***UPDATE See the new Topaz DeNoise 6 reviewed and compared to DeNoise 5 HERE***
Smooth, yet refined with an impeccable attention to detail.
Noise has become less and less an issue for digital photographers over the last few years as sensor technology gets better, in camera processing is capable of shouldering more of that load than ever before and digital asset management software/RAW converters are up to the task for much of the noise reduction needs. Still, with analog to digital information translation, there is an inherent signal:noise issue that can always be further helped by a good noise reduction software. Add to that, pro-sumer 35mm format models breaching the Medium Format pixel counts, or compact sensors pushing the pixel pitch to near immeasurable dimensions, there is, and will be a need for a manual noise reduction control through post processing. While Lightroom and Aperture have good noise reduction algorithms, they are the swiss army knife of image processing, giving you many handy tools, but what happens when you need a power tool? That little Swiss army knife’s mini-saw ain’t gonna cut through that noise riddled log for you, you’ll need a chainsaw. Enter, Topaz DeNoise5, your powerful, noise reducing chainsaw. You can download a free trial, or purchase DeNoise. If interested click HERE to go to Topazlabs.com. I’ve used Noise Ninja and Nik Define in the past, and I think both of those have just been pushed out of my workflow. Read on for examples and my take…
Here are 100% crops from the image above, click on any of the images to see a larger version:
Proof is in the pudding. No need for me to ramble on… Here are images processed using Topaz DeNoise 5 with 100% crops to show before and after. Click on any image to see it larger.
This first example was shot using my Canon 5D (the classic) and a Voigtlander 40mm f/2 Ultron SL lens. The settings were: ISO 3200 (H1), 1/500th sec, f/2.8, EV +/- 0
While this was a situation where I unnecessarily used ISO 3200, I still wanted to see how the files would hold up using DeNoise AND NIK Define2 which has been my go to noise reduction software up until now.
Now, I’m not sure if NIK has updated Dfine, but I’ve used it for the last few years with no complaints. I’ve lost most all interest in the NIK plugins over the last few years because A) they’re way too expensive and B) there are better comparable plugins (in my opinion), most of them for less money, and this is a good example. One thing I appreciate with Topaz DeNoise5 is how well it attacks the banding noise which has a user definable width slider to handle both vertical and horizontal banding. The NIK Dfine image to me was trickier to retain detail and it showed some pretty substantial posterization by comparison, and this was after I did my normal noise measuring and custom application within Dfine2. With DeNoise5 I started off with one of their RAW presets and adjusted a couple of the sliders. Saved that as my default “5D 3200” preset and viola. (see the screenshot lower down to see the DeNoise5 interface).
The next example was a shot using the Olympus OM-D E-M5 and 12-50mm kit lens ISO-6400, 1/40th sec, f/5.6, EV +/-0 and shows a very average example of using these amazingly high ISO settings to gain just enough room to capture kids indoors without flash. Downside? Noise. Until…..
I will say this, DeNoise5 does a great job with its presets, and further tweaks to the sliders are intuitive and effective. I do think that larger files will bog down the overall speed, and the refresh rate when making any changes, or dragging to another point in the image was sluggish, to be fair, there is a lot of information needing to be processed so I’m able to temper any frustration when I see the results. That being said, while I started with the heaviest preset available to obliterate the noise, I was still impressed with the amount of detail that it did in fact retain. As you can see in the screen shots below, there are quite a few perameters that are user adjustable. I found that going extreme on most any of these sliders, created unwanted artifacts and destruction of detail as you’d probably imagine it would do. With a light hand, I was able to bring back a little detail by decreasing the “Overall Strength” slider and slightly increasing the “Recover Detail” slider which proved to me, to be a great balance of detail and noise reduction. I then clicked on the “SAVE” preset button on the bottom left (below the preset bar) and entered “EM5 6400” so that I could easily revert back to this as a preset if and when I bring any of my other images in from this camera at this ISO setting (which I know I will be doing) and using this as a starting point.
Finally, I wanted to see how DeNoise, and Dfine handled banding. Recently I’ve written about the banding issues on the Olympus OMD EM5 in the OMD vs G3 article (click here to read) when shot at ISO6400 and I was curious to see if either of my noise reduction programs could remedy it. Now, this may be a more accurate test for any noise reduction software and a shooting scenario where you cannot really get away from shooting at very high ISO’s. This is where your darker tonal values will most certainly be challenged by noise. Small venue stage lighting is notoriously tricky and dealing with noise will always provide a task. Again, click on any of the images for a larger version.
