Category: Uncategorized

  • Waterfails: We test Pentax K-1’s Pixel Shift

    Punch Bowl Falls in the Columbia River Gorge of Oregon State offered a unique setting to push the limits of the K-1’s pixel shift technology.  This is a conventional single exposure utilizing no Pixel Shift. Pentax K-1, HD Pentax D FA 24-70mm F2.8 ED SDM WR 70mm, F16, 0.3 sec, ISO 100.

    Introduction

    It’s clear that the Pixel Shift Resolution shooting modes make a huge difference to image quality from our studio tests, but how does that translate to real world shooting situations?

    In previous iterations of this feature (in the K3 II) we found that movement within a scene could cause major issues during Pixel Shift Capture in ACR conversions. So if leaves rustled in the wind or your subject moved in any way the camera compensated poorly for the movement, which then resulted in pixel blur and artifacts in the regions where the movement occurred.

    The latest iteration of Ricoh’s Pixel Shift technology found in the K-1 takes four separate images that it uses to build the final full resolution image upon and if movement occurred in subsequent images, the camera is supposedly able to differentiate the pixels that moved from the first frame it shot and clone in pixels for that area from the sharpest (for that region) of the four single (demosaiced) Raws to produce a much sharper final product then was possible in the Pentax K3 II.

    To determine just how much the technology has improved in the K-1 over the previous iterations, I took the camera out into the field to really push the limits of the Pixel Shift Resolution shooting modes.

    The Gorge

    The Columbia River Gorge, located along the Columbia River that borders Washington and Oregon state offers a variety of amazing scenic locations in addition to a number of challenging shooting environments. In order to test the capabilities and limitations of the Pixel Shift resolution mode in the K-1, I decided to focus on the seemingly endless amount of waterfalls that parallel the Columbia River about an hour or so outside of Portland, OR.

    In an effort to get the most bang for my buck I decided to hike along the Eagle Creek trail, heading toward Punch Bowl Falls. There are a number of beautiful spots to take in the views and two waterfalls (Punch Bowl and Metlako falls) to make the trip more than worthwhile. The weather conditions were, well, pretty miserable. My fiancé and I experienced moderate rain throughout most of the day which definitely put a damper (literally) on the trip.

    Most of the Eagle Creek trail is carved into some fairly steep cliffs and snakes its way back along the deep river valley that lies several hundred feet below the trail. 

    Pentax K-1, HD PENTAX-D FA 28-105mm F3.5-5.6 ED DC WR, 28mm, F6.3, 1/50, ISO 100

    Pixel Shift Real World Testing

    Our first stop was Metlako Falls located about a mile from the trail head. The access trail to Metlako is poorly marked, so it’s definitely easy to miss if you aren’t careful. I chose this waterfall because it offers an excellent opportunity to examine not only how the Pixel Shift deals with movement but also detail resolution as well.  

    The widget below gives you an idea of the detail resolution that is possible when you are able to use the Pixel Shift Resolution shooting mode. The files were shot at an aperture of F10 to avoid the effects of diffraction at smaller apertures. The two files below were shot either (1) without Pixel Shift Resolution, or (2) with Pixel Shift Resolution with Motion Correction turned on.

    $ (document).ready(function() { ImageComparisonWidget({“containerId”:”reviewImageComparisonWidget-22760863″,”widgetId”:363,”initialStateId”:null}) })

    As you can see, despite the issues with movement, there are benefits to shooting with the Pixel Shift Resolution mode turned on. The detail in the foliage is much better with Pixel Shift, and a nearly 2EV theoretical increase in dynamic range means greater latitude for Raw processing of contrasty scenes. 

    We already knew from our studio tests that in static environments with little to no movement, Pixel Shift mode allows you to attain a huge amount of noise-free detail in a single file.

    Unfortunately, at least when it comes to landscapes, finding truly static shooting environments can be difficult, if not impossible. To examine how the Pixel Shift modes handles movement, I took three separate exposures within a few seconds of each other, at the same settings, using the same lens for each test. The photos were shot with either (1) Pixel Shift Resolution mode turned off, (2) Pixel Shift Resolution Mode turned on without Motion Correction, or (3) Pixel Shift Resolution mode turned on with Motion Correction. The following are ACR conversions of the Raw files.

    $ (document).ready(function() { ImageComparisonWidget({“containerId”:”reviewImageComparisonWidget-38320267″,”widgetId”:362,”initialStateId”:null}) })

    From looking at these shots, it’s obvious that Pixel Shift Resolution shooting modes are challenged by motion of any kind. Immediately obvious in the ACR conversions above are the large patches of green and magenta artifacts in the ‘On’ shot, and cross-hatching in both ‘On’ and ‘On w/ Motion Correction’ shots. The lack of large color artifacts in the ‘On w/ Motion Correction’ shot is really just dumb luck: recall that a similar Raw above had such artifacts. Note that artifacts are drastically reduced in out-of-camera JPEGs with Motion Correction on (not shown), but the issues with the Raws might crop up whenever motion is present in the scene. 

    We continued our trip up the Eagle Creek trail to the lower Punch Bowl falls trail where you can get up close and personal with one of the most gorgeous waterfalls in the Columbia River Gorge. I decided to focus on just the Motion Correction Pixel Shift Resolution shooting mode for this test as I really wanted to see how it handled subtle water movement away from the main waterfall compared to the normal shooting mode.   

    Normal Shooting Mode (Raw scaled 50%)

    HD Pentax D FA 24-70mm F2.8 ED SDM WR, 70mm, F16, 1/4, ISO 100

    Download Raw (PEF) file

    Pixel Shift Resolution with Motion Correction (Raw scaled 50%)

    HD Pentax D FA 24-70mm F2.8 ED SDM WR, 70mm, F16, 1/4, ISO 100

    Download Raw (PEF) file

    In the above image you can see how poorly the Pixel Shift mode fared in this test. It had a great deal of difficulty even in areas where movement was a lot more subtle such as water lapping at the rocks in the lower left-hand corner of the frame. The green and magenta artifacts can be seen here along with a fair amount of pixelation in any area that showed movement during the imaging process.

    If you don’t mind JPEG output, there is some good news. You can retroactively add motion correction in-camera to an image shot as part of a Pixel Shift burst.

    After applying the Pixel Shift to the file you can export it as an out of camera JPEG. With that said, I don’t know too many landscape photographers that shoot in JPEG or want to produce out of camera JPEGs, since they’re mostly un-editable from a post processing standpoint.

    Out of Camera JPEG Pixel Shift (50% crop)

    No Motion Correction

    HD Pentax D FA 24-70mm F2.8 ED SDM WR, 70mm, F16, 1/4, ISO 100

    Out of Camera JPEG Pixel Shift (50% crop)

    Motion Correction

    HD Pentax D FA 24-70mm F2.8 ED SDM WR, 70mm, F16, 1/4, ISO 100

    There’s another benefit to Pixel Shift we briefly alluded to: since information from four Raw files are essentially combined, noise is decreased, since you’re effectively averaging 4 frames. This nearly 2 EV increase in dynamic range is evident in the example below: while the non-pixel shift file still shows respectable noise performance, thanks to the very high dynamic range sensor, the Pixel Shift file is even cleaner, particularly at 100%. 

    $ (document).ready(function() { ImageComparisonWidget({“containerId”:”reviewImageComparisonWidget-54913034″,”widgetId”:364,”initialStateId”:null}) })

    So what does this mean?

    After analyzing the results from this test we determined that there really is no difference (outside of shooting variables such as water current and wind) between the Motion Corrected Raw files and the Normal Pixel Shift Raw files with no Motion Correction applied. It appears that Motion Correction is a JPEG only process that the camera’s JPEG engine is able to apply to the Pixel Shift files (even retrospectively for Pixel Shift Raw files). 

