I just got this new Panasonic Lumix FZ18 digicam. It’s amazing what under $300 will buy. This is a “bridge” camera, with a built in 18X optical zoom lens.
But what they say about the three different kinds of zoom are confusing. Here are a few lines excerpted from the manual:
"Taking Pictures with the Zoom
Using the Optical Zoom
You can make people and objects appear closer and landscapes can be recorded in wide angle with the 18X optical zoom. (35 mm film camera equivalent: 28 mm to 504 mm)
Using the Extended Optical Zoom (EZ)
This camera has a 18X optical zoom. However, if the picture size is not set to the highest setting for each aspect ratio, a maximum of 28.7X optical zoom can be achieved with no picture quality deterioration.
Extended optical zoom mechanism
When you set the picture size to 3M (3 million pixels), the 8.1 million pixels CCD area is cropped to the center 3 million pixels area, allowing a picture with a higher zoom effect.
Using the Digital Zoom
Further extending the zoom
A maximum of 72X zoom can be achieved with the 18X optical zoom and the 4X digital zoom when D.ZOOM is set to ON in the REC mode menu. However, if a picture size that can use extended optical zoom is selected, a maximum of 115X zoom can be achieved with the 28.7X extended optical zoom and the 4X digital zoom."
I have always understood the difference between “optical zoom” and “digital zoom”. It’s intriguing that in this information age “digital” has become synonymous with “fake”, but that’s another thread. Are they claiming that the “Extended Optical Zoom” and the “Digital Zoom” aren’t both just in-camera cropping?
I’m experimenting but am curious what the Doper take on this is, whether from those who have used this camera or its sisters, or from those who can read.
If I’m reading correctly, “Extended Optical Zoom” is a 3MP crop of an 8MP image. “Digital Zoom” would be a crop upsampled to fill 8MP.
edit: I should add, what they mean is that if you’re using a smaller image mode (like 3MP), you’re not just limited to using the full 8MP and then downsampling the image to 3MP. The “extended” zoom allows you to take a 3MP crop of the sensor when you’re using a smaller image mode, rather than forcing you to take an image on all 8MP of the sensor and then sizing it down to 3MP.
Also, there’s no good reason for you to be using the smaller resolutions on your camera when memory is so cheap these days. Use optical zoom as your benchmark.
So the two ways they might implement Extended Optical Zoom in their example where you’re only saving 3MP anyway are:
1X using all 8MP and no optical zoom, then downsampling
1.6X using 3MP and no optical zoom
…
28.7X using 3MP, and 18X optical zoom
or
1X using all 8MP and no optical zoom, then downsampling
…
18X using all 8MP and 18X optical zoom, then downsampling
28.7X using 3MP, and 18X optical zoom
where the … are the range covered by the optical zoom for a fixed number of pixels used. If I had to guess from what they wrote, they’re doing it the first way, but that would depend on the person writing the manual being someone who knows how it works.
Yes, extended optical zoom is doing exactly what you would do on a computer by cropping the image. It can be useful if you don’t want to spend the time on the computer doing the same thing or if you want to get a view of your subject that you can’t quite get with real optical zoom. Yes, it’s mostly a gimmick but not completely. I always prefer to do my cropping in the computer because I have more control and PhotoShop does a better job then my camera.
>Yes, extended optical zoom is doing exactly what you would do on a computer
Telemark, I want to know whether “Extended Optical Zoom” and “Digital Zoom” are not both just in-camera cropping. I already think in-camera cropping is not useful unless its disadvantages are outweighed by memory worries, which these days is pretty unlikely.
Do you know what the difference is between “Extended Optical Zoom” and “Digital Zoom”?
I have experimented a very little bit. When I set the image size to less than “8M” (eight megapixels), the zoom control appears to go smoothly from the same wide angle to a longer telephoto, and the zoom multiplier in the display and the little rectangular bar (looks like a “progress bar”) both go smoothly from “1X” (meaning wide angle) to “27X” or whatever. Whether or not I have chosen a size smaller than “8M”, if I enable “Digital Zoom”, the little rectangle has an extra region on the right and the transition to that region has a little pause, all of which seems like other digicams I have used. So, the user experience playing with controls and looking through the viewfinder or at the screen is that “Extended Optical Zoom” feels like optical zoom but goes further with smaller image sizes, and “Digital Zoom” feels like it does on other digicams. I haven’t experimented with the images themselves, to study their size and resolution and whatnot.
I’ve answered that question above. Extended Optical Zoom is a 3MP crop. There are no interpolated pixels, as the file size itself is much smaller. Digital Zoom is a crop that’s interpolated to the full 8MP size.
I do not know 100% about the nomenclature on that specific camera, but I work in the industry and I hope I can shed some light on the subject.
It sounds like “Extended Optical Zoom” is an option to make the optical zoom range higher by selecting a smaller image size. Most cameras give you the option of generating a smaller image size. At 1.0x optical zoom this is accomplished by taking the full frame, and scaling it down to the desired size. To accomplish a higher optical zoom when desired image size is smaller than the sensor frame size you can do a combination of
[list=a]
[li]Moving the lens to zoom in with the lens[/li][li]Crop some of the image off digitally[/li][/list]
before down scaling to the desired size. Extended optical zoom takes advantage of (b) to give you a seemingly longer zoom range with higher quality but at a lower output resolution. Doing this would generally indicate that they are confident in either the quality of their lens or their various distortion/shading correction algorithms.