With the above example, I tried adjusting the NIK Dfine version to the best of my ability. I’d either lose tons of detail, or I’d not get rid of enough noise. The result was the best that I could do balancing those two factors. The Topaz DeNoise version seems to me to be in all ways better, and certainly more effective with the vertical banding. The two versions took about the same amount of time, and the controls within the Topaz software were so much better and more acute. For me, it’s a closed case.
With a plugin like this, I won’t worry about shooting in situations where I may need to crank my ISO to get the exposure I need. Much of my shooting ends up in low light situations, so a good noise reduction plugin is important, and I am very happy to say that DeNoise 5 will be my noise reduction plugin of choice, replacing Nik Dfine2 in my workflow (which had unseated Noise Ninja a couple years ago for me). I feel it will be that extra little specialized bump that can help me tone down, or entirely eliminate noise all while doing a great job at detail retention. The fact that there is so much room to adjust with this software between this toning down or complete annihilation of noise makes it such a wonderful tool for me.
DeNoise5 is compatible with Adobe Photoshop CS3-CS6, PS Elements 6-10 and is also compatible with Aperture 2&3, iPhoto and Lightroom 2-4 via Topaz Fusion Express. DeNoise5 retails for $79.99 and I’d highly suggest giving it a try. With a fully functional 30 day free trial, you can see for yourself how useful it may be for you. Use the TopazLabs.com link below to go straight to the plugin page to see what’s on offer.
Get a free trial code to use the fully functional program for 30 days, or purchase DeNoise, along with any of the other awesome, affordable Topaz plugins here: TopazLabs.com
For more reviews on some of my favorite Topaz plugins click on the links below:
Topaz Clean 3, bang for your buck! Edge and texture control.
Topaz Black and White Effects, another great budget plugin.
Thanks for the read and happy shooting,
It’s coincidental you publishing this as I bought this about a month ago to use for a wedding I shot. Really, I did it in self defense as my older computer can’t practically use Lightroom’s noise reduction any more and I’m not willing to upgrade yet..
In prior versions of Lightroom, the “re-render everything with any change” behavior of Lightroom was acceptable — but with version 4 the noise reduction cause my computer to pause for very long intervals after any change once I apply noise reduction. Once I apply noise reduction in LR, all subsequent edits slow to an absolute crawl.
I’m not really sure that Topaz is any faster… except that I apply it ONCE and I don’t have to wait for noise reduction to be reapplied on any change I make to a photo after that.
I’m really glad I bought it though because as you stated it’s a specialty tool which does its job very well indeed. I’m very happy with it. It’s the first plug in I bought from these folks though I use LR Enfuse from Photographer’s Toolbox for my HDR and I have a couple from Image Trends: Shine Off and Fisheye Hemi.
The two from Image Trends are really quite useful for some special situations. Shine Off takes off hot highlights on faces and is great for those times you absolutely have no choice but to fire your flash directly at a shiny face. Fisheye Hemi semi-defishes a fisheye shot (it straightens vertical lines only) which has an amazing effect on people and makes them look natural while keeping the perspective and the horizontal curvature of the fisheye view. You have to see it — it’s quite unique. They have to be used in Photoshop or Elements though — they don’t work in Lightroom directly and there is no “bridge” interface like Topaz’s “Fusion” product to allow direct access.
Great stuff as always, Tyson, and thanks for your hard work.
I’ll have to check out the Image Trends plugins! I have acquired a digital toolbox of plugins, some of which I rarely ever use, but for those times when one of those tools is necessary, nothing else will do (most of the time anyway). I really like how affordable the Topaz plugins are compared to NIK, etc. I’ve found that there are certain tradeoffs with that price reduction in certain cases, but DeNoise is not one of the programs that asks you to sacrifice anything. It’s really pretty amazing.
Thank you for taking all the time to read through the various articles John. I really appreciate the ability to talk about anything to do with photography and digital processing. There is so much to learn and the more I can get myself to try and critically look at things, the more I find I need to learn. That I get anyone interested enough in my ramblings to start conversations that help me find cool new tools is icing.