    That explains why it appears that the Motion Corrected Raws actually looked worse in some cases than the non-motion corrected Raws in some of the examples provided above; there really wasn’t a difference in what the camera ‘did’ per se but more in the shooting conditions at hand (water motion, wind, current etc). This also explains why the motion corrected in camera JPEGs looked much better than the non-Motion Corrected JPEGs that the camera produced. 

    With that said the waterfall Raw examples just provided too many variables (wind, variable water motion and current) to really nail down a firm explanation of what we were seeing from this small sample set of images.  We are currently working on a supplemental piece with some nice concrete image examples and data that will really solidify what the differences are between the Pixel Shift Motion Corrected and Non-Motion Corrected Raw and JPEG files. 

    Conclusion

    Overall, the results of this test were honestly a bit disappointing. I think that everyone here, myself included, had hoped that Pentax was able to get the Pixel Shift movement issues we saw in the K3-II resolved, but it looks as though the company still has a lot of work to do. The amount of detail in the Pixel Shift files is markedly better than those taken with the mode turned off, but for landscape work (assuming anything short of totally still conditions) artifacts caused by movement in the scene almost negate the benefits. It should be noted that the color artifact issues can be rectified with processing software such as Silkypix, but the issues with pixel blur wherever motion occurred are still apparent even when using other post processing tools.   

    In absolutely perfect conditions, the K-1’s Pixel Shift shooting modes can offer outstanding results, but if there is a chance of any movement occurring in the frame I would definitely recommend staying away from them or at least backing your files up with non-pixel shift exposures as well. We’ll be posting more sample images from my trip to the Gorge soon, and and be on the look out for our forthcoming full review of the K-1 as well!


    Please note that all of the images published in this article were processed using Adobe Camera Raw 9.5 (unless otherwise stated). We’re currently examining the appearance of Pixel Shift files processed using other Raw converters and we’ll continue to update this article with our findings.

    Articles: Digital Photography Review (dpreview.com)

  • Photomyne app scans multiple prints at a time

    Last week we wrote about the Unfade app that lets you to scan and restore old photos using an iPhone. One of Unfade’s downsides, depending on your personal preference, is its simplistic approach that uses an almost fully automated process and hardly allows for any user input. The good news is that there are alternatives for those who want a little more control and efficiency.

    One of them is Photomyne. The app is currently available for iOS, with an Android version to be released in the near future. Its main advantage over Unfade is the ability to scan more than one print at once, allowing for faster throughput. You can simply lay out several prints on a desk or scan directly out of an album. Of course this means reduced scanning resolution compared to Unfade’s one-print-at-a-time approach, but most users of this kind of app can probably live with the image quality. 

    Once images have been scanned, Photomyne gives a few more retouching and restoring options than Unfade and sharing features are on board as well. A ‘discover’ feature lets you see other users’ scanned images in an Instagram-like image feed, as long as they have decided to make them public. Collaborative albums and cloud-only backup are currently in the development pipeline. Photomyme comes in several variants: you can either install a free version that lets you pay per session via in-app purchase, or a paid version for $ 5 that gives you a range of basic features. On top of that there is a $ 12 per year premium plan that includes unlimited saves, backup and access to your scans from desktop computers. 

    Articles: Digital Photography Review (dpreview.com)

  • Virtual Reality: It’s not just for gamers anymore

    The Nokia OZO is a state of the art VR camera.

    Virtual reality, or VR, has the potential to redefine the way we interact with images, including still images, movies, and other forms of visual storytelling. It’s already being adopted by major news organizations to take you deeper into stories, Hollywood studios that want to generate more immersive entertainment, and content creators who want to share experiences that don’t work as effectively on a flat screen.

    VR is still at a stage where it’s mostly of interest to early adopters, but it’s an exciting time to get involved with this new medium. In particular, it’s great time for photographers and filmmakers to start thinking about VR as this technology will likely impact the way we share our work, tell stories, and even remain competitive in business over the next several years.

    Timing is everything

    My first VR experience came many years ago when a technology incubator next to the molecular biology lab where I worked asked for volunteers to test a new ‘human interface technology.’ I found myself standing in a room full of computers, wearing a large headset with wires hanging out, and with something that looked like a hockey glove on my hand. A nerdy grad student was on hand to guide me through a virtual world.

    The graphics in this world consisted of nothing more than rooms full of poorly shaded spheres, cubes, and cylinders. There was no illusion of reality, but I could navigate through doors and wonder around. I later discovered that the grad student was actually from the psychology department and that I was, for all intents and purposes, a lab rat trying to find my cheese in a virtual maze. The grad student never revealed whether I did a better job of navigating the maze than the real rat, but the VR experience stuck in my mind. My gut told me it had potential.

    “I was, for all intents and purposes, a lab rat trying to find my cheese in a virtual maze…”

    As Mark Banas’ recent article discusses, and as my own ersatz rat experience confirms, VR has been around for a while and has even enjoyed some success, particularly in the gaming world. However, the technology behind VR may finally be sophisticated enough to give it a fighting chance of being useful to a wider audience, particularly photographers and filmmakers. When I discuss VR in this article I’ll be specifically referring to VR in this context – for photographers, filmmakers, and also visual storytellers.

    I’m not a hard core computer gamer or a rabid VR enthusiast; it’s likely that some of you reading this article know a lot more about VR than I do. However, I suspect I’m fairly representative of the typical photographer/filmmaker who’s followed VR from afar with a healthy bit of skepticism, waiting for someone to make a convincing argument that VR is relevant to me.

    VR has enjoyed some success in the gaming world, but as a visual storyteller I’ve been waiting for someone to demonstrate how VR is relevant to me.

    That’s not to say that I haven’t experimented with VR, it’s just that until recently it never seemed terribly compelling to me as a content creator. Almost every VR experience I tried boiled down the same basic formula:

    1. Videographer places a VR camera in an iconic location and captures video from a single spot.
    2. Viewer puts on a headset and watches video until he or she gets bored.

    At this point it seems like I’ve (virtually) stood around a lot of places: the pyramids of Egypt, next to the Eiffel Tower, Machu Picchu… you get the idea. But the key words are ‘stood around’. The experience can be interesting at first, but after about 30 seconds you’ve spun in a circle, looked up and down, and pretty much seen all there is to see.

    But that’s not what you do when you go to one of these places, in real life. You want to explore, to learn something, to understand the story of the people or the place that you’re visiting.

    Viewing an iconic place, such as Machu Picchu, using VR can be interesting. However, unless you show your audience something unique, help them understand the place, or immerse them in a compelling experience, they will quickly lose interest.

    Virtual Experiences

    For VR to gain any type of traction it needs to go beyond this ‘stand there and look around’ model – and, fortunately, it has. This was particularly noticeable at the NAB trade show in April where VR technology appeared to be everywhere. There was even a Virtual and Augmented Reality Pavilion that served as a hub for numerous VR companies, including makers of capture devices, display systems, and even content creators.

    My personal VR epiphany occurred at a technology showcase run by Kaleidoscope VR, a VR studio. In a roped off area dozens of people sat in chairs spread across the floor, each engrossed in some virtual world. What set the experience apart from most other VR demos I’ve seen was that the focus was on putting viewers into immersive stories and experiences.

    Visitors trying virtual reality at the Kaleidoscope VR showcase.

    ‘Content is King’ may be one of the most overused phrases in modern media, but it keeps getting recycled because it’s fundamentally true. Lack of good content is why VR always seemed dull or gimmicky to me in the past, but my experience at the VR Showcase proved that with the right content VR can be incredibly compelling.

    The first ‘film’ I selected was a VR experience called ‘Notes on Blindness: Into Darkness.’ Based on the audio diaries of a man named John Hull, who recorded hours of observations about how he learned to ‘see’ the world through sound after losing his sight in 1983, ‘Notes on Blindness’ isn’t, strictly speaking, an actual film. (It is, after all, an effort to help the viewer understand what it’s like to be blind.) Instead, it uses audio and 3D animations that mimic the real world.

