducing artifacts and the possibility of seeing pixels when viewed close is an artistic choice.

I hope this helps explain my previous answer

Dave