To me, the best examples are the debanding on the curtain plus much less graininess on new BWHN’s face and the old reliable magnetic letters. That’s impressive. I’m left shaking my head at how that’s possible to remove the banding and not destroy the picture.
I realize you have to work with sliders and all but how to “recover detail”? Not sure that’s actually possible but it certainly appears so to the eye. Magic! Thanks.
I know! It is pretty amazing what modern software is capable of. When combined with what current sensor tech is providing, there are very few situations where an image should be ruined by noise, white balance or general exposure issues assuming we get a decent file to begin with.
I have just downloaded the 30 day trial and compared DeNoise5 to Noise Ninja, which I have used for years. The difference was quite staggering.
I shoot in RAW with my E-M5 and ‘Exposing to the right’ have found that the E-M5 has amazing abilities of keeping the highlight detail and, of course, removing much of the noise in not just the dark areas but mid tone plain coloured areas too. (min +1 stop but sometimes to +2.5 stops)
However there are times, for example in early morning pre-dawn shots, when noise becomes part of the equation and in the couple of tests I’ve run on such images, DeNoise5 not only cleans the noise excellently but the detail remains and doesn’t become blurred or posterised.
For that I can cope with the extra processing time.
Thanks for the pointer Tyson.
I used NN back a few years ago too. I’ve not kept up with any updates, but I can say that for me DeNoise out performs Dfine, hands down. I’m really impressed.
This is off-topic (the software looks very good, but…) are you shooting RAW? There’s no way that wine-glass image is 6400 RAW without any NR. See these shots, from E-M5 at 3200 shot in RAW, exported with minimum processing:
Whole large image (for those who want the big thing — 5.1MB):
There’s basically NO chroma noise in your wine glass shot, or the shot of the fridge and the floor. Either you’re starting with jpeg, or your raw converter is doing a lot of NR for you, OR there is something wrong with my E-M5 😉
Captured as RAW. Converted with noise reduction as my default profile for the OMD EM5 upon import in Aperture 3.4 which I have set up.
Like I mention in the article, Aperture/Lightroom, etc do a good job at handling noise, or at least the initial conversion from RAW, but DeNoise is the tool that can help retain detail while literally eliminating noise in most cases. Had I tried to get entirely rid of the luminance noise in Aperture, I’d have seen the detail in the image disappear, and this is where a specialized noise reduction plugin comes in 🙂
While I could have shown an unaltered RAW file, I was more concerned with how this plugin would work with the way I convert my RAW files, in a real world setting so to speak. The shot of my son in the jumper was artificial to an extent as I could have opened the lens up and decreased the ISO as the light would certainly have allowed for that, but the stage shot was a legit ISO 6400, lens wide open situation to get just enough light and sensitivity to get the jumping singer frozen.
Cool, I understand. Thanks for the reply.
So you’re effectively doing two passes of NR. What might the results be if you don’t do any in Aperture and just let DeNoise do everything?
I’ll have to try 🙂
My gut tells me that DeNoise will do just as well, but would be interesting to see.
Also, look at this:
See the nasty noise at 6400 in RAW?
Yes, with ACR’s NR turned off it is somewhat messy, yet better than the others (NEX7 / NX200). I think effective NR is a multi pronged attack where the first run should be when converted, albeit minimally as to not compromise any detail, or as little detail as reasonable, then, if needed a proper NR software to fine tune it.
I’ve been very impressed with the files from the OMD EM5 since I’ve got the processing set up this way, much more so now with DeNoise making decently exposed ISO6400 shots entirely useable.
Thanks for the links and comments.
As I mentioned in my previous post, the starting point of noise reduction for me is to shoot in RAW and “expose to the right” (by between 1 – 2.5stops). This is not possible in JPG and reduces considerably the noise that I have to remove using LR and DN.
I would also mention here that for static shots (still life / macro etc) I also take a number of shots and use the “Averaging” facility in my Photomatix Pro to remove the noise, almost completely. I think the extended Photoshop offers “Averaging” too.
