Tag Archives: High ISO

3781 – A Ship Awaiting Us

ISO 3200 on an Olympus camera is noisy. It’s not unusably noisy, but if you pixel-peep, you undeniably see noise.

It’s not really worse than the noise of a high-end Nikon or Sony camera with between 30 or 45 megapixels, it’s just that you have only half of the pixels and are more likely to peep.

The current trend is to keep pixel size constant and to produce sensors with the same density and different sizes. After all, in most cases we talk about variations of the same Sony technology.

In the end we look at our images at the same sizes, regardless of the camera they’ve been made with. At least for screen view, we downsize radically. While downsizing, we also downsize the noise. More pixels means more downsizing of noise and therefore less apparent noise. That’s why the essentially same sensors in “full frame” cameras get better noise ratings than their Micro Four Thirds counterparts.

Of course you can also have big sensors with a small number of pixels. Sony tried that with one variation of their A7 line. Theoretically the noise should be lower, but if you look at DxO test results, the high pixel count sensors still lead. After all, downsizing seems to be the better strategy and it is more versatile in any case.

Speaking of DxO, this image has been converted with DxO Optics Pro. The algorithm is much too slow for in-camera processing. For each pixel it looks at 1000 neighboring pixels. I’s applied math and it works extremely well, even with half the pixel count 🙂

3627 – Open the Door

My new PEN-F has a 20 megapixel sensor, and while I can’t complain about its quality, it is only a very incremental change up from the E-M1.

On the other hand it is incredible how much of a game changer DxO is. This image has been taken with the PEN E-P5 (same sensor as the E-M1) at ISO 6400, f4 and 1/6s. I’ve shot it at night in a very, very dark part of the street. It looks like, well, maybe ISO 800. And it’s not only the lack of noise, it’s also the purity of colors. Buying DxO has given me the equivalent of maybe eight years of sensor development. That’s incredible.

OK, that’s just to put the improvements of the PEN-F into relation 🙂

3504 – RS at Night

F3.2, 1/13s, ISO 6400. Dark again, but not only is the blue in the image much cleaner than it was from Lightroom, this time you have to look at the red fringe in the background. It’s clean and vivid, while in Lightroom it was completely washed out.

Can you improve a correctly lit daylight image using DxO PRIME? No, you can’t. To the contrary, it will cost you detail. Not much, but it will.

Above ISO 800? Seems like a safe bet on Micro Four Thirds. At ISO 1600 pretty much every camera will profit, and at any higher ISO it will work pure magic. You see me impressed.

3503 – The Circus is in Town!

F1.8, 1/13s and ISO 6400? I suppose you agree: it must have been dark there 🙂

By the time you read this, the image should already be on Flickr in full size. Look at it at pixel level.

Not impressed? Ok, now look at the small red inscription on the green glass in the upper left corner. Look at the small yellow reflexes on the chrome against the dark sky. That’s what I mean when I praise DxO PRIME’s color purity. I’ve never seen anything like that before.

2578 – Truth Doesn’t Make A Noise

A few days ago DxO came out with their flagship RAW converter product DxO Optics Pro in its all new and shiny version 9. Five years ago, back in 2008, I have already evaluated and then bought DxO Optics Pro 5.3, but later I lost interest in the program. Photoshop was faster, offered me much more flexibility, and using plugins like Noise Ninja or Topaz DeNoise, I could get comparably good noise reduction.

I looked into it later with version 7, but at that time I used the Panasonic LX5, and although that camera was supported, DxO did a mediocre job on its files.

With version 9 DxO has introduced a new, improved noise reduction mode called PRIME, and that’s the reason why I got interested again. Can DxO Optics Pro 9 possibly give me an extra stop in low light?

Actually for me the interest is rather academic. So far I have set my Olympus OM-D E-M5 to go no higher than ISO 1600 in Auto-ISO mode, and along with the excellent image stabilization and my preference for static subjects this has sufficed all the time. In fact most of my images are made at base ISO, and only a few go higher than ISO 800.

Currently I use Lightroom’s built-in noise reduction capabilities. Lightroom is not quite as good as Topaz DeNoise, but as long as I don’t go higher than ISO 1600, I see no difference. Apart from that, Lightroom behaves very predictably. If you fiddle with the sliders in Topaz DeNoise, you get a slightly better result if you do the right thing, but if you don’t, it can be easily worse.

As I remember DxO, it could also be customized in many ways, but the analysis that determined the parameters in auto-mode was normally sufficient to produce a good result. If at all, DxO smudged away a little too much detail for my taste.

What I also remember is, that DxO was slow. It just didn’t feel like a program that I wanted to use all the time. This is another point that needs revisiting with DxO Optics Pro 9.