    Each scene begins mostly in darkness, accompanied by Hull’s narration of where he is and what he’s hearing. As different sounds enter the space, he describes them and indicates directionality, using phrases like “Behind my right shoulder I hear a car starting,” or “To my left I hear somebody running,” in a way that prompts you to move your head around to look. In essence, he’s directing you even if you don’t realize it. After a while you discover that Hull is able to draw a mental picture of what’s around him based on subtle cues such as the different sounds raindrops make when hitting objects, like a window or a teacup. As he speaks, scenes are gradually revealed in a manner reminiscent of The Matrix, but also rely on your imagination to complete the mental image.

    I hope I never experience real blindness, however for the first time in my life I feel like I might have a very basic understanding of what it’s like for a blind person to try to ‘see’ the world using their other senses. The experience was more powerful than I anticipated.

    The trailer for ‘Notes on Blindness’ (above) will give you a rough idea of what I’m trying to describe here. The VR experience will be available for download on June 30 if you want to try it yourself.

    I also viewed ‘Witness 360: 7/7,’ a VR film that follows the experience of Jacqui Putnam, a commuter on the London Tube during the terrorist bombings of July 7, 2005. Shot documentary style, you see the places Jacqui went that day, including riding on the Tube itself, and hear her vivid descriptions of what happened. When she mentions something like “The person sitting next to me,” you turn and, sure enough, there’s a person sitting next to you that roughly matches her description.

    The experience was more tense than I expected. I knew what was going to happen, and yet as I stood there on the train next to everybody else – real people who just happened to be on the Tube – I kept thinking to myself ‘These people are about to die.’ The fact that I could look around and feel immersed in the situation, able to see the things Jacqui was describing, generated a visceral reaction that I’m not used to feeling while watching a documentary. It felt personal.

    Despite very different subjects and creative approaches, both these VR experiences had one critical thing in common: neither one would have worked as effectively on a flat screen. They depended on a VR environment to achieve their impact. 

    By now you’re probably asking how still photography fits into the VR world. Quite nicely, it turns out. Of course, the obvious applications are things like real estate photography, where 360º views can be critically important to attracting eyeballs. The real estate industry has been finding ways of doing this for years already, and new tools will only make the experience better and smoother. But it’s the creative possibilities that are really interesting.

    One thing photography has always been good at is closing distance, i.e. taking you to a faraway place you may not be able to visit in person. It’s the reason we know what most of the world looks like despite never having been to most of it. VR has the potential to take this a step further. In the same way that color photography allowed us to see places differently than we could in black and white, VR will allow us to see places in an immersive way that we can’t experience with a two dimensional picture. I’m not suggesting the VR is better than a still photo any more than I would suggest that a color photo is better than a black and white one. My point is that they are different, and each allows us to experience the world in a way the others don’t.

    Nice kitty. Imagine using a VR camera to put your viewer right into the middle of a pride of lions.

    Photo: Jeff Keller

    The key to VR still photography will be figuring out how to leverage the strengths of the medium. For example, most people enjoy a great landscape or wildlife photo. If you show me a beautiful Serengeti landscape with a lion in it I’ll probably love the photo. However, if you show me that same landscape in VR it might not be as compelling since I can’t see it all at once. However, if you let me stand right in the middle of a pride of lions eating a wildebeest you’ll get my attention, because that’s something I haven’t experienced in a normal photo.

    What all of these examples highlight is how important it will be for artists to take new and different approaches to capturing, editing, and presenting their work. It’s an open canvas, and one that’s still largely undefined.

    VR Requires New Grammar

    As Oscar-winning cinematographer Emmanuel Lubezki told us a couple months ago, the grammar and language of VR have yet to be written. This is true in a literal sense. Words like ‘framing’ and ‘panning’ simply don’t apply to VR. New words are needed, and nobody has agreed on what those words are yet. The one thing they do agree on is that VR will require different approaches appropriate to the medium.

    This can be seen in the films described above, particularly ‘Witness.’ For example, the conventional documentary formula is to intercut interview footage with b-roll, but that never happens in ‘Witness.’ The convention works in flat films because you can lock the viewer into a rectangular frame and demand their attention. But what happens when the viewer has the freedom to look anywhere they want? Maybe they will get distracted by a picture hanging on the wall, or something happening outside a window. ‘Witness’ solves this by relying entirely on voiceover while featuring a few location shots of Jacqui Putnam throughout the film. This is just one example of where traditional filmmaking techniques don’t translate easily to VR, and there are many others.

    It isn’t the first time content creators have faced this challenge. In the early days of television, studios often tried to repackage shows made for radio into TV, such as American soap operas. Daytime soap operas on radio were aimed at homemakers who could listen while working around the house. Studio executives had reservations about whether soap operas would even work on TV since they would require the homemaker to actually watch a screen.

    Early soap operas were produced for radio; when TV came along producers had to figure out how to take advantage of the new medium.

    By Photo by G. Nelidoff, Chicago, for CBS/Columbia Broadcasting Company. (Library of Congress) [Public domain], via Wikimedia Commons

    Some production elements from radio didn’t translate well to TV. For example, producers had to re-think product placement and advertisements in shows – the very reason for the existence of soap operas in the first place – because having an actress pick up a box of laundry detergent and talk about its virtues in the middle of a scene just didn’t seem believable on TV. It took a few years before the industry perfected the formula.

    The reason I point this out is because we’re still in the early days of VR. It’s easy to look at VR as it exists now and think of it as a gimmick, a tool for gamers, or a toy for tech nerds. And that’s OK – people thought similar things about TV at one point, but once content creators figured out how to effectively use the medium there was no turning back. I suspect the same will be true of VR: once the language of VR is fully developed, and hardware for consuming content becomes more convenient (it will), there’s a lot of opportunity to do creative things that may not work on a flat screen.

    Why VR is not 3D Television

    I mentioned above that VR seemed to be everywhere at trade shows like CES and NAB this year. That’s an encouraging sign, but it’s worth noting that ubiquity of a technology at an industry trade show does not equate to commercial success. I need only mention 3D television to make my point, and several people have dismissed VR to me as just ‘the next 3D TV.’

    I believe VR is a much more promising technology than 3D television, and will ultimately be more successful, for a couple of important reasons.

    One advantage of VR compared to 3D TV is that viewers can at least try it with an inexpensive viewing device, such as Google Cardboard, and a smartphone.

    3D television struggled with a classic chicken-and-egg situation. Networks were reluctant to invest in infrastructure to produce 3D content without some assurance that there would be a critical mass of audience; consumers were reluctant to invest in $ 1,000+ devices without some assurance that content would be available. For studios, this was potentially a very expensive experiment that carried a lot of financial risk. Also, many consumers had only recently upgraded to HDTV, and it was a tough sell to convince them to invest in new hardware so soon. By now everyone knows how this ended.

    The stakes around VR are different. First, most VR content is being distributed through platforms like YouTube or on mobile devices where production standards are less stringent than for broadcast television. VR also has the advantage that the lowest common denominator for viewing content is the smartphone, meaning that most consumers already have a screen on which to watch content (alone or when combined with an inexpensive viewing device).

    The lowest common denominator for viewing VR is the phone most of us already have in our pocket.

    Photo: Dale Baskin Photography

    On the production side of the equation, low cost capture devices ranging in price from a few hundred to a few thousand dollars are easily accessible to content creators. It’s not an unmanageable risk for someone running a successful YouTube channel, or even an indie filmmaker, to invest a small amount of money to try the technology. Similarly, it’s easy to envision news media such as The New York Times, USA Today, or even your local TV station sending reporters into the field with a $ 1,000 VR camera to bring immersive experiences to their apps and web pages.

    In one particularly telling move, last November The New York Times sent free Google Cardboard to all of their print subscribers – over one million of them – to insure that they could use The Times’ new VR app. The Times followed up a few weeks ago by announcing that they would also send free Cardboard to many of their digital subscribers as well. When the barrier to entry is so low that a content producer can afford to give all their subscribers a free device on which to consume their content it’s a great indication of how accessible the technology can be.