I am experimenting with ETTR, and I think it’s a fantastic idea. I haven’t hard-set my OMD to ALWAYS shoot that way, like this man recommends:
The software I’m using (Darkroom, on Linux) doesn’t really do well at “auto exposure” back to normal settings, so am finding it a lot of work to “fix” each shot. For just a few shots (as in when I’m trying to make “art”) it’s no biggie, but I shot a wedding recently, and re-tweaking the exposure on hundreds of images just sucks.
(Yes, I realize, I should probably get more professional software if I am serious. Darkroom is actually extremely good, it just doesn’t have enough of a manual to know how to use its powerful features well)
The averaging idea is very interesting. I’ll have to try that. Thanks!
I’ve not used Darkroom, but does it not have a batch processing engine? That (along with a non-destructive, non-copying, adjustment workflow) is worth the price of Ap3 or LR4 in and of itself, although I don’t know if either program could be run on a Linux machine (I’m entirely uneducated on Linux).
For sure, it does. And it’s non-destructive; it uses the xmp files to store its “recipes” just like LR.
I hear what you’re implying; set up a preset that is 90% of the tweaks necessary to just happen on import, then tweak a little. That’s what I’d love.
The problem seems to be that there is no way (that I can figure out) to tell Darktable how to “auto expose” as recommended by Pekka in that article. But maybe if I lock my OMD to +1EV, and figure out that corresponds to “exposure -8%” or whatever, I can do that. I’ll keep working on it.
I’m not using Darkroom to save money; it’s just a practical matter. I have a Mac laptop, and a Linux workstation (for work). The latter has my color calibrated display, and is faster (and more RAM-laden) so I do my photo work there. Linux could MAYBE run Ap or Lr, but it would be wonky.
As much as I love working with free software, and as much as Darktable really is extremely high quality software, there’s basically no support: no books, no videos, no tutorials, no classes, no de-noise plugins. So I’m left just fumbling with the knobs, hoping to make stuff look nice. 🙂
Thanks for your blog.
I don’t use Pekka’s method, but prefer to use the Histogram for my ETTR. Pekka moves until he gets the red blinkies. I am more flexible in how far to go for each shot and have found that this has given me more movement to the right. I would expect a minimum of +1 stop. Anything less is not going to produce results and Pekka’s enlarged comparisons of noise seem to show rather more noise in the ETTR shot than I would expect.
I will, however run some trials using Pekka’s method to compare.
It may be that I am less ‘academic’ about colour results, preferring my human eye to tell me if a result is OK.
Also, as the training co-ordinator for my Camera Club, most of my club members have histograms either in live view or after the shot that can be used for ETTR but not too many have the blinkies.
Using my E-M5, I have done some comparisons using Histograms and Pekka’s method of ETTR (as I understand it).
I found that by adjusting the histogram upper indicator setting down to 245 as he advises, the red blinkies cut in so early that they are almost at the same point as what I would call a normal exposure, making the ETTR gain when using them almost negligible. This is probably why I saw little difference in the noise comparisons he shows in his report.
The E-M5 sensor is far more forgiving and capable than that.
Also he advises to make the contrast and saturation both -2 stops. These changes to the screen display seem to have no bearing on the appearance of the blinkies at all from what I can detect and, of course are irrelevant for RAW images.
So I will continue to set a ‘normal’ exposure and adjust to the right at least 1 stop using my histogram (with setting at 255) as an indicator until I am happy that the E-M5 will cope (using my experience).
I would suggest an exercise using a tripod, of taking a contrasty scene with some shadows and setting a good looking manual exposure in live view suitable for jpg production, capture it as a ‘master’, then capture subsequent shots as you adjust the shutter speed to expose to the right at third stop intervals up to 2 stops (6 shots). The shots will start to look very pale and washed out but stick with it.
Throw them all into a photo editor (for me that is Lightroom) and run the ‘Auto’ tone correction across them all, which theoretically gets them all back to a ‘normal’ state in which all images will look very similar and see at what shot the ETTR exposures start to lose detail in the highlights. Take that shot or the one before, note how many stops higher than the master it is, and compare the shadows noise with the master shot and see the difference.
By trialling as above you can gain an idea of how much ETTR you can use with your camera before the RAW data starts to lose the highlight detail.
Reducing the shadow noise when capturing the shot in this way means that the image is cleaner from the word go and any preliminary post process sharpening is not also sharpening the noise.
Michael, I’ve had very good results with Pekka Potka’s methodology, but I think the difference is in when the user decides when to stop cranking up the exposure.