The Windows installer is a big download with 317 megabytes. It comes complete with the .NET runtime 4.5, and the package seemingly contains both 32 bit and 64 bit versions. The actual correction modules for supported lens/body combinations are not contained, they are downloaded as needed.

Well, my first impression when starting the program was, that it is even slower than I remember. It may be that I am spoiled by Lightroom’s fast operation, but fact is, that DxO is maddeningly slow. It is slow to start up, and once you “develop” an image with PRIME noise reduction enabled (and that’s the one thing that I am after), it begins to remind you of early home computers. I’ve tested on two dual core computers, my laptop with 2.4 GHz, and a somewhat aging desktop with 3.2 GHz, both running an up-to-date Windows 7 64bit on 8 GB memory. On both computers it took between 10 and 15 minutes to render a single image.

Insane? Well, depends on what you want and how you look at it. DxO tells us that the PRIME algorithm looks at about 1000 pixels in the neighborhood of each pixel rendered. Clearly this explains the times, and we will see, that the results are indeed very good. How good? Well, I’d say DxO delivers the best noise reduction performance that I’ve ever seen. If you care to look at pixel level, it is clearly better than everything else, and it delivers that stunning performance even while running in full automatic mode.

I guess it’s time now for some examples. Let’s begin with an ISO progression. The images were taken at daytime in an indirectly lit room. I’ve used the OM-D E-M5 with Olympus 45/1.8 at f8, the camera mounted on a tripod, release delayed by 2 seconds. In order from top to bottom the ISOs are 200, 1600, 3200, 6400, 12800 and 25600. Each image has a crop from the original out-of-camera JPEG on the left and the DxO version on the right. You can click on the images to open them in full size.

There is clearly a an advantage for DxO above ISO 1600. Even at ISO 3200 the original JPEG begins to get fuzzy, and from there on it only gets worse. Low frequency color noise turns up, and from ISO 6400 up the colors get more and more washed out.

Both versions lose fine details, but DxO delivers smooth images without color noise. The detail retained seems pretty near the maximum possible and DxO holds colors pretty well.

This is not what I use high ISOs for though. Let’s look at the same progression, but this time it’s dim, artificial light, in other words, the typical light that forces you to use high ISOs.

Now it gets clearer, doesn’t it? You can see an obvious advantage even at ISO 1600, and the difference in luminance noise, color noise and purity of colors is stunning once we get to ISO 12800. I would still consider ISO 25600 unusable for my purposes, but the DxO version of ISO 12800 is actually pretty good.

Care for one more progression? Here is an outside scene at night:

This is pretty much the worst case for photography: most of the image is almost black, the few lights of some remote houses are extremely bright. There is dimly lit foliage on some remote trees.

We can see that DxO eliminates dead pixels. That’s fine but expected of a RAW converter. Again we see better colors in the DxO version. Better, you ask? Well, they match what we got at base ISO, thus DxO must be right, right?

You can compare the images in detail and in full size from a Flickr set that I’ve created.

Especially when you look at high ISO night images, you can see that the high ISO camera JPEGs show some banding in the darkest spots. Just look at the sky in the ISO 25600 night shot. Now look at the same image developed with DxO. Quite a difference.

Now let’s look at a part of the ISO 200 night image in Lightroom. I’ve applied a steep curve to extremely lighten up the shadows. Color noise reduction is on Lightroom’s default, sharpening and luminance noise reduction are turned down to zero. As to be expected: there is noise in the shadows, and if you look at the roof, there is horizontal banding as well. Of course you can’t see that in a normally exposed image. It’s just what you get when you abuse a digital image sensor.

Finally let’s have a look at a screenshot from Photoshop. You see the two ISO 25600 versions above, left the camera JPEG, and below the two ISO 200 versions, again left the camera JPEG. To all four images I have applied the same extreme curve.

Immediately you see that DxO has set a much more definite black point. This black point is retained even at extreme ISOs. At base ISO the camera JPEG shows a mottled pattern of magenta/green in the shadows. There are also sharpening artifacts. Not so with DxO.

To come to a verdict is not so easy.

DxO Optics Pro delivers very good image quality. It’s noise reduction is definitely the best that I’ve seen so far. If you frequently shoot at high ISOs and want to retain every bit of detail while keeping colors natural, DxO is the best tool on the market. It will be most interesting for people who shoot action or use long lenses in low light.

As an all-purpose RAW converter DxO fails miserably, at least for me. It is just too slow in general operation, even without PRIME noise reduction, and once you apply PRIME, it is impossible to use it for more than a few select images.

