Most people reading this will be familiar with the specs for the iPhone X tele lens, but just in case you missed them, here they are.
- 6 Elements
- 56mm equivalent
- 12 megapixels
- Fixed Aperture f2.4
- Image stabilised
- Autofocus via focus pixels
- Body and face detection
Note 1: It has not been possible to find a final spec on the sensor dimensions or pixel size. However, all information from Apple and my own tests indicate it is superior in all ways to the sensor in the iPhone 7 models.
Note 2: The sensor in the 8 and X models may be the same with only the lens and image stabilisation being different, though Apple has indeed hinted that the X version is superior. Hopefully, I can provide a definitive answer to this in the third part of the series when I compare the two with real-world images.
Note 3: No portraits in this article, yes I know lots of people will use this lens for portraits, but they tell you next to nothing about chromatic aberration, edge performance, cross-field resolution etc. Check out the previous article if portraits are your game. Anyhow lots of people will use the tele lens for all sorts of regular images, so this is for you folk.
A Request: If anyone really does know the exact sensor size I’d be keen to know also as it will help me to calculate exactly how the depth of field pans out, please drop me a line. I can make some estimations based on the focal length, but I don’t know for sure that the reported 6mm of the X tele lens is precise either.
This review of the iPhone X tele lens only uses RAW/DNG files, no jpegs, basically I want you to see what the lens/sensor module is really capable of. JPEG shooters will not obtain the same results, but I assume you already knew that.
The files were captured via ProCamera, but the results should be pretty typical of what’s possible via any decent app capable of DNG capture. Test files were processed in Iridient Developer on my Mac or Lightroom CC on my iPhone X
Remember that RAW/DNG files can be adjusted to provide pretty much any colour rendering or white balance setting you desire, so stating that the RAW files produce a more mellow colour is meaningless, it all depends on how you choose to cook the files’ ingredients.
Ok, let’s dive into the good stuff.
Many photographers get hung up on resolution and judge a lens purely by how much detail it can record, I find that approach a little simplistic but for all you pixel peepers, the answer is yes.
Yes, this iPhone telephoto lens resolves very nicely indeed.
To judge resolution, I take several photos of real-world 3D subjects; I certainly don’t go shooting flat charts, brick walls and printed artwork. Testing lenses in this way doesn’t tell you much at all unless you’re working with macro lenses. It quite surprises me that people would think that close up flat field subject rendering is a credible way to test a lens at all, anyhow, we have lots of photos at varying distances of real-world stuff here.
Next, all the tests have to be done in DNG (RAW); otherwise, I’d be just testing the JPEG processing of the device rather than the optics and sensor.
I shoot at the lowest ISO to ensure that if the device is adding noise reduction to the file, it’s minimal and also so I can get the best result in the RAW converter.
I check the files out using Iridient Developer with all noise reduction turned off, I also post-edit the extracted files in Photoshop to determine their malleability.
I try to find the interpolation algorithm that works best for the file at hand, Iridient Developer offers several options, they give slightly different results and looks. In the case of the iPhone X files, V3 Ultimate and V3 Detail+ seem the give the best results.
Finally, I view the image at different percentages, 200, 100 and 50%
So moving on, what have I found?
The central resolution is about as high as you could reasonably expect for a small 12-megapixel sensor, it has no problem resolving fine textural information or low contrast detail.
I reckon if you were to look at the resolved detail and compared it with any decent 12 or 16-megapixel camera/lens combo offering the same angle of view you’d be unlikely to find anything to mark the iPhone X down on. (more on that later)
Furthermore, there is sufficient resolution to uprez to larger than native print sizes and files offer a reasonable degree of crop-ability for most purposes.
The results here are pleasantly surprising; there’s little drop off in detail as you move towards the edges and corners, I’m confident that no one is going to have an issue with soft corners for any application.
In particular, the very even cross frame resolution would make for high-quality image stitching results.
I’ve examined a wide array of images, even landscapes with very fine grass out in the corners show high resolution, the only trade-off I can see is a slight loss in corner contrast, but most of that would be due to vignetting and the resulting correction.
