Big Mike's Digital Photography Blog

Olympus E-620 ISO 3200 test

The following series of images were taken at ISO 3200:

1. JPEG, Noise Filter = Standard

2. JPEG, Noise Filter = Standard

3. JPEG, Noise Filter = Off

4. RAW, via Adobe Camera Raw

5. RAW with Photoshop Reduce Noise filter

6. RAW with Photoshop Reduce Noise filter, resized to 1200 x 900

The first image shows us the well exposed scene which was shot at ISO 3200. Now it’s very important to get your exposure right when working at ISO 3200, because the camera has much less dynamic range at this sensitivity. Using center-weighted metering, I had to use +1 1/3 EV exposure compensation. If I had just let the camera do its thing and underexpose the image, and tried to correct the underexposure in post processing, I would essentially be working with an image that was push processed to ISO 8000, so the results wouldn’t be as good as we see with the well exposed image.

The second image shows us a 100% crop of the JPEG saved with the Noise Filter set to Standard. This is the setting I recommend for the Noise Filter (although I admit that I haven’t really experimented much with Low or High).

In the third image, the Noise Filter was set to Off. The result is absolutely horrible! What has happened here is that the JPEG processing engine has applied sharpening to the noise and made the noise look horrible. This happened even though I had in-camera sharpening set to -2, the lowest value. This is impossible to fix in post processing. The image has been ruined (although, admittedly, a lot of detail has been preserved). The E-620 JPEGs have this problem no matter what ISO you use. Even at ISO 200, if you have the Noise Filter set to Off, the noise is sharpened and looks ugly. Don’t set the E-620 Noise Filter to Off!

The fourth image was developed using Adobe Camera Raw (the ORF file was first converted to DNG because I don’t have the latest version of Photoshop). There is a lot more chroma noise than in the JPEG in which the Noise Filter was set to “Off.” Obviously, even when turned “Off,” the Noise Filter is still really on with respect to chroma noise.

I also looked at the ORF file in Olympus Master and discovered that Olympus Master did the same lousy sharpening as the in-camera JPEG image. As far as I can tell, Olympus Master does nothing but give you the exact same JPEGs you would get straight from the camera, so what's the point of it? A big thumbs down for Olympus Master. It’s not worth bothering with RAW files if Olympus Master is all that you have.

The fifth image, in my opinion, is the best result. I took the raw image, and used the Reduce Noise filter in Photoshop. There is a good compromise between noise and resolution. You could get even better results (not a whole lot better, but better nevertheless) if you used a dedicated noise reduction program or Photoshop plug-in such as Neat Image.

Finally, to create the sixth image, I took the RAW file that was processed in Photoshop and had the Reduce Noise filter applied to it, then resized it to 1200 x 900, and took a crop from the 1200 x 900 image. Now what’s the point of that you ask? Well on my monitor, a 1200 x 900 image is approximately 12.5 x 9.5 inches, so I figure this is a pretty good approximation of what would happen if tried to make an 8 x 10” print, and it’s also pretty much the largest size you would ever use for publishing an image on the web.

In the sixth image, the noise is barely noticeable. Thus my conclusion is that a well-exposed ISO 3200 image from the E-620 is perfectly usable for 8 x 10” print purposes or web viewing if you shoot RAW. But my other conclusion is that Olympus has done a really bad job with its in-camera JPEG engine, because I like the unfiltered result from Adobe Camera Raw better than any of the poorly processed in-camera JPEGs.

E-620 vs. G9 at ISO 400

After my outdoors shootout in which I discovered that at ISO 200 the Olympus E-620 was less sensitive than the Canon G9 at ISO 100, I decided to do a higher ISO test. ISO 400 is the highest ISO at which the G9, in my opinion, gives you a useable image that’s not ruined by noise. I wanted to see how well the E-620 did at a comparable sensitivity.

I was really surprised to discover that the E-620 and the G-9 seem to have the exact same sensitivity when both cameras are set to ISO 400!