For instance, if you only crank it until a few isolated points turn orange (as I did initially) there is very little benefit. The advantage of the 245 setting is that you can make more of the highlights orange and still not lose anything to speak of (or nothing at all).
At least, this was my experience. I needed to trust the E-M5 and crank it up more knowing that I was seeing clipping well ahead of saturation of the sensor.
The primary reason I prefer Mr. Potka’s method is simply that I don’t have to have a histogram on my display interfering or distracting from framing. Furthermore, as I’ve become better “trained” at using it, I find it very quick and trustworth for run-and-gun photography with the E-M5 where I can’t review my shots. I’m learning to use it in a way that I can trust myself and the camera.
All that being said, I don’t think either method is an absolute that can substitute for using the chosen method for lots of shots in lots of different kinds of scenes and “training” ones brain as to how aggressive you can be pushing to the right.
Which is like everything else in photography: time and experience teach you what works best for you and your style.
I thought of something else, lol. The need to use the muted setting and turning down the contrast and what not is essential to the method or the blinkies come on too soon. I don’t know if some folks thought that step was an option, but I found out it is a key adjustment to make it work.
I found that irrespective of whether the contrast and saturation were up or down didn’t affect the moment that the blinkies turned on.
They responded only to the Histogram limit being changed.
That’s an interesting finding. Since they used 8 bit limits (0-255) for the values that suggests they are setting the limits based on the rendering engine which provides the live view and not the sensor — basically the jpeg engine — hence the value would change.
The “blinkies” would be typically oriented around jpeg shooters and not RAW shooters when they are available which I believe is why Pekka did everything he could to “compress” the RAW data into the rendering that was being evaluated. By making those adjustments, you get the widest range of sensor values into the resulting rendering.
My experience was that I had no luck with this method until I changed the rendering, which I did not do at first. I wonder if we are using the same firmware? I had 1.2 from the start since I bought my camera fairly late in the campaign. I just went up to 1.5 but I haven’t shot anything to speak of since then.
But it’s mostly academic for me since I found I actually like the flatter rendering in my viewfinder anyway, lol.
Further tests I have made show that changing the saturation from -2 to +2 or changing the contrast from 0 to -2 makes no difference to when the blinkies start.
However, I have now found that moving the contrast to +2 brings the blinkies on 1/3 stop earlier, so it does affect the display. I didn’t see this before as I never use that setting and I only changed from 0 to -2 as per Pekka’s method.
Also I found that selecting Vivid, Natural or Muted made no difference but selecting portrait also brought the blinkies on 1/3 stop early (by enhancing skin tones probably)
So, adopting Pekka’s setting would ensure that blinkies were showing appropriately for those who have Portrait and/or +2 contrast selected.
The 6 shot test I mentioned previously was at 1/3 stop intervals which covers a +2 stop range.
It doesn’t use the blinkies or the histogram but should show the capabilites of your sensor when shooting in RAW.
Good idea, Michael (6 shots!)
My impression is that the changes to the viewfinder/display rendering in Pekka’ s recommendation are merely aesthetic… to make it so that what you look at in the finder (if you’re at +1.7) is watchable.
Having ‘been with’ Olympus digitally since the E-1, I consider that I have been rather spoilt in terms of live view. Even the word ‘lazy’ comes to mind.
I have always shot in RAW and 90% of the time in Manual. I would compose and set the exposure by viewing the image on the LCD screen and press the shutter. Voila!
Through the E-3 and E-5 I did the same thing and 95% of the time a successful outcome.
At that time I started to review the direction I would take (stick with E-Series; move to full frame etc…) and started to read more articles and blogs about sensors and stuff.
Then they announced the E-M5 and I made the decision to not only stick with Olympus, but to sell all my E-Series bodies and lenses and go M43.
I have the view that I now have a much smaller and lighter ‘David’ against the competition’s ‘Goliaths’ and, with that, I adopted a rather more ‘what can I get out of this camera to close the gap’ approach.
With a basic ETTR method ( I use the histogram but however it is achieved) I found that my ‘David’ has a very good slingshot and can knock down a few giants on the way. Of course, it’s not all about the equipment (he said trying to be modest) but I love this camera and its capabilites and the versatility it gives me in a hobby I am passionate about.