Thus: used as a special noise reduction tool, DxO Optics Pro 9 earns my recommendation. Whether you need such a tool is up to you.

The Song of the Day is “Truth Doesn’t Make A Noise” by The White Stripes. Hear it on YouTube.

2039 A – Our Darkness

This is not a great image and it’s not meant to be. Quality-wise it is about as good as I would have got from the Nikon D200, using for instance the Tamron 17-50/2.8 VC wide open at ISO 1600. Only this is the OM-D, f8 at 1/10 s and ISO 12800.

I wouldn’t have needed to go that high, I just wanted to see what I get at such a high sensitivity (or rather amplification). Only Lightroom was used. Taking it into Photoshop and blending layers with different strengths of noise reduction applied, I could probably have gotten an even better result. No need though, in practice I would rarely go that high. It’s nice to know that I can though.

The Song of the Day is “Our Darkness” from the 2007 Richard Hawley album “Lady’s Bridge”. Hear it on YouTube.

1796 – Compared To What

Some days ago Juha posted his findings about the new “High ISO Noise Reduction” setting, introduced with version 2.0 of the Panasonic DMC-LX5’s firmware, that was released on September 13.

Juha compared two images at ISO 1600, one developed from RAW using Aperture, and the other a JPEG from the camera, with “HIGH ISO NR” set to “on”. The image was a “twigscape” with slightly out-of-focus background.

Well, I thought I try it myself, but I use my own RAW workflow, trying my best to wrestle the most out of the RAW files, adapting the level of noise reduction to the particular image and trying to sharpen in a way to retain as much detail as possible.

The first comparison is with an image pretty ideal for noise reduction. There are plenty of hard contrasts, and in between we see lots of artificial, uniform areas. ISO is 1600 and you see two exposures: At the left is a JPEG out of the camera with “HIGH ISO NR” set to “off”, in the middle a second JPEG with “HIGH ISO NR” set to “on”, and on the right you see the second image developed from RAW.

The RAW file was converted using the Adobe DNG Converter, then developed in Adobe Camera RAW 4.6. In ACR I’ve completely turned off noise reduction, color as well as luminance. Then, in Photoshop CS3 I have used Topaz DeNoise 5 with a “RAW strong” preset, duplicated the layer, set it to “Luminosity” mode and sharpened that with “Unsharp Mask” (radius 0.3, strength 500, threshold 0). Basically that’s it and the result is pretty obvious.

There is a slight improvement from “HIGH ISO NR = off” to “HIGH ISO NR = on”, but the difference between that and the version from RAW is striking.

OK, that was easy, now let’s look at something notoriously hard, let’s look at foliage. Well, grass or fur would have been even better, but this detail from my living room window in Vienna, seen though the telephoto end of the LX5’s zoom range, is representative enough of a typical “landscape” situation. It was a gloomy, cloudy day (and cold!), so there are contrasts, but much less accentuated, there is not much differentiation in terms of color, and there is a lot of fine detail.

To the left is the JPEG with “HIGH ISO NR” set to “on”. This is a situation that is much harder for noise reduction algorithms – and it shows. On the RAW side I have used much less aggressive noise reduction, and the sharpening layer was tempered with an edge mask. You may have noticed in the last example that the colors from RAW were more vivid than from the JPEG, and so it was here, but while that looked attractive on the artificial surfaces, it looked gaudy and wrong here. Thus I have desaturated the yellows, mixed a tiny amount of magenta into the yellows and finally darkened the image slightly. In the end I got something sufficiently similar in color to the JPEG from the camera, that the eye can fully concentrate on noise and detail.

I think the result is again a win for RAW, although with this material it is less dramatic. There is more detail, yes, but there is also more noise. I’ve tried a few things, different settings of Topaz DeNoise, different sharpening, fiddled with the edge mask, but whatever I did, when I got less noise, the shadows became invariably mushy, just like in the JPEG from the camera. I don’t say it’s the optimum, but I think it’s as far as I would go for a normal Image of the Day. Every tiny bit of additional progress would have to be bought with obscene amounts of time.

So, here we are. I see an improvement possible by setting “HIGH ISO NR” to “on”, but it is not comparable to what you can reach by going through RAW and using a high-quality noise reduction program that you can parametrize according to the image’s needs.

We can also see that ISO 1600 is possible on this camera, albeit only for certain subjects. Artificial surfaces with hard edges and not much texture work best, for landscapes though I would not go higher than ISO 400.

The Song of the Day is “Compared To What” from Roberta Flack’s 1969 debut album “First Take”. Hear it on YouTube.

1351 – Darkness, Darkness

The two villages down along the coast from Sestri Levante, Moneglia and Fornaci, are only accessible from the north via two long tunnels.