Canon G9, ISO 400, f/3.5, ¼ second

Histogram for the Canon G9 photo

Olympus E-620, ISO 400, f/3.5, ¼ second

Histogram for the Olympus E-620 photo

I don’t understand why the cameras behaved so differently between this test and the outdoor shootout test.

You can take my word for it that, at ISO 400, the Olympus E-620 with its bigger pixels is significantly less noisy than the Canon G9. The E-620 blows away the G9 at ISO 400. (But unfortunately, the E-620 front-focused here, just as my E-500 did, even though the E-620 and 14-54mm combo focuses perfectly on a white sheet of paper with a focus test printed on it. The Canon G9 has much more dependable focus.)

E-620 and JPEG sharpening

I’ve previously complained about the Canon G9’s oversharpened JPEGs. Does the E-620 have this problem? The following are 300% crops of text on paper:

G9, lowest sharpening	E-620 default sharpening	E-620 lowest sharpening

At the default setting, the E-620 image has observable sharpening halos, but the problem is much less severe at the lowest setting of -2, with the output looking significantly less processed the the Canon G9. The default setting is reasonable given that unsophisticated consumers like the sharpened look better, and more experienced photographers can simply adjust the settings.

HOWEVER, even at -2 the E-620 is applying a heavy amount of sharpening, which can be seen if you set the Noise Filter to Off because that causes all of the noise to be sharpened. This is demonstrated in my E-620 ISO 3200 test. The only way to turn off the sharpening is to shoot RAW.

Olympus E-620 vs. Canon G9: shootout!

I previously compared a Canon G9 to an Olympus E-500 DSLR, and the Canon G9 had much greater resolution than the Olympus E-500.

Now that I have a new Olympus E-620 DSLR incorporating Olympus’ latest technology, I was eager to see if Olympus finally made a DSLR that could produce a higher quality image than a Canon G9 point and shoot camera.

The Olympus image was shot at 54mm focal length using the 14-54mm f/2.8-3.5 lens, ISO 200, f/7.1, 1/400 second, and saved in RAW format. The RAW was converted to a DNG file using the Adobe DNG converter, and then opened in Adobe Camera Raw 4.6. I used an unsharp mask to make the image a little sharper.

The Canon image was shot at 22mm focal length, ISO 100, f/5.6, 1/1000 second, and just saved as a JPEG. No unsharp mask is needed because the G9 applies sharpening to the JPEG before saving it—too much sharpening in my opinion.

Canon G9

Olympus E-620

When you compare the images from the two cameras, the E-620 image seems to be exposed an extra 1/3 EV compared to the G9 image. However, based on the camera settings, in which the E-620 has 2/3 higher f/stop, 1 EV higher ISO, and 1 1/3 EV of slower shutter speed, one would expect the E-620 image to be exposed by an extra 1 2/3 EV. There is a 1 1/3 EV discrepancy. In other words, the E-620 is shamefully overrating its ISO compared to the G9. ISO 200 on the E-620 is equivalent to ISO 80 on the G9. Can this really be true?

According to the good folks at DxOMark, the G9’s true measured ISO at ISO 100 is 139. DxOMark hasn’t profiled the E-620, but the E-620 is just the consumer version of the E-30 which DxOMark did test, and DxOMark says that at ISO 200, the true measured ISO of the E-30 is 128. Thus DxOMark does agree with me that the similar E-30 camera is less sensitive at ISO 200 than the G9 at ISO 100. My real world test confirms the DxOMark testing. [HOWEVER, when I did a test at ISO 400, both cameras had the same sensitivity at ISO 400.]

Now let’s compare some 100% crops:

Canon G9

Olympus E-620

Canon G9

Olympus E-620

From looking at the full images you can see that there was slightly more zoom used on the E-620 shot, so on that basis you would expect the E-620 to show slightly more detail. But if you compare the first set of crops, which is from an area near the center of the image, it’s really hard to say if one camera has resolved more detail than the other.