In the dim past, many smartphone lenses showed rubbish native contrast, especially if you pointed them towards bright light sources. This behaviour was due to a combination of poor lens coatings and the effects of the front cover glass.
Some early iPhone lenses scratched up badly after mild use. These days the front cover lens is very tough and highly scratch resistant. Additionally, the coatings seem to be vastly better. Even when the iPhone is pointed towards light sources contrast remains commendably high, this is especially the case with the telephoto lenses on both of the new iPhone X and 8 models.
The advantage of the higher contrast is seen in the shadows, which respond nicely to editing, provided you have good exposure files can be coaxed into displaying excellent tonal gradation in the shadows.
Note, as always the lens needs to be scrupulously clean for best results, something often overlooked. I’ve had many students who ended up with hand cream and fingerprints all over their smartphone, obliterating any semblance of resolution. That said, the telephoto lens is a little more forgiving of dust and small levels of lens born pollution.
One nice side benefit of the newer iPhones is the better sealing of the phone and its internal workings. Older models sometimes built up dust under the front lens cover requiring a pull down to fix; I’ve not seen this on an iPhone 6 or later model, still, it’s something to look out for if you are buying a second-hand phone as it plays havoc with contrast rendering.
Flare is closely related to contrast; any lens can be made to flare if you try hard enough, usually when shooting into the sun or towards some specular light source. Most older smartphones show horrible flare in particular when shooting against bright white cloudy skies.
It is easy to confuse flare with localised over-exposure, this often occurs with light sources. Here the immediate area around the light source loses gradation and bleaches out, this is usually a result of the sensor design and technical limits, but it may be made worse by poor lens coatings and designs which cause internal reflections.
I really pushed the tele lens to see if I could induce flare, and frankly unless I did something idiotic the flare was a non-issue, basically specular highlights remained specular, no flare-fests were evident.
The good news is that you will get excellent flare-free results without having to shield the lens, which has not always been the case. Certainly, my previous iPhone 6S often needed a little hand shielding many situations.
It can be a little tricky to test the vignetting with iPhone Raw/DNG files, you see all of the converters will apply an auto-correction under the hood before you do any editing to the image.
Does this matter? Yep, it does, vignetting causes flow-on effects further down the editing chain, less real vignetting makes for a more pushable file.
To see what the native vignetting is you have to remove the effect of the base profile from the image, I can do this in Iridient Developer.
I can tell you that the native vignetting level is minimal in smartphone terms and the fall off is quite even showing no sudden banding. The extreme corners certainly show a bit extra darkening, but it’s easily corrected and unlikely to show up as an issue in real-world images. In short, the native vignetting performance is the best I have seen from any smartphone lens I’ve tested so far.
Most older Apple smartphones in RAW show a significant red colour shift as your move towards the outer edges of the frame and this is made worse with underexposure or raising the ISO setting.
It’s not so much a product of the lens but rather a sensor issue which is exacerbated by vignetting, which of course means the exposure in the corners is always less.
You will almost never see this effect in JPEGs or indeed most RAW conversions because it’s dialled out early in the processing chain as part of the profile, but it has a flow-on effect for colour reproduction, colour noise and to a smaller degree, sharpness of the edges and corners. It can also make high-quality panorama stitching very difficult.
Naturally then, less native red-shift would mean a better-edited result. Again the news is excellent, the Tele lens module shows minimal colour shift, you can just see it in cyan blue skies or if you have neutral greys out towards the corners. Finally, I can say the redshift is a non-issue and a big hooray for that.
Honestly, I couldn’t find any; I tried all sorts of photos and subjects, I looked at the image at magnifications up to 500% with the input profile disabled, nada, nothing. What can I tell you, well only that this is an impressive performance?
You may see a tiny bit of bluish fringing on white highlights if the exposure is set high enough, but this isn’t CA, instead, it’s the effect of very localised blooming, and even then I had to enlarge the image to 300% to notice anything.
This is a little tricky; the distortion level is minimal, low enough that I doubt anyone would ever notice it unless you were into the masonry arts and shooting brick walls to get your jollies.