    In November the New York Times sent Google Cardboard to all their print subscribers, and a few weeks ago announced similar plans for digital subscribers.

    Are we there yet?

    At this point I probably sound pretty enthusiastic about VR – which I am – but I’ll also provide a reality check and let you know that we’re not quite there yet in terms of the technology.

    VR depends on belief; the belief that you’re somewhere you’re not. One of the things you figure out really fast with VR is that in order for it to be believable, every part of the experience must work. This includes image quality, the general viewing experience, audio, and even the space you’re in and how you interact with it. If any part of the experience is incomplete or breaks, then the experience becomes less believable.

    As photographers, you’ll notice this immediately when it comes to picture quality. We’ve become spoiled by high resolution, high dynamic range sensors that are almost magic relative to what we had just a decade ago. VR cameras aren’t there yet. Resolution is limited (usually 4K, but spread across the entire 360 degree field of view), highlights may be blown or shadows lost, and the richness of color we’re used to just isn’t there. However, in the same way that early digital photographers managed to create great photos with 2MP and 3MP cameras, VR content creators are finding ways to work within the limits of their tools.

    One of the most accessible VR cameras available today is the Ricoh Theta (above) which lists for $ 260. Alternatively, the Nokia OZO (seen at the top of the article) lists for $ 60,000. The one thing they both have in common is that neither captures the same resolution, dynamic range, or richness of color we’re used to with modern ‘2D’ cameras.

    Viewing devices will need to improve as well. Not only are they large and unwieldy, but the thing that makes VR so accessible – your mobile phone – is also one of the bottlenecks to the experience. Magnified by VR lenses, video looks pixelated and low resolution by today’s standards, sometimes exhibiting a ‘screen door effect’ similar to looking through a mesh screen. When Sony introduced a smartphone with a 4K display I initially thought of it as marketing overkill. In retrospect, I don’t know if they had VR in mind, but VR is a case where a 4K phone really could provide an improved experience. Suddenly, I like the idea of a 4K smartphone.

    Audio is a much bigger challenge. There’s an old adage in filmmaking, which I’ve discussed on the site before, which says that an audience will forgive a bad picture, but they won’t forgive bad sound. That’s actually more true in VR than on a flat screen because having an immersive experience is absolutely dependent on it. Audio for VR is in its infancy, and spatial audio that matches what you see, including directionality of sound as you move around, is critical to creating a believable experience. One studio executive I spoke with at NAB told me the biggest challenge in creating believable sound for VR is that 70% of viewers don’t actually wear headphones when watching VR, but instead rely on the speaker built into their phone, insuring a suboptimal experience.

    Immersive audio, including directional audio, is crucial for VR experiences that include sound. One challenge for content creators is that 70% of VR consumers today use only the tiny speaker built into their phone.

    Finally, there’s the disconnect between the virtual world and the real world. Its frustrating when you’re immersed in a VR experience at NASA Mission Control and you reach for the control panel only to find empty space, or when you’re standing in an open field but inadvertently bump into a wall. Some of this will be addressed through technology, but it’s also one of the challenges content creators will have to address through creative choices.

    The good news is that all of these things are solvable problems, and smart people are working on them.

    The Future is Now

    As exciting as VR is, I’m not suggesting that it will replace traditional, two dimensional media such as photography or television. There’s plenty of room for both mediums to exist side-by-side.

    I recently shared this thought with a friend, who thoughtfully responded “Then what type of stories are best told in VR versus other mediums?” After thinking about it I realized this was probably the wrong question to ask. After all, if you replace the word VR in that sentence with any other form of media, like TV or print, it doesn’t make sense. You can tell a story with any medium, but the challenge is figuring out how to leverage the strengths of each for maximum impact. In that sense, VR is no different.

    Also, as much as I’m excited by VR, I really don’t want to have complete control over my viewing experience for everything. I want master filmmakers and photographers to craft a story in their image, or to show me the world as they see it, without necessarily giving me the freedom to mess it up. I can’t imagine how The Godfather would be any better if I had the freedom to look around the scene instead of watching it the way Francis Ford Coppola shot it. On the other hand, I look forward to as-yet-uncreated projects that allow me to participate more freely in the experience.

    What’s potentially most exciting are the VR applications that haven’t been invented yet. I can’t wait for the day when NASA puts VR cameras on landers going to Mars or the moons of Saturn, allowing me to stand virtually on the icy surface of Titan, gazing out over a methane sea.

    And now if you’ll excuse me, I’m off to make my first VR film.

    Articles: Digital Photography Review (dpreview.com)

  • Asus announces Zenfone 3 Deluxe with stabilized 23MP camera

    Asus has launched three new models in its Zenfone line, out of which the Zenfone 3 Deluxe is arguably the most interesting to mobile photographers. It comes with an impressive camera spec sheet that includes a 23MP 1/2.6-inch Sony IMX 318 sensor, F2.0 aperture, 4-axis optical image stabilization, electronic video stabilization in video mode and an AF-system that combines contrast-detect, phase-detect and laser technologies. At the front there is an 8MP sensor with F2.0 aperture but video shooters will have to make do without a 4K mode. 

    Under the hood the Android OS is powered by a Snapdragon 820 chipset and up to an enormous 6GB of RAM. Storage ranges from 32 to 128GB and is expandable via a microSD slot. A fingerprint reader, 3,000mAh battery with QuickCharge technology and a Type-C USB connector are all features you would expect on a current flagship device although the 1080p resolution of the 5.7-inch display cannot quite keep up with the Quad-HD displays of most competitors. All the technology is nicely wrapped up in a full aluminum unibody with “invisible” antenna lines that don’t disturb the overall design language. 

    Along with the Zenfone 3 Deluxe Asus has also launched the standard Zenfone 3 that comes with a less powerful chipset, a smaller 5.5-inch display and a 16MP camera. The Ultra model ups the screen size to a massive 6.8-inches and features the same camera as the Deluxe. No information on availability has been released yet but the Zenfone 3 will cost you US249, the Ultra will set you back $ 479 and the Deluxe is the most expensive new model at $ 499.


    Press release:

    Taipei, Taiwan (30th May, 2016) — ASUS Chairman Jonney Shih took the stage today during the Zenvolution press event at Computex 2016 to unveil Zenbo, the first ASUS robot, along with a stunning portfolio of third-generation mobile products designed to provide users with revolutionary functionality for pursuing their passions. The incredible line-up includes the all-new ZenFone 3 family, featuring ZenFone 3 Deluxe, the new flagship ASUS smartphone with advanced camera technology that takes mobile photography to the next level; ZenFone 3, a feature-packed smartphone that brings premium design and empowering performance to users; and ZenFone 3 Ultra, an incredibly powerful smartphone with a 6.8-inch Full HD display that excels at entertainment. Also announced were ZenBook 3, an ultra-sleek and lightweight notebook with a premium aluminum design, along with ASUS Transformer 3 and ASUS Transformer 3 Pro, the world’s most versatile PCs that feature an unrivalled combination of mobility, convenience, and expandability.

    While revealing ASUS Zenbo, Chairman Shih said, “For decades, humans have dreamed of owning such a companion: one that is smart, dear to our hearts, and always at our disposal. Our ambition is to enable robotic computing for every household.”

    Joining Mr. Shih on stage, Intel’s Corporate Vice President and General Manager of the Client Computing Group, Navin Shenoy said, “For nearly thirty years, Intel and ASUS have been collaborating to bring some of the most innovative PCs and devices to market. We are excited to continue that collaboration on the new ZenBook and Transformer 3 family powered by Intel Core processors, and we look forward to working closely with ASUS on expanding beyond traditional clients into new, emerging markets like robotics.”