When you compare the second set of crops from an area near the edge of the image, the E-620 beats the G9. The G9 image shows a little bit of chromatic aberration and is slightly softer. But before you proclaim the E-620 the clear winner at the edges, I should point out that on my E-500 camera, this lens showed no CA at the zoom end used to take the above image, but there was quite a bit of CA at the wider end of the lens.

In both sets of crops, the image from the E-620 has more contrast. It’s impossible to determine if that’s the result of bigger pixels, a better lens, the tone curve applied by Adobe Camera Raw, or the fact that the lens on the E-620 had a lens hood.

From a resolution perspective, I call this test even. The 12 MP DSLR resolves the same amount of detail as the 12MP point and shoot camera with a much smaller sensor. The bigger pixels don’t provide a noticeable advantage when it comes to resolution. If there is slightly better image quality from the E-620 (and you need to do some determined pixel peeping to see it), it's probably because the Zuiko 14-54mm lens is slightly better than the lens which comes with the G9.

DYNAMIC RANGE

Because of the bigger pixels, the DSLR is supposed to have greater dynamic range than a point and shoot camera like the G9. Is this the case? Unfortunately, I can’t really see it in the image. You will notice that there are some very dark shadow areas in the lower left side of the image. We can try to bring this up using the Photoshop Shadow/Highlights adjustment for the G9 JPEG, and the Adobe Camera Raw “Fill Light” slider for the E-620 RAW file.

Canon G9

Olympus E-620

It seems to me that the G9 JPEG captured more detail in the shadows compared to the E-620 RAW file which is supposed to have much greater dynamic range. Once again, the bigger pixels of the E-620 were unable to demonstrate an advantage over the smaller pixels of the G9.

My finding of no DR difference is also consistent with the test results at DxOMark. According to DxOMark, at ISO 100 the G9 has a DR of 9.4 EV, and at ISO 200 the Olympus E-30 (which has the same sensor as the E-620) has a DR of 10.07 EV. That’s not a very big difference, and I did expose the E-620 shot by about 1/3 EV higher than the G9 shot. If DxOMark is correct about the Olympus DSLRs having an extra 2/3 EV of DR at these ISO settings, then the extra DR is to be found in the highlights, because I don’t see any extra DR in the shadows.

CONCLUSION

With the E-620, Olympus has finally achieved parity with the Canon G9 point and shoot camera with respect to resolution and overall low-ISO image quality. But there is still no reason to use the E-620 for typical landscape or travel photos taken during daylight hours. The Canon G9 is a lot more convenient to carry around and it has better exposure metering.

Want a second opinion? Read the Canon G10 vs. Hasselblad H2 comparison at Luminous Landscape:

In every case no one could reliably tell the difference between 13X19" prints shot with the $40,000 Hasselblad and Phase One 39 Megapixel back, and the new $500 Canon G10. In the end no one got more than 60% right, and overall the split was about 50 / 50, with no clear differentiator. In other words, no better than chance.

UPDATE

Although I discovered here that the E-620 overrates its ISO, relative to the G9, when the E-620 is at ISO 200 and the G9 at ISO 100, I later experimented with ISO 400 and was surprised to discover that both cameras behave nearly the same at ISO 400 (with respect to creating images with the same apparent exposure). Very strange.

Olympus E-620's secret mirror lockup function

I say that the E-620’s mirror lockup function is a secret because of how difficult it is to find it. I tried to find “mirror lockup” in the index of the (extremely poorly written) instruction manual, but to no avail. Only by searching on the web was I able to discover that the mirror-lockup function is called “ANTI-SHOCK” in the Olympus world. Someone at Olympus thought it was a good idea to change the name from what photographers are used to calling it to some new Olympus-only name. They also thought it was a good idea to spell it out in all capital letters.

Even after you know the correct name, the mirror lockup function continues to be a big secret. In order to turn it on, you have to go into the Custom Menus and change the setting from Off to some time setting (I used the default of two seconds) which is the amount of time between the mirror going up and the shutter triggering. The Custom Menus themselves are sort of a secret, because they also don’t appear unless you enable them.

