 |
| FZ80 f4 20mm hand held ISO 80 One forum member thought the grain in this photo to be simply unbearable. Yet when printed out even at large size it is barely visible. |
Mirrorless digital cameras like the FZ80 produce two kinds of noise.
The first is audible noise consisting of various artificial sounds in the form of electronic beeps and the real sounds of the stabiliser, mechanical shutter, focus motor and zoom motor working.
By the way the FZ80 can be configured to operate very quietly by switching off those beeps.
The second is digital noise which is visible in images. This is the “noise” component of the signal to noise ratio produced by the sensor.
There are two kinds of digital noise. The first is chroma (color) noise the second is luminance noise.
Chroma noise has been very successfully suppressed in JPG files from modern cameras and in RAW converters with RAW files.
So I will be mostly talking about luminance noise which appears as grain in RAW files and either grain or some form of speckles in JPG files.
The FZ80 uses an 18 Mpx sensor with a diagonal measurement of about 7.67mm which has only a slightly greater area than one of the buttons on the back of the camera. How they fit all those millions of photo sensitive sites on there I have no idea. But amazingly they do and the results are pretty good.
However each of those photo sites is microscopically small and that means images from these small sensor cameras have more digital noise (lower signal to noise ratio) at any given ISO sensitivity setting than pictures from cameras with a larger sensor.
A few days ago I posted a photo on the Digital Photography Review Panasonic Compact Camera forum. One respondent said he felt the level of grain present in this photo was “simply unbearable” and that he could not possibly consider using this camera.
Of course every individual has to make his or her personal judgement about these things and that is fine.
But now I will tell you something very interesting. I printed up this image on my Epson 4880 machine to an actual image size of 410 x 540 mm. This is quite a large print by any standards.
It looks strong on the wall with good color and acutance. The grain is only just visibleif I get up close and peer at the print with my spectacles on.
This leads me to two discoveries of great importance to the evaluation of grain in photographs.
1. Grain is much more obvious in images viewed at 100% on a large, hi res monitor screen than it is in a print, even a large print. I think this is a manifestation of a fundamental difference in the way grain appears in the two viewing media.
The corollary of this is that an image which may appear “unbearably” grainy on a monitor at 100% can look perfectly fine when printed out.
2. I find that a rough but useful guide to evaluating on a high quality monitor how any image will look in a print is to view it at 50%. If the image looks good at 50% on screen it will look good in a print at any size.
Further to the statement above I want to share with you another discovery which I have made. On camera review sites and user forums I often see a statement that [camera x] can produce prints up to [some stated size]. My experience over the years does not lead me to support this notion at all.
I find that if an image is sharp and clear to start with and good upsizing strategies and printing technique are used it will print up to pretty much any size and still look good at normal viewing distance.
When I look at pictures made with the FZ80 I am reminded of my many years in film photography when I mostly used 35mm SLRs.
Prints from the FZ80 have a level of detail, acutance and grain very like those made from 35mm film with a good quality prime lens.
That level of imaging capability was good enough for me and millions of others for about 50 years and it is still good enough for my purposes.
Of course in the film era there was no such thing as an amazing 20-1200mm lens or the miraculous ability to change film speed with the turn of a dial.
Of course you could argue that in the digital era many cameras deliver better picture quality than was (or is) available from 35mm film. That is true enough in the technical sense but you have to ask what is the purpose to which this better quality will be put ? And if the answer is online sharing or small prints then all that quality is not necessary, in fact redundant.
Summary
Small sensor cameras including the FZ80 produce images with more visible grain at any given ISO setting than camera with larger sensors.
Whether that level of grain is acceptable to any individual photographer is partly a matter of personal preference.
This may be adversely affected by too much pixel peeping at 100% on hi res monitors.
I take the view that more relevant issues are the photographer’s skill in managing grain at the capture phase of use and also in the post capture phases.
I will be posting more about this shortly.
 |
| Another arcade. A bit less light here so I used ISO 320. With thoughtful work in Adobe Camera Raw the grain is well controlled I think |
Cameras compared
A question often asked on user forums and one which has exercised my interest is whether any of the cameras which use the same sized 7.67mm diagonal sensor have less apparent digital noise than others.
The conventional wisdom often stated on user forums and by reviewers is that sensors with fewer pixels will have less digital noise than those with more pixels.
I own and have tested cameras each with the 7.67mm sensor size but having 12, 16, 18 and 20 Mpx.
As far as I am aware all these sensors are made by Sony although, unfortunately neither the camera makers nor Sony care to confirm this for reasons unknown to me.
Test method
Cameras: Panasonic---FZ300 (12Mpx), FZ70 (16 Mpx), FZ80 and TZ80 (18Mpx), Nikon ---B700 (20 Mpx).
Test subject: This is books on shelves with a page of newspaper to judge resolution. The camera is on a tripod, and the subject is photographed with RAW capture at ISO settings from 100 to 1600. The same focal length and lens aperture is used throughout.
Preparation of the files: In order to compare like with like the files must be rendered comparable in size, color balance and lightness.
To equalise the size one could upsize the smaller files or downsize the larger ones. The usual practice is the latter so I do that. I then adjust color balance and lightness (which by the way vary considerably from one camera to the next) so they are as nearly the same as I can make them.
Next in the Detail tab in Adobe Camera Raw I adjust the Amount, Luminance and Color sliders to zero.
Viewing the files Now I open all the files in Photoshop and view them side by side at 100% on screen, using Window>Arrange> Float all in windows.
They are all the same size, color and lightness so can be meaningfully compared.
 |
| The white faced heron is one of the few Australian birds which will stand perfectly still while hunting for breakfast. This one was in deep shade early in the morning. It let me get fairly close but I still needed 800mm on the lens, 1/80 second and ISO 1600. This is the full frame. RAW capture processed in Adobe Camera Raw and Photoshop. You can see there is quite a bit of grain but for ISO 1600 I think this is more than acceptable. |
Results
Color Noise: I found that the 20 Mpx files from the Nikon B700 had the least color noise with the others being all about the same. This finding is of somewhat academic interest however as I could easily remove color noise from all the files with the [Color] slider. The B700 just needed less adjustment than the others.
After removing color noise I went on to evaluate for sharpness and luminance noise.
Resolution/sharpness: Files from the FZ300 and TZ80 had slightly less resolution than the others but I had to look carefully at 100% to pick it.
Luminance noise: The B700 produced just faintly less luminance noise (grain) than the others but the difference was so slight it would not be detectable in ordinary photography.
The relationship between the cameras was the same across the ISO spectrum from 100-1600.
Conclusion
For all practical purposes there is no difference between the imaging performance of the cameras tested.
Two of the cameras had just slightly less sharpness/resolution than the others but I suspect from other (lens) tests that this is mainly due to differences in lens sharpness.
The conventional wisdom that fewer pixels produces less noise is not supported by my tests.
Actually the camera with the most pixels (B700) produced less color noise and just faintly detectably less luminance noise than the others.
Summary
Fewer pixels do not produce less noise.
More pixels do not deliver more resolution.