    ASUS Zenbo, the ZenFone 3 Series, the ASUS Transformer 3 Series, and a line-up of other all-new ASUS products are on display at the ASUS showroom at the Nangang Exhibition Hall at Taipei World Trade Center. Visitors to Computex 2016 are invited to visit the showroom to experience the revolutionary functionality of these latest ASUS innovations for themselves.

    ZenFone 3 Deluxe — World’s First Full-Metal Smartphone with Invisible Antenna Design

    ZenFone 3 Deluxe is the flagship model of the ZenFone 3 family and the ultimate expression of ASUS smartphone design. It is constructed with a strong and light aluminum alloy unibody, and has a rear surface free of unsightly antenna lines and an ultra-thin 4.2mm edge.

    ZenFone 3 Deluxe features a 5.7-inch Full HD (1920 by 1080) Super AMOLED display with a gamut of over 100% NTSC color space for rich, vibrant colors, even in harsh, outdoor lighting. An ultra-thin bezel gives ZenFone 3 Deluxe a 79% screen-to-body ratio for a maximized display in a compact package. Inside, ZenFone 3 Deluxe has a powerful Qualcomm® Snapdragon™ 820 Series processor, Adreno™ 530 GPU, and integrated X12 LTE modem, as well as up to 6GB RAM to deliver the best performance and fast connectivity for demanding apps, games, and media.

    ZenFone 3 Deluxe raises the bar for mobile photography with its incredible 23MP camera featuring the latest Sony IMX318 image sensor, a large f/2.0 aperture lens, and 4-axis optical image stabilization for high-resolution, blur-free, and low-noise photos in almost any lighting condition. It also features 3-axis electronic image stabilization for steady 4K UHD videos. Coupled with an ASUS TriTech autofocus system that automatically selects 2nd generation laser, phase detection, or continuous autofocus to provide accurate and nearly instant 0.03-second focusing and subject tracking, as well as exclusive PixelMaster 3.0 technology, ZenFone 3 Deluxe captures truly stunning photos and videos.

    ZenFone 3 Deluxe has a built-in fingerprint sensor that’s perfectly positioned on the rear of the phone to sit underneath the user’s finger and unlocks the phone in just a fraction of a second. Quick Charge 3.0 technology reduces battery recharge times and a reversible USB 3.0 Type-C port that makes connecting charging and accessory cables effortless.

    ZenFone 3 Deluxe also excels at audio with its powerful five-magnet speaker and NXP smart amplifier that provides clear, defined sound and also protects the speakers from damage. When listening over certified headphones, users can enjoy Hi-Res Audio (HRA) that provides up four-times-better sound quality than CDs. 

    ZenFone 3 — Agility, Beauty, and Clarity

    Winner of a Computex 2016 d&i Award, ZenFone 3 is a feature-packed smartphone that brings premium design and empowering performance to everyone. Built around a gorgeous 5.5-inch Full HD (1920 by 1080) Super IPS+ display with up to 500cd/m2 brightness, ZenFone 3 delivers an incredible visual experience that makes apps, videos, and games look their best. With a narrow bezel, ZenFone 3 provides a 77.3% screen-to-body ratio for a maximized viewing area in a slim and compact body. The front and rear of the phone are encased with scratch-resistant 2.5D Corning® Gorilla® Glass that gently curves to make the edge of the phone completely smooth.

    ZenFone 3 is equipped with a 16MP camera with ASUS TriTech autofocus that automatically selects 2nd generation laser, phase detection and continuous auto focus to achieve precise focus in just 0.03 seconds, resulting in sharp images in any condition.

    ZenFone 3 is the first smartphone worldwide to be powered by the new Qualcomm Snapdragon 625 octa?core processor — the first Snapdragon 600 Series processor with 14nm FinFET process technology, an integrated X9 LTE modem, and 802.11ac MU-MIMO Wi-Fi connectivity — PC-grade graphics and up to 4GB RAM that together deliver outstanding mobile performance with improved efficiency and battery life. ZenFone 3 has a built-in fingerprint sensor that’s perfectly positioned on the rear of the phone to sit underneath the user’s finger and unlocks the phone in just a fraction of a second. 

    ZenFone 3 Ultra — Unleashed, Unlimited, and Unrivaled

    Winner of a Computex 2016 Best Choice Golden Award, ZenFone 3 Ultra is a smartphone designed for multimedia lovers, featuring a 6.8-inch Full HD (1920 by 1080) display with a 95% NTSC color gamut for rich, vibrant images even outdoors in harsh lighting. It is the world’s first smartphone to have ASUS-exclusive Tru2Life+ Video technology, which harnesses a high-end 4K UHD TV-grade image processor to optimize every pixel in each frame before it is displayed, resulting in superior contrast and clarity. ZenFone 3 Ultra also excels at audio with its two new powerful five-magnet stereo speakers and a NXP smart amplifier that provides clear, defined sound and protects the speakers from damage. When listening over certified headphones, users can enjoy Hi-Res Audio (HRA) that provides up to four-times-better sound quality than CDs and the world’s first smartphone with virtual 7.1-channel surround sound with DTS Headphone:X.

    Like ZenFone 3 Deluxe, ZenFone 3 Ultra has an incredibly slim and elegant full-metal unibody chassis — the world’s first to have no antenna lines. An ultra slim bezel gives ZenFone 3 Ultra a 79% screen-to-body ratio, maximizing the viewing area while minimizing its overall size and weight. ZenFone 3 Ultra is equipped with the same high-resolution 23MP camera with ASUS TriTech autofocus system as ZenFone 3 Deluxe. Powered by the Qualcomm Snapdragon 652 octa-core processor, Adreno 510 graphics, and up to 4GB of RAM, ZenFone 3 Ultra delivers outstanding mobile performance. A built-in fingerprint sensor is perfectly positioned on the front of the phone beneath the user’s finger and unlocks the phone in just a fraction of a second.

    ZenFone 3 Ultra also has a high-capacity 4600mAh battery for long-lasting performance and Quick Charge 3.0 technology for rapid recharge times. ZenFone 3 Ultra even works as a power bank with 1.5A output for quickly charging other mobile devices.

    Articles: Digital Photography Review (dpreview.com)

  • Never miss a video: Subscribe to DPReview on YouTube

    We’ve been producing more video content than ever before, including tons of content from our last year’s PIX show, our ongoing series of long-form Field Tests, overviews of the latest cameras and lenses, as well beginners’ technique guides and interviews. We post videos right here on our homepage when they’re first uploaded, but the best way of not missing anything is to subscribe to DPReview’s channel on YouTube.

    We’ve organized our content into playlists, so you can head straight for the stuff that most interests you, whether that’s long-form gear reviews or interviews, short overviews of the latest cameras and lenses, or beginners’ technique guides. 

    Articles: Digital Photography Review (dpreview.com)

  • To St. Helens and Back: Olympus TG-Tracker Shooting Experience

    Olympus has been in the rugged camera business for a very long time, with its first model, the Stylus 720SW, released way back in 2006. Ten years later the company has made the leap to action cams. 

    The TG-Tracker is a camcorder-shaped device that can capture 4K/30p and 1080/60p video as well as timelapses. The F2 lens has a whopping 204° field-of-view ‘on land’ and 94 degrees when you take it diving with its included underwater lens protector. It features a 7.2MP, 1/2.3″ BSI CMOS sensor paired with the company’s latest TruePic VII processor. (If 7.2 Megapixels sounds a bit low for 4K, you’re right – the camera has to interpolate in order to produce 4K as well as 8MP stills.)

    The TG-Tracker captures every data point you could possibly want from an action cam.

    Design-wise, there are two things that stand out. First is the camera’s flip-out (but non-articulating) 1.5″ LCD, which is mainly used for menu navigation. Second is what Olympus calls a built-in ‘headlight,’ capable of projecting up to 60 lumens of light.

    What really makes the TG-Tracker unique, as its name implies, is tracking. It records location, altitude or water depth, temperature, orientation, and acceleration. All of this data is shown on graphs in the app, allowing you to see the pictures you took at a certain altitude or in a specific area of the map.

