F30 Digital Zoom vs 3x Optical Zoom

Prompted by a series of posts that say that Digital Zoom could be better than from cropping and expanding from Max 3x Optical Zoom photo, the following tests were conducted.

Arguments for Digital Zoom to be better are:
a) Fuji actually have 12M pixels and interpolating in camera before committing to 6MB jpg could give better result.
b) With the proper framing enabled by the digital zoom, the metering would be more accurate and will result in a better picture.

The pictures taken at the various zoom are shown on the left. For comparison purpose, a Panasonic FZ3 (3 M) with 420mm zoom is included.

The subject for comparison purpose is the wooden bird. The pictures are cropped to have just the bird. For the 3x optical zoom, the cropped image is upsized using bi-cubic at 10% step until the image is roughly 640x480. The mid digital zoom picture (282mm) is cropped and not re-sized at all. The max digital zoom picture (695mm) is down-sized using Bi-linear(2pass) in 10% step till image is about 640x480. The Pana FZ3 shot is cropped and shown as it is without any resizing. The Pana's image is slightly more yellowish compared to the actual Bird color. F30's bird greyish color is more accuate.
I have set the shutter speed to be roughly an inverse of the focal distance to compensate for the camera shake for zoom shot. At higher zoom, the effect of camera-shake is strong. I have to replace the 1/320 shot which shows some blur due to camera shake with a 1/250 one which gives a sharper image. I guessed I must be holding the 1/250 shot better than the 1/320 one.

What can we conclude ? I think the 695mm picture is worse, and may be the the 282mm is very very slightly better than the sized-up 113mm shot. But they could not compared with the FZ3 shot at all.

The recommended rule might be stay with optical zoom. The digital zoom needs one to have higher shutter speed to compensate for camera-shake that reduces the light input. Mid range digital zoom perhaps can be used when the lighting is adequate.

F30 Tips

1. Metering Mode and Focus Mode
Pattern Metering tends to over-exposure. May need to use -1/3 EV for Outdoor Shot to prevent blown-out, and even more -1 EV for high contrasting light.
Average Metering mode tends to under-expose but good for preventing blown-out.
Use center-Metering with Center Focus else may lead to surprising results.

2. Use ISO 1600 -1 EV than ISO 3200 according to Dynamic Range Chart if you dislike the in-camera processing for ISO3200 picture. Use ISO 1600 auto or ISO 400 auto according to the lighting conditions of your photos.

3 . Use smaller aperture to reduce Purple Fringing PF. Sweet spot for F30 is about F5.6.

4. Always do Noise Reduction First in Post-Processing work flow.
a. Preserve noise patterns so that stored noise profile can be applied.
b. Subsequent operations will create fewer artifacts due to removal of noise earlier.

5. Macro - Turn on macro when focusing distance is less than 60cm.

6. Sharpness - wide is best. Aperture F3 to F5.6 give best too. Stay away from F7 to F8.

7.Good F30 Tips Blog:
http://littlezumbari.blogspot.com/
http://www.pbase.com/arn/fuji_f30

Dynamic Range and ISOs

The DigitalCameraInfo.com review of F30 has a chart on F30 Dynamic Range over ISOs...

http://www.digitalcamerainfo.com/... .../Testing--Performance.htm

It is interesting to note that at High Quality (0.1), F30 has a dynamic range of 6.8. It stays well as ISO200 with 6.6, but begins to drops at increasing rate when ISO is increased, reading from the chart, approximately 6.1 at ISO 400, 5.4 at ISO800, 4.8 at ISO1600, and 3.9 at ISO3200.

What I learned:

1. Dynamic range decreases with increasing ISO.

From ISO 100 to ISO3200 the drop is 3EV. From 400 to 3200 is 3EV. From 800 to 3200 is 1.5EV and From 1600 to 3200 is about .9 EV.

2. If the drop from 1600 to 3200 is .9EV loss of dynamic range and 1600 to 3200 is 1 EV of sensitivity. It makes sense to say that using 1600 with -1EV underexposure is about the same as using ISO3200.

3. Looking at the Dynamic Range of other Camera over the ISOs, F30 performed quite well. For example: Nikon D80, reading from the Chart,

Nikon D80 7.5 at 100, 6.8 at 200, 5.8 at 400, 6 at 800, 4.8 at 1600, 3.5 at 3200.
FinePix F30 6.8 6.6 6.1 5.4 4.8 , 3.9

But this does not mean F30 is as good as D80 as there are factors to picture quality.

"Raw" Processing - More Details

The test result in the past post was shared in the DPReview Fujifilm forum and many comments were received.

The key point is - 'underexposure' causes lost of color information and it is not advisable to go far too low from correct exposure, and certainly not ISO 100 for shots that need ISO3200. Some think one should just stick with in-camera processing.

To understand this better, we plot the histogram/levels curve for the same photos as below:
The correctly exposed ISO 1600 has a smooth distribution curve. The under-exposed ISO100 has a distribution curve concentrated on the left. To get the right photo, we need to 'spread' the distribution out. The typical way is use to gamma correction. I tried it out first but it looks terrible. See the 1st photo below. The curve becomes spikes showing loss of information. However, PhotoBrush has a EV correction function and I tried it with max 4EV and than another 2EV to get the final result show above. The key part (right side) of the curve is quite smooth (due to interpolation when the image is saved) but you can still observe some spikes in the left part of the curve showing worse interpolation at the left end (darker side) of the histogram. This is due to JPEG non-linear tonal curve compression method. The end-result is, in my opinion better than ISO1600 photo - smoother, more contrast and better letters details. But the ISO1600 preserves better texture and to experts, a better photo. But you need to make your own judgment and decide on your photo taking and processing technique.