Each of those tunnels is actually a succession of tunnels, with the longest one maybe four kilometers long, and the interesting thing is, that they have only one lane. There are semaphores on each side and fixed times when they go green.

Here you see the entrance to the first tunnel in Riva, a part of Sestri Levante with a long beach and some ugly industry, most likely wharves. There is a minimum speed of 40 km/h in the tunnels and a maximum of 60 km/h.

After having arrived in Moneglia, I thought that it was a shame that I had no images of the tunnel itself, so I set the camera to manual focus, high speed continuous mode, speed priority with 1/250s, f2.8 and hoped for the best.

In the next long tunnel I was prepared. I only had to raise the camera through the car’s open roof, press the shutter and see what I get. It was not even distracting from driving. But then, I probably wouldn’t do it with a policeman in sight 🙂

The image with the boats was taken in Sestri Levante, after sundown on our last evening there.

The Song of the Day is “Darkness, Darkness” from the Cowboy Junkies’ 2004 album “One Soul Now”. This is some more music that I learned to know via Bill Birtch’s blog. Thank you Bill 🙂

YouTube has the song.

1231 – Twisted II

This is an image of yesterday evening. It’s 17mm, ISO 2000, f4 and 1/15s. I’ve used Noise Ninja against the noise and Topaz Detail to accentuate local contrast, mostly in the foreground.

Other than that I have made a change to my buying habits that was long due. My last three music purchases were digital downloads. Funnily enough this came totally by chance. While I installed software on my laptop (the one that I just use while being on the train to Carinthia), I wanted to download the latest version of JetAudio. JetAudio is the proprietary music player of Cowon, the company that produces the excellent iAudio players. I own one of them, because Cowon is one of the few companies who care to support the free OGG music format, and because I have originally started to rip my CDs to OGG, I currently have about 30.000 tracks in this format. OGG was originally superior to MP3, and it is a format designed and implemented as open source. Unfortunately Apple obstinately refused to support the format in their players. Well, so did I buying their hardware. I even filed a bug report and wrote them one or two mails, but of course I got no reply and nothing has changed.

Anyway. JetAudio was my preferred software player so far. It’s pretty OK but nothing special. Think of WinAmp or similar programs. Now, as I installed software recently, the download server for JetAudio was unbearably slow, so I decided to look for an alternative. I found Songbird, a Mozilla-based iTunes look-alike with similar functionality and all the usual Mozilla features like plugins and skins. I had looked into it at least two years ago, then it had not really impressed me, but in the time since it has really grown up.

Well, here I was with a new player, and like iTunes is linked to the iTunes store, Songbird has plugins for music stores. Here in Austria it offers integration with 7digital, a UK-based service and of course Jamendo.com and Magnatune.com. Integration of Jamendo does not even need a plugin, and in fact I currently hear the album “The Green Waltz” by “The Dada Weatherman“, an album released under Creative Commons, an album that I have downloaded from Jamendo for free. For 10€ I could also buy the album in WMV format, i.e. uncompressed full CD quality. I have no idea how many people do that, but I seriously consider it. It’s excellent music.

The three purchases from 7digital were Ella’s “Twelve Nights in Hollywood”, on CD available at around 50€, as 320kBit MP3 offered for about half of it, a Greatest Hits compilation by Blondie and the debut album of Melody Gardot. Well, it looks like my plastic buying times are over 🙂

Regardless of the music that I’ve talked about so far, the Song of the Day is “Twisted“. Not the Annie Ross classic that we had in “578 – Twisted“, not Joni Mitchell’s respectable version from “Court and Spark”, but instead the completely unrelated song from Annie Lennox’ 2003 album “Bare”. Hear it on YouTube.

1167 – Vienna Waits For You

I’m on the train to Carinthia. It’s been a little complicated, the train was shorter than expected, I stood on the wrong end and had to go through the seven cars, all through to the front, but now all is well. I sit in the only car with electric current, in the right place at the window to have space for the mouse, I can work and I have an internet connection. The place was marked as reserved, so although hundreds of people must have walked by, nobody sat down. I found out though that the reservation was for the other direction, Villach to Vienna. A lot of luck, I’d say 🙂

I saw this bicycle yesterday when I left work. It looks like the bikes that those foreign, underpaid people use, who deliver our newspapers since the privatization of postal services. I have no idea why someone has sprayed “Vienna” on it, but it’s always nice to know where you are 😀

The Song of the Day is “Vienna” from Billy Joel’s 1977 album “The Stranger”. 1977?? Oh dear, I’m growing old. Hear it on YouTube.