    There are two other neat tricks the camera can do thanks to all these sensors. First, if the accelerometer detects a sudden change in equilibrium, it will put a chapter marker in your videos. Also, the TG-Tracker can detect when the camera goes underwater and switch the switch to the appropriate white balance setting.

    All of this metadata is viewable in the Olympus Image Track app, which is where you can preview your photos and videos and then transfer them to your mobile device (save for 4K video)

    To see how the TG-Tracker functions in the real world, we sent it to Mount St. Helens, an 8363 foot-tall stratovolcano most famous for its major eruption in 1980. But before we get into that, let’s take a look at the design and what it’s like to use this action camera.

    Articles: Digital Photography Review (dpreview.com)

  • Looking Sharp: A focus stacking tutorial

    This was taken in the deserts of Southern Utah at Snow Canyon State Park near St. George, Utah. 

     Photo: Chris Williams Exploration Photography

    Background: Understanding the Process

    One of the questions that I receive quite often from beginner photographers is “How do I produce tack sharp images from front to back in challenging shooting conditions or in cases where the foreground fills up nearly half the frame?” The answer in short is to utilize a process known as focus stacking. The answer is simple but the process can be very labor intensive from a shooting and processing standpoint.

    Even when working with a large depth of field there may be occasions where you still have to take more than one image due to low light, wind and or large foreground subjects. Most technique books will tell you that this is easily accomplished by stopping down to a very small aperture such as F22, focusing on a set focal point at about 1/3 of the way into the frame or determining the spot from your hyperfocal distance (the distance between a camera lens and the closest object that is in focus when the lens is focused at infinity), focus at that spot and take your image.

    Here’s one of the images that I used during focus stacking.  This was shot at 35mm and at f/2.8 due to conditions at the time of shooting.  As you can see it was impossible to achieve sharp focus throughout the image.

    In principle this sounds like a very quick and easy fix to a somewhat complex problem that plagues most landscape photographers. In practice, however, it comes with a couple of big issues. Not only can using an aperture this small decrease image quality due to diffraction, it doesn’t actually deliver the maximum possible depth of field. That’s where stacking comes in. In the digital age we can now improve upon this technique and produce higher quality, tack sharp images from the front to the very back of the frame.

    Before diving into this process I should mention that you may not always need to use this technique and this process is really up for interpretation in regard to what you define as a ‘sharp image’. The lens quality and aperture play a huge role in whether or not you wish to go through the work of focus stacking an image. In general, I always use this process now as I want my images to look tack sharp even when printed at very large sizes.

    Selecting the Aperture

    Choosing the aperture is an important first step to this process. I always try to choose an aperture in the ‘sweet spot’ of the lens. This is a bit of a loaded term; you have to decide whether you are after maximum resolving power or greater overall focus in your image. These differences may be subtle in some cases, but regardless I always try to choose optimal sharpness over depth of field in an individual image, bearing in mind that I’m going to be stacking multiple images. I would rather take a few extra shots to ensure that the RAW files are of the highest quality I can achieve given the conditions.

    Generally speaking, the sweet spot of a lens is about 2.5 to 3-stops from the maximum aperture. This does vary from lens to lens however; for example, I normally shoot between f/8 and f/11 when possible on my Canon 16-35mm f/2.8L ii to achieve the sharpest results possible (which I know is a few stops greater). The bottom line is to get to know your lens! You can even run it through an aperture progression and compare the images to find that coveted lens ‘sweet spot’.

    Behind the Lens

    Focus stacking can be very tricky when you’re out in the field. I use the following steps to ensure that I don’t miss a focus point and to make adjustments on the fly in changing conditions such as light, wind and rain.

    • Find your desired composition and make sure that your tripod is in safe position that won’t allow for any movement or shifting while you’re performing the in camera focus stacking.
    • Place your camera on your tripod, turn on live view, switch your lens to manual mode, make sure that your camera is set to manual and dial in your composition.
    •  Once you have the composition dialed in lock your camera down on your tripod and make sure that your ball-head or camera mount is completely locked down so no movement can occur during this process (if some movement does occur you can try to correct this using Auto-Align in Photoshop, but I always try to avoid this to the best of my ability).
    • Use either your camera’s in camera timer (set for 2-10 seconds) or use a remote shutter trigger to avoid any camera movement issues.
    • Once you are 100% sure that you’re happy with the composition it’s time to adjust your settings; I always fire a few test shots to ensure that I can freeze the foreground subject (flowers etc.) and to see what I can get away with in regard to ISO/Aperture/Shutter-speed while still yielding an acceptable result with respect to proper exposure and the signal to noise ratio.
    • This may sound trivial but play with your settings a bit to find the right exposure/sharpness balance; aim to keep the ISO at base and the aperture as close to the ‘sweet spot’ as possible (f/8-f/11 in most cases) this may not be possible depending upon conditions, so make adjustments as you see fit.
    • Adjust your CPL or lens filter (ND etc.), if you’re using one, to give the foreground more pop etc.

    The Progression

    Now comes the fun part: how do you make sure that you have everything in focus and that you don’t miss a focus point? There’s definitely more than one way to go about doing this, but I generally focus on the foreground elements that are closest to the lens; normally near the bottom 1/3 of the frame in live-view.

    I normally zoom in to my area of interest in Live-View and dial in my focus point using manual focus.  After taking the image I repeat the process for my next focus point.
    • While in live-view, zoom in to the bottom most portion of your foreground subject and manually adjust the focus until everything is sharp and adjust your settings depending upon conditions (wind etc.)
    • To find your next focus point stay in live-view, zoom in to the same area you just photographed and move up in the frame to find your next focus point
    • Repeat this process until you reach the background elements in your image and take your final exposure.
    • You may want to bracket your last exposure to keep the highlights and shadows from clipping or to catch a sunstar; if you’re shooting on a camera that has a great deal of dynamic range (like a Sony a7r/ii or Nikon D810) then you may be able to do this with one exposure; it’s completely up to you

    The toughest thing about focus stacking in the field is dealing with changing conditions such as wind, rain and light. When wind is an issue I always run through at least two or three focus stacking progressions to ensure that I have a sharp frame at each focus point. Even in perfect conditions I still run through a focus point progression at least twice to ensure that I haven’t missed a point. There’s nothing worse then getting home after a long day of shooting only to find that you completely missed a focus point.

    Once you take your camera off of your tripod or move your tripod, it will be a huge pain to get everything lined up again and you may not even be able to in most cases. Double and triple check your exposures to ensure that you’ve got all of your focus points nailed down before moving your camera and packing up your gear.

    Articles: Digital Photography Review (dpreview.com)

  • Huawei P9 camera is nice, but it’s still a smartphone camera

    To dismiss the Huawei P9’s imager as ‘just another smartphone camera’ would be unfair. Its dual 12MP cameras aren’t your average dual cameras – Huawei claims that by using one monochrome and one RGB sensor, the camera is 100% more sensitive to light than your standard smartphone camera. It offers Raw capture, full manual exposure control and laser-assisted autofocus. Oh, and Leica had a hand in designing the module. All-in-all, there’s plenty going for it camera-wise. 

    We approached the P9 eager to give it a try, but with a healthy dose of skepticism. And keeping our expectations measured, we were pleased by the Huawei P9’s results. For one thing, shooting with is a nice experience. There are multiple grid overlay options, a level gauge (praise hands emoji!) and you can easily toggle between full manual and full auto shooting. It’s also possible to shoot exclusively with the monochrome camera, which opens up other creative possibilities.

    The P9 offers many of the comforts of an advanced point-and-shoot. So how’s the image quality? It’s pretty much as we expected. Good light JPEGs look just peachy on phone-sized screens, but at 100% they don’t look exactly ‘100 emoji’ 100. Grass turns mushy and halos start appearing around high contrast edges. Things obviously get quite noisy by the time you get to ISO 3200.