The histograms of the other two post processing software are shown below. We can see that the Picasa one is made up of more spikes. The FinePix Viewer seems to interpolate quite well but till under-exposed and can be adjusted further.


In summary, when you under-exposed your shots, you will loose color information and hence don't under-expose too much. I believe you have the alternative to the in-camera ISO3200 option with a few stops lower ISO 1600(auto), 800 or 400(auto).

"Raw" Processing with Fujifilm FinePix F30

Compact camera like F30 does not have the real unprocessed raw file. Usually, only high end prosumer like the F6500 or DSLR has them.

However, I discovered that we do have some sort for 'raw' file capability. That is to take picture at set at ISO100 (the lowest possible), choose Shutter priority, get the focus right, choose the right shutter speed for the object (inverse of focal length i.e. 1/30 at wide, 1/60 or 1/100 at zoom end of F30 for hand motion blur), and let the metering choose the aperture (usually max F2.8 in indoor shot). If there are bright light, then the metering will choose smaller apeture like F5 or even F8 - but you are ok for you are shooting with ISO 100 already.

notes: True Raw file is the data captured in the CCD before any processing by the camera. It contains 12 bit (some high end camera may have more bits) information for each pixel. The file in F30 has been converted into Jpeg format. In jpeg, there is only 8bit for each color pixel (or 255 levels).

Under in-door and low light condition, at ISO100, you will get quite a dark picture. But that is ok because you can use software to get the picture you want. That is what raw processing do ... software vs in-camera hardware processing built into the camera. Please note that due to under-exposure, the dynamic range is reduced and so you are loosing information and details. After post-processing, you get a more plastic looking nevertheless less noise photo... you loose textures.

Below is a series of different ISO photos taken - fixed at 1/30, F2.8. The correct exposure is 1/30 at ISO 1600. You will see at ISO 3200, the camera choice of F3.2 is over-exposed a little. At ISO 100, you could hardly see the photo.





However, by using software like PhotoBrush I can process the ISO100 photo with exposure +6EV, and then use Neat Image to remove the noise. The resulting photos vs the correctly exposed ISO 1600 is in fact slightly better! Slightly more details and with Neat Image more smooth.

However, expert in Fujifilm Forum pointed out the ISO1600 photo is a better picture with more details - look closer at the area surrounding the white spot of the bottle. You can see the reflections of the surrounding areas better. I was looking at the wording on the CDs which appeared clearer to me (that is because the dynamic shades have been lost with under-exposure resulting in flatter look and contrasting look and appear better when just looking at letters). Saying it another way, loss of colors range due to under-exposure will give a flatter and more plastic looking photo after processing (there is lost of textures).
Similar result is obtained by using Picasa2 choose "I'm feeling Lucky" button or in FinePix Viewer choosing Auto Adjust. The procoessed ISO photos are shown below. Even with just Piscasa one-step processing, the photo turned up to quite nice. ... rich in color and low noise!
I have tried this same thing on my Panasonic FZ3 (3Mp) camera. Under expose at 1/30 ISO400 and 1/8 correct exposure (FZ3 has image stabilisation) and obtain similar result ... software correct an under-exposed picture is good!
I took the above shots was for comparison with F30 for noise processing - image stabilization vs high ISO capability. Forgetting about the difference in size 6M vs 3M and just look at noise. F30 wins.( Comparing F30 ISO1600 vs FZ3 1/8 more light exposed ISO400).

Comparing ISO 3200, 1600, 100+6EV after Neat Image default is shown below. Please not it is not fair to ISO3200 as the metering overexposed the image. The ISO100+6EV could even be brightened a little to compare with the ISO 1600 shot.


This series of test was conducted to see if we could do better than the camera use of ISO3200 under low-light conditions. The answer seemed to be yes but need your own software post-processing. With software like Picasa, the post-processing is not that much time consuming or difficult. So, we could use ISO400(Auto) for a good balance between ready-made and post-processing effort and get better than ISO3200 photos. Will be trying this for real world low light photos to see if this works out well. ... try to use the highest possible ISO to minimize loss of colors details due to under-exposure.

DV and DP are getting closer

Now, 2006 April, 5 years from the original post, the gap between Videocam and Digital Camera is now much closer.

Video: Digital Camera is doing 640x480 30 fps. Videocam has moved to mpe2 720x540 format.
Storage: SD card can now do 2GB to 4GB. Video Tape is 60min for 2GB too.

Videocam have better audio input, zooming and focusing at the same time.

DVandDP

There are good sites on Digital Videos and there are good sites on Digital Photography but there seems to be little interests on have DV and DP combined.

While strictly speaking, DV and DP should be differentiated since DV is more about Motion and Audio and DP is about still.

With DV, the action is exciting. With DP, I can take my time to browse and enjoy the photo.

But if I want to do both, what do I buy ? A DV cam with still capture ability or a DP cam with video capability.

Physics is the limitation. A system designed for Video capture is not suitable for capturing the details of a still. There are compromises to be made. So if you really are for video, go buy a proper DV cam. If your need is for still, go for a really good Digital Camera.

But what if I am 80% still and 20% motion ? Which one should I pick?

It is greyed needs like this that make our lives interesting or difficult.

I attempt to look at some DV and DC and shared my experience.

See http://groups.msn.com/DVandDP for details.