On a couple of occasions I thought I could notice a tiny bit of pincushion distortion and then on another occasion I felt it was barrel distortion. It may change a little as the focus distance changes, but really there’s no distortion worth thinking about.
The great news is that if you wanted to create stitched panoramas, the telephoto lens should make the job very easy. Oh and just in case you were wondering, yes you can use the 2X lens to take panoramas using the standard Apple camera app.
To test the noise level, you need to be able to see how the RAW/DNG files look without any noise reduction applied. Different applications handle the noise reduction in various ways and some like Iridient Developer, provide several options to deal with noise.
It might surprise you to learn that some noise reduction methods can actually make the iPhone images look noisier because they alias the noise, it also depends on whether you’re after a “clean down-scaled web image”, a screen image for the iPhone itself or something optimal for large or small prints.
Ultimately the only way to get a handle on things is to judge the inherent noise in the unprocessed raw file and then extrapolate how that might translate for different output needs.
In the case of the dual lens iPhones we need to do this for each lens because obviously, the sensors are an integral part and different in design (I suspect anyway).
Ideally, you’d want to use the iPhone at the lowest ISO setting to obtain the lowest noise levels, which is what I have done for this section, the next part deals with what happens when you ramp the ISO to higher levels.
For this test I use ProCamera, I also took shots using both UniWB and TLC methods for comparison and examined the images initially with all profile settings off, in other words with no sharpening or noise reduction applied, either early or later stage.
The lowest ISO option using ProCamera is just 16, though the app does not actually list this. It’s not uncommon to be able to shoot within the range of ISO 16 through to 32 under bright sunlight.
TLC shots in Bright sunlight at ISO 16 will give a shutter speed in the range of 1/100 to 1/250 sec, which is hardly a challenge on the stabilised iPhone X. To see any significant noise you’ll need to zoom into a 300% view or greater, most of the noise will be fine luminance noise, but the deepest shadows will show some chrominance noise that is slightly blue shifted.
Essentially the TLC files are noise free, and all noise reduction could be disabled, meaning you can then apply whatever sharpening you need to extract the detail level you want.
UniWb files typically end up with a better level of RAW exposure because the histogram is no longer acting like a filthy lying mongrel, you may get as much as two EV extra exposure this way. The idea is to set the exposure so the brightest required details just about to clip. Unlike the TLC versions, you’ll get some minor chrominance noise which will tend to cut in from the middle grey tones and increase as you go darker on the tonal scale.
My take is the optimally exposed UniWb files show very little noise to worry about at the lowest or moderate ISO settings; you could just dial in a minimal amount of noise reduction and then sharpen as needed, provided you don’t do something completely stupid the image will look great.
Just so you know, “completely stupid” might mean turning the saturation up to eleven, applying crazy amounts of medium radius sharpening, trying to recover deep shadows that we expect to be left well and truly in the dark.
Now, most people won’t shoot neither TLC or UniWB, they’ll just let the app do whatever it likes, so what exactly might they or you expect.
For regular exposure they will find 100 ISO is the point where noise becomes a bit annoying and needs cleaning up, 50/64 ISO will show fine filmic noise that for prints and downsized images could be left alone and might actually be helpful, anything over 400 ISO will undoubtedly be a bridge too far.
Underexposure vrs ISO Gain
With many regular DSLR and Mirrorless cameras shooting in RAW, it’s possible to shoot using the lowest ISO and merely brighten the image in the RAW converter. You will often get a result little different than what you would have achieved by ramping the ISO in the camera. The main benefit is that such images often retain a better level of highlight detail.
I thought it would be a neat experiment to try this with the Tele lens on the iPhone X and then see what could be drawn out using a desktop RAW file converter.
The fundamental noise levels were pretty much the same either way, but I did notice a couple of anomalies, in bright conditions the underexposed images seem to be sharper and more detailed than the higher ISO versions for the same shutter speeds.
I suspect the iPhone may be applying noise reduction to the DNG files as the ISO is ramped up, but I cannot prove this an Apple don’t of course document this anywhere.
The loss of clarity really turns nasty at around 320 ISO and beyond 400 looks horrible in almost every respect to my easily offended eyes.