    $ (document).ready(function() { ImageComparisonWidget({“containerId”:”reviewImageComparisonWidget-42824869″,”widgetId”:365,”initialStateId”:null}) })

    But wait, there’s Raw! You can see some of the benefits in terms of detail retention in the studio scene widget$ (document).ready(function() { $ (“#imageComparisonLink2550”).click(function() { ImageComparisonWidgetLink(2550); }); }), but definitely check out the difference it makes in some of our real-world samples. It should be noted that Raw capture is only available in Pro mode, and can’t be used when shooting with the monochrome camera.

    TL;DR, I’d be pretty happy if the Huawei P9 camera was the camera I had with me all the time. But the dual camera system hasn’t solved the low light troubles inherent to small sensors.

    Articles: Digital Photography Review (dpreview.com)

  • Critical eye: Picturesqe aims to highlight your best frames and throw out the worst

    Picturesqe is a software application that’s designed to speed up the process of selecting the best image from a sequence of frames and the most successful pictures from a day’s shooting or a project. Load your images into the program and it applies artificial intelligence and machine learning to pick the best shots based on its own criteria and common preferences in photography.

    The idea is that the program analyzes the pictures as you import them, highlighting some and rejecting others, in an automated process that takes the strain out of determining which are the frames to use. All you have to do is review those Picturesqe selects and chose the best of three, for example, instead of the best of the 26 you shot in that sequence.

    I have to say at this very early stage that I’m not very keen on a machine making artistic decisions for me. It’s enough to battle the wants and tastes of a pushy modern camera without then allowing a canned algorithm to replace the subtle process of appreciating line, form, focus, light, composition and atmospheric exposure. That is quite a lot to expect a computer program to do, no matter how ‘intelligent’ it is. Nevertheless, I endeavored to give Picturesque a shot.

    What it does

    The beginning of the process involves importing your images into Picturesqe, which sends thumbnails to its servers to create groups and then determine which pictures are the best of each group.

    Picturesqe is a desktop application that runs on PC and Mac platforms, using support from the cloud to run its more power-intensive tasks. Once installed the program invites the user to import images either from an external source, such as a memory card, or from files already stored on the hard drive. As images are imported they are analyzed so that they can be formed into groups of similar-looking images and then arranged in order so that the best frames of each group are positioned at the top left of the screen.

    The process of grouping and ordering is done via an algorithm that exists in the cloud – or Picturesqe’s servers – so thumbnail images are sent from your machine for inspection and the information gleaned returns to Earth to inform the arrangement of images in the Picturesqe user window.

    The import process doesn’t actually copy files from the memory card onto your main desk drive but acts as a filtering system in between the two – a sort of staging post – so that you’ll eventually only have to save the best of your images. This way you’ll stop clogging your machine with pictures that will never be used and never seem to get deleted.

    Does it work?

    I imported the contents of a memory card into the software and it sorted the 460 image into groups according to content and what it determined would be the pictures I would want to choose the best of. Pictures it couldn’t find groups for, and lone images, are left floating freely. While most grouping exercises go well there are exceptions, such as this group in which the images clearly do not belong together.

    I found the results of an import and a dose of analysis to be a little mixed. At first I was impressed that Picturesqe was able to divide the contents of a memory card that contained street images, portraits and product shots into a number of mostly sensible groups. Images that feature the same color in about the same place get grouped easily, and those that contain the same objects in approximately the same composition are also bundled together with a decent degree of accuracy. Pictures that fall between stools are left ungrouped for us to leave floating on their own, to manually delete or to add to the group we think most appropriate.

    Opening a group reveals what the program thinks of your pictures, as it orders them according to its perception of their merits and labels those it doesn’t think much of with a waste paper basket. Those with a sensitive nature should avoid this stage, as the program’s mathematical opinion will not necessarily reflect everyone’s perceptions of their artistic brilliance.

    I had to remind myself that it was a machine I was dealing with so I didn’t have to take things personally

    In some instances I was glad of Picturesqe’s help in selecting the best frames from each group, but in others I was left slightly mystified by the way my images had been treated – like entering a camera club competition. Pictures that I thought were the better of the group were often not placed ahead of others I thought less pleasing. On more than a few occasions pictures I had previously selected for printing were marked out for shredding by the algorithm in the cloud. I had to remind myself that it was a machine I was dealing with so I didn’t have to take things personally. I hope Picturesqe has good insurance to pay for user’s trauma therapy. 

    I wouldn’t have grouped all of these images together as, while they are all from the same shoot, they represent the pool from which I’d hope to pull three quite different pictures. It doesn’t make sense, to me at least, to group uprights and landscape format images together.

    The criteria for the grouping process is a little more open than I’d like as, for instance, it doesn’t seem to take into account the orientation of the image. When shooting a portrait, for example, I’ll shoot poses and locations in upright and landscape formats, and generally edit to offer the sitter one of each. To me then it doesn’t make sense to group uprights and landscape format images together as though you’d pick one or the other instead of one of each. On the whole though the grouping process does a pretty decent job of working out which pictures belong together, and will only rarely drop an odd frame into a group in which it clearly doesn’t belong.

    Groups can be edited, of course, once the program has made them, and we can reject individual images from the group so that they float free in the main browser window. A process of drag-and-drop then allows us to manually create our own groups by bunching images together.

    This is how the images look when they are first imported into the program – as they should. Picturesqe though then drops the camera-dictated color, contrast and styles for a much more ‘Raw’ looking image. Removing image characteristics doesn’t make picking a winner easier.

    The application works with Raw and JPEG files, and I was impressed it is able to display the latest Raw files from the Panasonic Lumix DMC-GX80/85 even before Adobe can. It does it though at the expense of any in-camera processing settings so we see the Raw ‘Raw’ data with distortions that are usually corrected before we get to view the picture. Anything you’ve shot in black and white using Raw will also be displayed in color, though for a short few seconds images are shown as we intended before Picturesqe renders them flat and with native colors. Thus we have to make our judgements without a good deal of the qualities we might often weigh-up when selecting the best of our images – such as color.

    I found this quite irritating. JPEGs are rendered just as they should be though, but when fed simultaneously-shot Raw and JPEG files we only get to see the Raws but with the rendering of the JPEGs – though I found the relationship between what I’d shot and what was displayed a little inconsistent across camera brands. If you only shoot Raw though you get to look at slightly flat and unsharp images.

    Judgement Day

    It seems that the ‘best’ picture (top left) has been selected by the software because the subject is quite central in the frame. The picture the software likes least (bottom right) shows the subject against a distracting background.

    It’s not completely clear on what grounds the program and analysis makes its judgements as it orders our images from good to bad. I spotted that it seems to like a subject that’s in the middle of the frame, and one that sits on a third, and that it can tell within a group when the subject is against a clear background and when it’s against one that’s a bit distracting.

    Magnifying the eye of the sitter from this group of portraits makes it clear that the software doesn’t place sharpness at the top of its list of priorities. The sharpest picture (bottom center) is actually marked with a waste paper basket, indicating that that Picturesqe thinks I should dump it. The shot it recommends is a good deal less focused.

    Things that humans might take into account though the software doesn’t seem to – such as exposure, over-powering flash, focus and whether there are distractions in the background. It can work out when eyes are closed though, and it knows the difference between someone facing the camera straight on and someone at an angle – but it always prefers face-on as though that’s the safe formula for a ‘good’ picture. It can’t tell an unflattering angle from a complementary one though, and doesn’t distinguish between a well-lit subject and lighting that makes a nose look big.

    The zoom feature identifies the part of the subject that you want to inspect and just magnifies that bit. The clever bit is that it doesn’t just magnify the same area of the frame of each picture, but actually identifies the part of the subject – wherever that subject is in the frame.

