Downsampling reduces noise?

Started Jun 11, 2012 | Discussions thread
Senior MemberPosts: 8,364
Re: Downsampling for noise comparison of 16mp and 24mp cameras
In reply to Robsphoto, Jun 14, 2012

Robsphoto wrote:

TrojMacReady wrote:

Robsphoto wrote:

philbot wrote:

Robsphoto wrote:

I won't get into the old argument of what is an adequate ppi, but I do discuss it here:


It's a bit flawed to not take into account the actual 'information' contained within the image when determining size, otherwise your exercise is purely mathematical..

And that's the point.. pixel level vs image level, you are stuck on pixel level, Jesper is simply approaching it from the practical image level..

Arguably you are both correct in what you say, but you have to wonder which actually serves a purpose in the real world..

Well let's take a "real world" example to illustrate the point I’ve been making. Let's say that I’m a salesperson in a shop that sells the Sony range of cameras, including the A55 and A77.

A customer comes into the shop and says "I’m interested in seeing prints from images taken by both the A55 and A77, particularly ones taken in low light.”

To illustrate the difference in print quality, I might then choose an image that was taken in low light by both cameras at exactly the same time, of exactly the same subject, and using the same exposures etc.

I would be careful to print the images from the A55 and A77 at the same pixels per inch (ppi), and on the same printer. Now, if you select the same ppi (which I think it’s important to do), and if you decide to print the full-sized images from both cameras (A55 = 4912 x 3264 pixels; A77 = 6000 x 4000 pixels), then the print sizes (if you select 200ppi) are about 24.5” x 16.3” for the A55, and 30” x 20” for the A77.

I think you mean DPI.

No, I prefer ppi, but let's not argue this point as it's really not all that relevant to the overall point I'm making, but see this write-up for example:

Your printer will just get mighty confused when you give it commands involving PPI.

And you don't think the client will wonder why you're comparing differently sized pictures first? A smart one would.

To take a more extreme example, if you were comparing a print from an 8mp cell phone with a print from the A77, I think the customer would want to see the largest size print that each camera can realistically produce. To show equal sized prints from both makes little sense to me.

You'd have to differentiate between what you actually want to do. The largest possible print or the best low light capabilities. Keep a constant. If you want to show the largest print possible from any camera, you're not going to use a high ISO shot taken in very low light. And in better light, the A77 will get away with larger prints, before noise ruins the output.
In low light, they will both be limited around the same size.

I realise that you could downsize the A77 image to the same image dimensions of the A55, and this would mean that both the A55 and the A77 photos would be printed at 24.5” x 16.3”.

But, from a sales point of view, this doesn’t show off one of the reasons why you might select the A77, namely that the A77 has an image width that is about 22% greater than that of the A55.

You'd show that where cameras can actually use that resolution in general, better light. Not even the most expensive camera is at it's best producing good detail in giant prints at high ISO.

OK, but in the example I gave, the client specifically asked to see a low light print, because after reading lots of posts from this forum, this aspect was of concern to the client!

You'd make the largest print possible before the image falls apart. And if the light is low enough, the A77 and A55 will eventually shake hands in this regards. Imaging Resource produces some nice print tables with sizes at the the bottom of their reviews showing exactly that for each camera.

Now, the question arises, which print will show more noise – the 24.5” x 16.3” print from the A55, or the 30” x 20” print from the A77? I suspect the answer may be that the A77 print will show more noise than the A55 print?

I am interested in your answer to this point..........

It's a rather rethorical question that has little meaning IMHO, see above and below.

This is the “real world” situation that I as a buyer want to consider, I don’t want to look at equal sized prints from both cameras. If the A77 print of 30” x 20” is a lot “grainier” or “noisier” than the 24.5” x 16.3” print from the A55, then IMHO this is an important factor to take into account when choosing between the A55 and the A77!

Using the same logic, the ancient D40 would be the easiest to sell, never mind the small prints it makes in comparison (4 times smaller). I know it's a small print, but it's cleaner!

Possibly, but a buyer wants to be sure that the extra money paid for the A77 doesn't merely produce a lot noisier prints than prints from the A55 when they are enlarged as described earlier!

As per Imaging Resource:

Reply   Reply with quote   Complain
Post (hide subjects)Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark post MMy threads
Color scheme? Blue / Yellow