Digital zoom generally refers to cropping of the image and then upscaling to the desired size. For consumer/prosumer digital cameras that do not save RAW, it can theoretically be better on the camera than scaling you could do in post processing since the scaling can be done in an optimum sequence of their image processing filters. I wouldn’t call it ‘fake’ – mostly useless, but it’s not fake.
Bottom line, it’s nothing you can’t do on any camera in the computer. It may save you some time in post-processing but it’s not giving you something that you didn’t already have the ability to do yourself.
I have the older Lumix FZ20 that is only 12x Optical and the only time I use the digital zoom is if the camera is stationary and I am effectively using the camera as a telescope. I can get the same zoom on my computer as the digital on the camera gives. (I think you already understand this from the Op)
I love the camera, but I see no difference in quality between the camera’s digital zoom work and just doing it on a PC. It is still all about optical zoom from what I can tell.
Now your camera is 8 megapixels, so what it can do much better than mine is do the digital zoom on the camera itself, reducing the quality and still give you as good a picture as with 36x zoom as I get with my 12x optical 4 megapixel camera that takes amazingly sharp shots. The bonus to this is it transfers a much small .jpg to your PC or external storage device and makes it easier for your photo tools to work.
I wondered from the outset, and hear several of the kind respondents saying things that sound similar to me, about whether what they mean is that I can set my control to use the entire 8 megapixel sensor, or use the central 3 MP region, or the central 1 MP region, et cetera. I think this would be useful if the viewfinder was showing this small region, so I could better see what I was taking such long pictures of. My last camera was an Olympus with a big zoom, and only 1.3 MP total, and I know it’s not unusual for people to set their cameras for lower resolution (though I don’t). Thus, it seems like such a partial-sensor feature could be useful enough to learn how to use, and not bad for a microprocessor-only feature addition.
I see, though, that when I set the camera for 3 MP, the view at the wide angle lens setting is every bit as wide as it is for 8 MP. So, that’s not the way to think of it.
If there’s some kind of subsampling that is refigured at each viewfinder refresh cycle during a high speed or low speed zoom, and it is all sort of spliced together, well, they made it surprisingly smooth.
Whatever they did increases the zoom range, the apparent focal length at the long end divided by the apparent focal length at the short end.
So does “digital zoom”, but in a way that is obviously pieced together and klunky.
At this point I’m mostly just curious about what they did, and also wishing I wasn’t confused about how digicams zoom. Smooth or not, I can’t really say this isn’t just a trick.
The most important thing I haven’t tried is looking at saved images on the PC.
What you need to keep in mind is that most digital cameras use a monochromatic CCD/CMOS sensor with a color filter on each pixel. Most common arrangement of color filters is called Bayer and look like this
...
... R G R G R G R G ...
... G B G B G B G B ...
... R G R G R G R G ...
...
So you have 8 million pixels. 4 million green, 2 million red and 2 million blue pixels. To produce your final image, the two missing color components per pixel need to be interpolated. Then your image is scaled to the desired size, and converted to YUV color space for JPEG compression. Depending on the DSP, algorithms, etc. different steps can be combined in different ways with all sorts of processing before, during and after.
If you think of this in terms of input and output geometries for the scaling step,
you will see that setting the output geometry to 3MP still allows you to use the whole wide 8MP input.
In these terms, consider these examples:
Normal = Lens: Wide, Scaler Input: 8MP, Scaler Output: 8MP
Optical zoom = Lens: Zoomed, Scaler Input: 8MP, Scaler Output: 8MP
Digital zoom = Lens: Zoomed, Scaler Input: 3MP, Scaler Output: 8MP
Extended optical zoom = Lens: Zoomed, Scaler Input: 3MP, Scaler Output: 3MP
Edit: I think the point of my ramble is that due to inherent interpolation and colorspace conversion, your image is always in some sense ‘scaled’ even if the size is not changed, so if you think of scaling as something that always happens but with different parameters, it’s easier to understand.
>think of scaling as something that always happens but with different parameters
How universally true is this with digicams? Is it the case with RAW output? How about if your camera has the kind of sensor with multiple layers so each pixel has three values read (the Foveon, Foveon 1/1.8" X3 Image Sensor: Digital Photography Review)? How about if, like my first digicam, there are advanced instructions for re-reading the sensor to remap bad pixels?
Not universally true, and it depends on how you define scaling and what you group with it. I consider Bayer interpolation to be equivalent to scaling.
If the RAW output is unchanged in size and uncompressed (or compressed losslessly), then no it is not scaled.
They do not require any interpolation, but they still may do it. However, if you are saving into any format that uses non-RGB colorspace (say a YUV422 JPEG), then even though the image is not scaled, some information is still discarded in the colorspace transform. The color-space transform does not have to be a per-pixel filter or even distinct from the scaling algorithm either.
Depending on how advanced the defective pixel algorithm is, it will have to interpolate away the bad pixels before applying any scaling (or as part of applying scaling).