    A very nice feature is the synchronized zoom that is great for checking focus across a number of images at the same time. With a collection of portraits, for example, you can zoom into the eye of one frame and the eye area of all the other frames get magnified along with it. What is particularly clever is that the eyes don’t have to be at the same coordinates in the picture for the program to magnify them, as it can identify similar objects in the images across the series.

    This feature worked brilliantly for portraits, allowing closed eyes and missed focus to be spotted easily and early on, and on occasions it worked even when the images weren’t all in the same orientation. The success rate is reduced for landscapes and more general scenes where the element you want to inspect isn’t as distinct or as obvious as an eye, but the application can concentrate on rocks, trees and buildings so long as the object you are checking has reasonably powerful edges and outlines.

    When the images are less regular, such as in these street pictures, the software finds it more difficult to identify the subject. Using the zoom feature didn’t allow me to compare the sharpness of the subjects in the street pictures, even though they have similar outlines and contrast in each frame.

    Export

    Once you have whittled your selections down to the best images in each group, or to the better groups in the collection, you are ready to export the pictures to their final destination. The editing process involves deleting the images that aren’t wanted – not from the card or from their home on your hard drive, but from the imported collection. No files are actually deleted.

    To do this you’ll have to click on the trash can icon on the image. And that will have to be done on every unwanted image. So, if you have shot a sequence of 100 images and only need one you have to delete 99. You can attend to each frame individually and delete each one in turn, or you can select them all and use the delete key on the keyboard.

    I’m a little unhappy about the solely subtractive nature of the process and feel I’d rather just select the best frame and export that without having to deal with the wreckage of the frames that didn’t work, but the truth is it’s more of a big deal in my head that in reality. Clearing away the duds doesn’t take that long.

    Images can then be exported to a chosen folder on your hard drive or directly into Adobe Lightroom for editing. 

    The research program

    At the export stage the first option in the navigation is to send the images to Picturesqe’s research program. The company wants users to send images that have been rated by humans so that it can compare the ‘right’ ranking and grouping with the way the software performed. The idea is that the company can study the differences and similarities and develop algorithms that select and grade more like the user does.

    The uploaded images, we are assured, will never be used for anything other than research, and the company takes only a 1500×1500 pixel thumbnail that is studied by a computer and not by a human. The idea is to build a database of how images are selected to allow the software to learn and get better at its job.

    In this example the software has compared a Raw file with a JPEG I processed from that same file, and has concluded that the Raw file isn’t just weaker than the JPEG, but that it should be deleted. Had I used the software to select which Raw files I would save for processing in the first place, this one wouldn’t have ever got to the black and white JPEG stage.

    Conclusion

    So, Picturesqe isn’t perfect. Not by a long shot, in fact, but it gets enough right that shooters who work with long sequences of the same subject will find it useful. In more general work it is not much of a chore to compare three or four frames to select which is the better, but if your photography involves long bursts of action or multiple frames of the same thing, then what it does will be enough to make it useful and to save you time. As the engineers seem to have prioritized portraiture it makes sense that this is the area in which it works best, and for which I’d most recommend it, but motor sports and general action would suit it too.

    The company doesn’t claim that Picturesqe can make judgements of taste

    As the selection process tends to judge on more formulaic principals you’ll have more success with images that are more about content than artistic ideals. The company doesn’t claim that Picturesqe can make judgements of taste, but that it uses math and obvious standards to rank images, and that is exactly what you get.

    There is a good deal of potential to make this a very powerful program for factual photographic subjects and the learning element of Picturesqe promises great things. Right now it needs a bit more work to make it of a standard that the majority of photographers will find it useful, but it is an interesting idea and even as it is will be useful for a good many.

    Fortunately we don’t have to take its advice completely, and when you work together with the program, combining its literal mind with your own taste and style, it can work very well if you shoot the right sort of subject.

    What we like:

    • A great concept
    • Easy to use
    • Very good for comparing magnified views 
    • It does have some success
    • Good for factual images

    What we don’t like:

    • Success rate just isn’t high enough yet
    • It doesn’t sort by orientation
    • Doesn’t seem to take focus into consideration


    Interview Q&A with Picturesqe CEO Daniel Szollosi

    We got some time with Daniel Szollosi, the CEO and founder of Picturesqe, and questioned him about the way the application works and some of the issues we picked up during the review process.

    DPReview: The process to get to the pictures photographers want to save and work on means they have to delete those that they don’t. So, in a collection of 100 pictures to get to one you want to keep you need to delete 99. Is there a way to just export the one you want instead of going through the process of deleting the 99?

    Daniel Szollosi: We realize it is not a perfect workflow concept, you are totally right. Originally our primary goal was to get rid of the digital trash – so image deletion was the main focus. Since then we have come up with a new workflow, which directly helps the selection of top quality photos. The new workflow is going to be released in the next version of Picturesqe.


    DPR: Raw files don’t always look the way they were shot – when they are supposed to be black and white, for example. Does that only happen when there is no simultaneously shot JPEG imported with it?

    DS: We know about this bug. It’s related to white balance, we have problems with setting the right value yet. 


    DPR: Your program displays Raw files from cameras that are new and which Adobe can’t display yet. How does that happen?

    DS: We’re using a 3rd party library for decoding Raw camera images. Another advantage of this library is that the user doesn’t need to install any kind of camera drivers. 


    DPR: The program doesn’t always detect when images are out of focus – is it supposed to or is that something you are working on?

    DS: We are definitely working on it! In the next version we expect a really big improvement regarding focus detection. We have developed a technology which seems to be better than the current state-of-the-art regarding local sharpness/blur evaluation. 


    DPR: Picturesqe seems to prioritize images where the subject is in the centre of the frame, or on a third. Is that part of the analysis? What other factors are taken into account?

    DS: We do take into account composition factors, like rule of thirds. The quality factors we take into account are:

    • Location of visually attended area
    • Exposure (globally and in the visually attended area)
    • Focus (global blur, wrong focus localization)
    • Lighting distribution
    • Color harmony
    • Composition
    • Optical distortions
    • Visual noise

    DPR: In some cases the program recommends deleting a Raw file but promotes a JPEG processed from that file to the top of the stack. Why does that happen?

    DS: Thanks for mentioning this, we have not thought about it. We resize the images to a smaller size when evaluating the quality and aesthetics and on this scale Raw information does not exist anymore. We can easily implement a filter which prioritize Raw images when compared to its JPEG descendant.


    DPR: Does Picturesqe assess exposure and the content of the background?

    DS: Semantically we do not interpret the content of the background, but visually we assess the background and the foreground separately. The quality attributes calculated from the foreground have more weights.

    Articles: Digital Photography Review (dpreview.com)

  • Dual-cam Duel: Sony and LG Innotek rumored to be competing for Apple orders

    Judging by the number of rumors buzzing around the internet it seems almost certain that at least one version of the upcoming iPhone 7 will come with a dual-camera setup. Word on the street until now has been that in addition to the standard model Apple is planning to offer a premium version of the larger ‘Plus’ model with a dual-cam designed by Linx, an optical engineering company that was bought by Apple last year.

    According to new information this might be, at least partly, incorrect. Financial publication Barron’s quotes Citi Research in saying that all 5.5-inch iPhone models coming this fall will have dual-cameras on the back, not just a premium model.

    Barron’s also quotes Chris Chan of Asia-based financial services group Nomura:

    ‘We think Sony may not be able to deliver its full share of dual cameras to Apple due to: (1) lower-than-expected yield, and (2) damage to its production facility from the April earthquake in Kumamoto. As a result, we think LGI will gain majority share of the initial dual camera orders from Apple…’

    Looking at the amount of information out in the open there is almost no doubt that whatever the manufacturer of the modules, we’ll have a dual-cam iPhone to play with in September. The question is if the dual-lenses will be used for optical zooming, as on the LG G5, or to enhance image quality and simulate a shallow depth-of-field, as we’ve seen in the Huawei P9.  We’ll only have to wait a few more months to find out…

    Articles: Digital Photography Review (dpreview.com)