Once “ANTI-SHOCK” is enabled, you can then access it in the Sequential Shooting menu (which also has the self-timer and remote control options). There’s no reason not to enable “ANTI-SHOCK,” because it’s not always turned on, you just get the option of selecting it in the Sequential Shooting menu. It’s not at all clear to me why, by default, the mirror lockup option is turned off, or why Olympus thinks there’s even a need to turn it off.

Below is a dramatic example of the benefits of mirror lockup. (In case you are wondering, I had IS turned off for both of these photos.)

Without mirror lockup

With mirror lockup

The above examples also demonstrate that a light and inexpensive tripod doesn’t keep your camera perfectly steady when pitted against a slapping mirror.

You should always use mirror lockup when shooting inanimate objects with a tripod, and you should never buy a DSLR without this feature. There are still some entry-level DSLRs on the market which don’t have this feature. Caveat emptor!

Olympus E-620: focus and white balance

Here’s a photo of a bottle of Johnnie Walker Black Label scotch, taken with the Olympus E-620 camera. The focus is on the label at the bottom of the bottle. The photo was taken with the 50mm f/2.0 lens at f/2.0 and ISO 200.

Compared to the E-500 (see similar photos in my review of the Olympus 50mm f/2.0 lens, the auto white balance is tremendously improved, but the “ESP metering” still underexposes the photo. Yes, I know that there are a lot of white and pale colors in the photo, but “ESP” implies a super-smart automatic exposuring which could figure that out. When I look at the histogram, it’s pretty obvious that the scene is underexposed. Thus the ESP metering on the Olympus E-620 DSLR is not as good as a compact camera like the Canon G9 (see the G9 comparison photo here).

We see, however, the E-620 is able to focus accurately on the label on the bottle. The center autofocus point was pointed at the bottom label, and the AF Sensitivity was set to Small in the Custom Menu. I haven't seen any advice on the web about the benefit of using Small or Normal AF Sensitivity, but I will keep mine set to small so it focuses on what I point the camera at. [UPDATE: I set the AF Sensitivity back to Normal, because with Small sensitivity the camera is more often unable to lock focus, even when outside in bright daylight.]

One thing that stands out in this ISO 200 shot is that there is a surprising amount of noise (the Noise Filter was set to Off) which became even more noticeable after I applied an unsharp mask in Photoshop. In fact, it’s noisier than the Canon G9 at ISO 80. I’m not saying the noise will be noticeable when the picture is reduced to normal print size or web-viewing size, but it’s surprising, nevertheless, in a negative way. Of course, it’s possible that the G9 is applying more noise filtering at the expense of resolution, so very careful pixel peeping will be required to see what’s really going on.

Olympus E-620, initial impression

I took my new E-620 camera out of the box, put a 14-54mm lens on it, and the first impression I get from playing around with it is that the camera is doo darn small.

Now I’m not generally in favor of cameras being big and heavy, because a small and light camera is easier to carry around. That’s why I don’t understand the aversion to plastic seen in a lot of camera reviews and in internet forums. If they can make a camera lighter by using quality high-tech plastic that does the same job as metal, then I say use the plastic!

That said, after you put the 14-54mm f/2.8-3.5 lens on any camera, it’s no longer such a small and light camera. The E-500 had a decent sized grip which made the camera easy to hold. To make the E-620 small and cute, the big grip was removed and placed by a pathetically small grip, which makes the camera much less comfortable to hold, and probably harder to hold still. The smaller E-620 also holds a smaller battery which means it will run out of power faster.

The smaller size is definitely a step backwards with respect to usability, without having any practical effect on the camera’s portability.