I really find myself in strong agreement about my E-M5.
Other than pocket cams, I never used any kind of active viewfinder until I started buying PEN’s for recreational photography. Having tried to use the live view on my D7000, my opinion was “who the hell want’s to use this!” But I broke down and bought a VF-2 for the PEN’s and it started me thinking.
You guys in the Olympus world have been spoiled (and Panasonic too, my daughter has an “old” G1) with such an excellent implementation of electronic viewfinders and live view. Regardless of whether we use the blinkies or histograms, the availability of that kind of exposure information in the viewfinder is a revelation!
When I bought the first E-M5, I already had an E-P3 for “recreational” use and two D7000’s I used for commercial stuff like weddings and so forth. When I bought the E-M5, I saw it as a replacement for one of the D7000’s. I had a bunch of Nikon lenses including several very expensive pro zooms.
But after shooting the E-M5, I sold all my Nikon stuff. I am totally sold now on the “Olympus Experience” with the the acquisition of the E-M5 and a lot of the fine primes. I have the 12, the Pany/Leica 25, the 45, and the 75 now and am blown away by all of them. I’m even a big fan of the little 9-18mm which I use for a lot of urbex and landscape work. I have the little pancakes from Panasonic as well, but I use them less and less now.
The link on my name goes to my blog where my last entry was how absolutely thrilled I am with my new camera system and leaving the “bigger iron” behind. I’m not a young man, and carrying too much bugs me, lol.
It looks as though we have followed similar paths to the E-M5.
I too have the 12, 45 and 75mm lenses and have only 10 mins ago taken delivery of a Voigtlander 25mm f0.95 which I am about to take for a drive around the block!
I will probably be getting the new 60mm macro later this month as a replacement for my much loved Four Thirds 50mm f2 macro lens, which is the only one of my old lenses that I have missed. I do love flower macros.
Next year sees the new 17mm f2 lens which will be identical to the 12mm and have the sliding focus ring which reveals the wonderful DOF scale. I only ever use that scale now capturing landscapes.
Photography has never been so much fun, and, as you say, in later life (I’m 62) that can be so rewarding when the burden is less!
PS. Tyson, thanks for a GREAT blog
Ooops…wrong. My last blog entry was on my recent vacation in the outer banks, sorry! The one below it is the one I was talking about.
I guess I’m gonna have to read through and try PP’s exposure method! I don’t know how I missed that one 🙂
Thanks for the continued conversation and insight guys.
Pingback: *50% off All Topaz Plugins! « Tyson Robichaud Photo-blography
Pingback: *40% off all Topaz Plugins through Dec 31st! « Tyson Robichaud Photo-blography
Pingback: *Against the grain on the new Olympus 60mm f/2.8 macro lens | Tyson Robichaud Photo-blography
Pingback: *Topaz Labs and 500px team up and offer discounts | Tyson Robichaud Photo-blography
Pingback: *Topaz Black and White Effects 2.1 Update, free for existing users, discounted for the rest of us! | Tyson Robichaud Photo-blography
Pingback: *Topaz Time – Huge discount on Topaz Plugin Bundle | Tyson Robichaud Photo-blography
Pingback: *The GX7 vs the OM-D E-M5, battle for my affection, Round 3 – IQ, Interface, Features and the Final Decision | Tyson Robichaud Photo-blography
Pingback: *Holiday Topaz Super Sale! | Tyson Robichaud Photo-blography
Pingback: *Topaz – Simplify 4, quick and easy digital painting filters with a click! | Tyson Robichaud Photo-blography
Pingback: *Helpful Topaz Workflow Suggestions | Tyson Robichaud Photo-blography
Pingback: *Topaz Clean, on sale through May | Tyson Robichaud Photo-blography
Pingback: *Topaz Adjust On Sale for 50% off! | Tyson Robichaud Photo-blography
Pingback: *The Entire Topaz Collection on crazy sale! $249.99 for all 15 Topaz Plugins | Tyson Robichaud Photo-blography
Pingback: *Yes! Topaz DeNoise on sale!!! 25% off through June @topazlabs | Tyson Robichaud Photo-blography
Pingback: *Topaz End of Year Sale, anything and everything 35% off w/ @topazlabs | Tyson Robichaud Photo-blography