On the other hand under low light conditions where slow shutter speeds are needed to grab those fleeting photons, the higher ISO DNG files seem to render better results than severely underexposed low ISO frames. I took many images in the 1/30 sec and lower range and in most cases with custom processing, I could get slightly lower noise, more shadow detail and better clarity from the higher ISO captures when compared to the underexposed low ISO versions. Just note, the shutter speeds for each of the test pairs were identical meaning the real exposure at the sensor level was the same in each case.
The extreme highlight detail was usually better with the low ISO frames once you applied the optimal processing settings, but not to the same degree it is in when taking shots in bright daylight conditions.
It became apparent in this process was that the optimal sharpening and noise control settings for each of the changes in ISO are considerable. It takes quite some effort to find the best settings….but in all cases, the final result was wildly better than equivalent JPEGs from the camera, which were all pretty mushy and “watercolour like”.
Under regular daylight (at higher shutter speeds) it seems the higher ISO ramps up the noise reduction enough to require considerably more sharpening in comparison to the lower ISO frames. While it’s entirely possible to obtain reasonable levels of resolution from the images shot at 200 ISO plus, the final results tend to look more forced and digital than the optimally processed but under-exposed low ISO frames.
For lower light levels I found I could get an excellent compromise by choosing an ISO of 160, in this case, highlight detail was still very recoverable, the shadows were fine and the noise looks comfortably filmic.
One final note on this, many folks would think the native colour noise (chrominance noise) from the higher ISO frames would be horrible. I found that not to be the case; sure there’s colour noise but its nowhere near as objectionable as it often looks in many of the processed DNGs and JPEGs I’ve seen on the web. The real issue is the processing methods, many of which exaggerate the noise, turning the fine colour noise into ugly clumpy blobs of colour with underlying rainbow clouds.
ISO Range performance
I find it hard to draw conclusions here as there are so many variables at play, the shutter speed used, the amount of light, whether you are targetting highlights or shadows and ultimately how you process the files. In other words, it’s nowhere near as simple as with JPEGs where you can do a straight ISO ramp up and then point your finger at the results and give the definitive nod of approval.
But you want answers so I will make some generalisations that hopefully help you get optimal results.
- Perfect focus matters more than with the standard lens, and under low light, the iPhone sometimes hunts a bit and settles on the wrong focus distance, if the focus is out a little you’ll probably try to ramp up the sharpening and that makes the noise look way worse.
- The tele lens needs to be much steadier than the wide angle one; it’s far more likely you’ll get motion blur unless you pay a bit more attention. I found that 1/30 sec was the point where problems start to crop up occasionally but everyone is different, your experience may be better or worse.
- Anything over 400 ISO is going to look rubbish regardless of what you do in the processing.
- 100 to 160 ISO is good for low light, even if you underexposure to keep the shutter speed practical.
- Under bright light the lowest ISOs (16-32) can be radically under-exposed (like three stops or more) and still give terrific results, the main advantage is this will allow for extended highlight rendering without the hassles or deficits of HDR. A “2 stops underexposed” image shot at 16-32 ISO can do a very nice impression of colour neg film with sympathetic editing.
- Under bright/moderate light there is a general drop in potential sharpness at ISO 125 and above, but it will be fine for many needs.
Finally, I was fascinated by the excellent results delivered by the telephoto lens on the iPhone X and decided to do a little further exploration and see how it compared to a regular high-quality camera.
I chose to compare it to my Olympus EM5Mk 2 paired with the Olympus Pro Grade 12-40mm f2.8 lens.
In the end, the comparison grew more significant than I planned and I made test shots with the standard iPhone X lens and also tried cropped and panorama shots for the iPhone to compare with using a regular zoom on your Mirrorless camera. All images on both devices were shot in RAW.
I guess, at this point, you are assuming the M4/3 reigned supreme, you might want to just hold off on that assumption, anyhow the article will be posted soon, then all will be revealed.
- TLC files are captured using pre light filtering, this is entirely explained in my book “Ultimate iPhone DNG” available on the iBooks store.
- UniWb capture increases the overall level of exposure to the maximum, again this is covered in “Ultimate iPhone DNG”