The cute and tiny size, at the sacrifice of the handgrip and larger battery, demonstrates Olympus’ gimmicky marketing-driven approach to selling cameras instead of actually trying to produce better photographic tools. Another gimmick is the useless "art filters." The R&D required to create the art filters could have gone into programming stuff that would be more useful. (There issue isn’t that there aren’t any real photographic improvements since the E-500, but rather that Olympus could have made a better camera for the same price.)

On the positive side, it seems to focus more confidently in my dimly lit apartment compared to the E-500, even when I point the AF sensor at dark shadowy areas.

Canon EF 50mm f/1.8 vs. Olympus 50mm f/2.0

In this blog post, I compare the Canon EF 50mm f/1.8 II to the Olympus Zuiko ED 50mm f/2.0.

But wait, you say. How can I write about the Canon lens if I don’t own it and don’t even own a Canon DSLR? Well luckily for me, both lenses were tested using the same criteria at dpreview.com, so I can use that as a basis for putting my own spin on the difference between these two lenses. Here are the links:

Canon 50mm review
Olympus 50mm review

Also, please refer to my previous blog post where I reviewed the Olympus lens.

PRICE

The Olympus lens sells for $450 while the Canon lens sells for only $110. That’s a huge price difference. Is the Olympus lens worth four times the price? Probably yes as a macro lens, and no as a portrait lens. But we’re just reviewing portrait lenses here.

IMAGE STABILIZATION

A big advantage of the Olympus lens is that it’s image stabilized. This is because all Olympus lenses automatically become image stabilized because Olympus builds image stabilization into the camera bodies. This can make a significant difference when shooting handheld in low light. For example, a shutter speed of 1/40 second is acceptable with respect to subject movement if you are photographing someone who is relatively still, but at this shutter speed you won’t be able to handhold a short telephoto lens without getting a lot of camera shake. But with image stabilization, 1/40 becomes a perfectly handholdable shutter speed.

FOCUSING

I can’t really do any tests because I don’t have a Canon system. Reports from various internet forums is that the Canon 50mm f/1.8 focuses slowly and noisily compared to more modern lenses which have “USM” focusing. But it only costs $110.

Unfortunately, the only Canon APS-C camera with micro focus adjustment is the 50D, so if you buy this for a Canon Rebel, and if for some reason the lens or the camera is out of alignment (in a way that’s correctable with micro focus adjustment), you won’t be able to fix the problem yourself. There's a good chance that Canon won't add this feature to future Rebel cameras because Canon purposely leaves stuff out of the Rebel cameras in order to incentivize people to buy the more expensive xxD cameras.

IMAGE SHARPNESS

When you put a lens designed for 35mm film on an APS-C digital camera, you normally expect to get a lens which doesn’t perform so well. This is because approximately 63% of the imaging circle is being wasted. However, this lens seems to be an exception. On 35mm-sized sensors, the corner softness of this lens is so bad, and affects such a large percentage of the imaging circle, that there is actually an overall improvement when this lens is placed on an APS-C camera because the center of the lens is very sharp.

Based on the dpreview test, wide open this lens gives an acceptable level of performance through most of the central part of the image. In fact, if you cropped the image into a square, the majority of the square would be pretty sharp except for the extreme corners. On the full 3:2 aspect ratio sensor, there is pretty bad corner softness outside the square area, but as long as you keep your main subject within the square, the important part of the image will be sharp enough. If you compose according to the Rule of Thirds, the subject will fall within the square.

In comparison, the Olympus 50mm lens is slightly sharper in the middle when wide open, and the sharpness is maintained through the entire area of the photo including the extreme corners. This is an impressive performance, but it’s not really that important in real life situations where the corner area of the photo will be out of focus anyway because of these lenses’ shallow depth of field when used wide open.

Dpreview also praised the Canon lens for having very little chromatic aberration, which is impressive for a $110 lens. The Olympus also has very little chromatic aberration in the focal plane, but of course that’s what you’d expect for four times the price.

BOKEH

The following image with this lens was taken with the Canon lens at f/1.8:

Source: Elliot Freeman

The main problem with the bokeh is that the circles of light in the out-of-focus areas have slightly hard edges.

The bokeh becomes much worse when the lens is stopped down:

Source: Gwen Harlow

The pentagon shapes are a result of this lens’ cheap 5 blade aperture. You don’t get this problem when the lens is fully open, because the aperture blades are fully retracted from the lens opening. Thus, you are better off using this lens wide open than stopped down a little bit.

A positive point about the Canon lens’ bokeh is that there is hardly any bokeh chromatic aberration, or spherochromatism if you prefer. These are the color fringes which show up on some fast lenses as magenta fringes on high contrast edges in front of the focal plane, and as green fringes on high contrast edges behind the focal plane. Below is a crop from a photo I took with the Olympus lens showing the green fringes:

In case you think this only happens on my camera, here’s a link to a sample photo from the dpreview review of the lens showing purple fringing caused by the spherochromatism in the bokeh: click here.

So my final bokeh conclusion is as follows:

(1) The Olympus lens has much better bokeh when the lenses are stopped down.
(2) Wide open, the Olympus lens has better black and white bokeh, but the Canon lens has better color photo bokeh.
(3) Neither lens has really great bokeh, but the Olympus lens has above average black and white bokeh.

WHICH LENS TO BUY?

If you own an Olympus system, you have to buy the Olympus lens because it doesn’t take Canon lenses. If you own the Canon system, you have a few more choices, but the Canon EF 50mm f/1.8 lens is a pretty compelling bargain.

WHICH SYSTEM TO BUY?

The problem with the Olympus system and portrait photography is the lack of fast prime lenses. A Canon DSLR with the Canon 50mm f/1.8 lens is overall a better deal as a portrait camera, given the inexpensiveness of the lens, and the fact that you get a little bit more background blur on the Canon system because the Canon lens is 1/3 of an f-stop faster, and you can get closer to your subject given the slightly wider angle view (50mm is equivalent to an 80mm “full frame” lens on Canons with a 1.6x crop factor, and 100mm on Olympuses with a 2.0x crop factor). It's true that you'll have soft corners using the Canon camera with the 50mm f/1.8 lens, but wide open the corners are usually out of focus anyway so this is usually won't impact the image quality.

Although Canon lacks image stabilization, you do get an entire extra EV of high ISO ability with a Canon camera because Canon image sensors just do a lot better at high ISO, so this makes up for the lack of image stabilization.

If Olympus were to come out with a reasonably priced f/1.4 portrait lens, the advantage would probably go back to Olympus. It’s unlikely that Canon will come out with any new EF-S lenses that will help with portrait photography because, clearly, Canon’s marketing strategy is to make you move up to full frame cameras and “L” lenses so Canon can make a lot more money selling the more expensive equipment. If Canon made its Rebel almost-as-good as a full frame camera with an “L” lens, Canon would lose money! Olympus doesn’t have any other system for you to move up to, so I don’t fully understand why Olympus hasn’t come out with a better selection of fast prime lenses.

Olympus E-500 focuses accurately on test chart

I created a makeshift focus chart in order to try testing my camera’s focusing system again. The following photo is a crop, reduced in size by 50% (along each dimension):

As you can see, focus is pretty much dead-on. Yet clearly, in my original review of the Olympus 50mm f/2.0 lens, the camera front-focused.

My best-guess conclusion is that the lens is in spec, and the Olympus E-500 DSLR focuses well when you point it at black text on a white background, but the autofocus is flaky when pointed at real-world objects like bottles of scotch. I will further guess that maybe the problem is that the center autofocus sensor is much bigger than the label, and is also picking up some of the glass, and the optical properties of the glass are confusing the autofocus.

These tests demonstrate the difficulties of using wide-aperture lenses with autofocus DSLRs.

Olympus E-620 on the way

Placed my order for an Olympus E-620. Ordered it from newegg (no sales tax to NY!). Hopefully it won't suck as bad as the E-500.

Even if I had wanted to buy it in person, no store in NYC has it in stock. What's the deal with that?