Infinite frames in a crop: compositional building blocks (or, the future of photography?)

_S21_L1008191 copy
Ion Orchard sunset. Leica S2, 70/2.5 Summarit-S

Here’s an interesting thought: sensor resolution is way past the point of far more than most people need – look at that recent Nokia 808 PureView and it’s 41MP sensor as an example. For most compact camera users, that’s just asking for trouble – file sizes, viewing, shooting technique…the list goes on.

What if we took a say 50MP compact 2/3″ sensor (not unfeasible) and just output no more than 5 MP from the sensor? You can zoom to your heart’s content before capture – either by taking a 1:1 crop or downsizing various portions of the frame; have a low light mode which uses pixel binning to reduce noise and/or improve dynamic range; and finally, make the whole thing speedy and lag-free by using a bright, fixed-focus wide angle lens. Or I suppose if you must have macro modes, make it AF. Finally, add some form of image stabilization to the mix, because that kind of pixel density is going to make most tripods look as though they have palsy. We’ve dropped the size, weight and cost of the zoom assembly; this makes for more compact, robust cameras. It also means that more engineering effort and perhaps even more money (though I wouldn’t put it past consumer electronics companies to want to hang on to every single extra penny and avoid R&D spending where possible) can be put into the lens, which should in theory improve optics.

The results probably wouldn’t be pretty at the pixel level, but the target audience is unlikely to care. And 5 good MP is more than enough for a clean 6×4″, or social media post, or email. Hell, it’d probably be enough for a decent A3 print if you use one of the pixel binning modes in good light. Think of it as digital zoom, but instead of getting degradation off the base pixel-level image quality, you get an improvement as you use more and more pixels.

However, I digress. As interesting as that idea is for the future of compacts – I wasn’t initially intending to suggest a new type of camera. What I wanted to explore are the building blocks of composition. Assuming everything is in focus, a scene – especially one shot from a wide angle perspective – can easily be decomposed into multiple elements, some of which contain the subject, some of which contain context, and some of which may turn context into a subject in its own right. All images are made up of these building blocks, and each can be treated as an image in itself. (Don’t try to compose this way in the real world, or you’re either going to go crazy or turn into a product photographer.) Examples may be – a face, a texture, a landmark.

Sometimes you don’t need more than a couple to create an arresting image – a portrait, for instance – sometimes you need as many as you can get to convey the chaos of the scene (crisis photojournalism comes to mind). What I notice in good photographers’ images is that all critical frames contain the elements required to tell the story, and usually in aesthetically pleasing relative placement, too. This is a learned skill; one can identify the critical elements required to tell a story, and then go out and look for a frame that contains them. I suppose the more control you have over the placement of the individual elements, and the more individual elements you can even identify in the first place, the stronger your image will be. Elements may break down into sub-elements – a wide shot may contain a portrait which can contain a face closeup; all of these things are building blocks.

This brings us full circle to the original starting point – how many images are there in an image? Let’s take the first image as an example:

_S21_L1008191 copy

This decomposes into what I think are five clear elements; the building and its relationship with the organic are the subjects.






What I find interesting is that each of these elements – or crops – could be an image on its own. And you can of course further subdivide these images down into their own building blocks. How many does one need to make up a strong overall image? One? Two? More? It’s hard to say; I think the answer is actually infinite. If you had infinite resolution, you could keep subdividing and cropping infinitely.

I know this is a fairly nebulous post, but stay with me here and you’ll start to see the genesis of an idea.

Let’s do the same, but in reverse. The frame below arguably works on its own as an image:


But then, if we include more context, so does this:


Or this:


And this.


We can keep pulling out…


…until we reach the initial (final?) frame.

_8001447 copy Welding. Nikon D800, AFS 28-300 VR

Each of these compositions on its own would have worked as a frame. Similarly, continuously pulling out adds more and more context, until you pass a threshold beyond which the context dominates the subject. I framed the final shot that way because that was the farthest I thought I could pull out to maximize context, but at the same time isolate the subjects in the frame. Anything more and the subjects would have been too small to identify cleanly. You’ve also got to consider the final viewing size: smaller would mean physically smaller subjects, which would encourage tighter framing. Similarly, images that appear flat and cluttered at smaller sizes frequently work well when enlarged due to the greater amount of detail we can physically appreciate.

I started this article postulating on whether increases in resolution would lead to a new generation of pixel-binning super-sensored compacts; the more I think about it, the more I think it might even apply to photography as a whole. We have long past the point of resolution sufficiency for most purposes; shrinking pixel pitches mean that lens maximum apertures and diffraction-impacted apertures are converging. Eventually you’re going to have to buy excellently designed and well-corrected fast lenses and shoot them wide open to absolutely maximize resolution from the sensor.

Perhaps this will give birth to a new style or form of photography – we capture everything first, and then decide what’s important to the frame afterwards. A little bit like Lytro, but for composition rather than focus. Of course, we can do the same today with the super high resolution DSLRs, but there are limits to how much we can crop, and the lack of suitable lenses to serve as a reproduction optic. Such a camera – let’s call it post-capture framing – would be responsive due to lack of focusing needs; hyper focal would probably suffice; offer decent low light performance due to the large maximum apertures required; and be reasonably compact – rangefinder sized, perhaps – due to an optical system designed and matched to the sensor. It’d probably be closer in proportions to a cube, actually – once the lens is taken into account.

The skill of the photographer at the moment of capture would then be reduced to timing and exposure; the latter would probably be mostly automated, leaving only timing; however, this doesn’t make things any easier. The bulk of the work would now be done in post-processing – not just the digital ‘development’ portion, but also the editing, framing and cropping. I’ve always thought of editing as one of the critical skills for a pro – one’s reputation is based as much on the photos you show against the ones you don’t; being able to see the frame ‘in the wild’ and capture the moment in time is not an easy thing to do. Post-capture framing would give you all the time in the world to do that, which would make things easier somewhat – but also offer more opportunities and flexibility for other compositions.

This would of course raise some interesting IP issues: if another ‘editor’ cropped the original photo to a much smaller, unrecognizable frame, whose photo is it? The person who pressed the button, and just aimed the camera in the right general direction, or the one who actually saw the composition in the master frame?

It’s a brave new world, and an exciting time to be a photographer. Personally, though, I’m one of those very anal people who seeks to maximize all of the image quality they can get out of the camera; which means I don’t crop unless it’s to a non-native aspect ratio. Understanding the use of ‘building blocks’ to create a final composition, plus knowing when you’ve got enough context (or too much context, overpowering the subject) is a very important skill to refine, regardless of the technology involved. MT


Visit our Teaching Store to up your photographic game – including Photoshop Workflow DVDs and customized Email School of Photography; or go mobile with the Photography Compendium for iPad. You can also get your gear from B&H and Amazon. Prices are the same as normal, however a small portion of your purchase value is referred back to me. Thanks!

Don’t forget to like us on Facebook and join the reader Flickr group!


Images and content copyright Ming Thein | 2012 onwards. All rights reserved


  1. One of the reasons I swapped the Nikon Z6 for Z7, the latter includes a 5:4 crop; I find 35mm native ration too long, personally. While some will say “what a waste of megapixels” or “crop in post”, there’s something instinctive about framing within those constraints to begin with, cropping in post becomes too calculated at times.

    • Strange there’s no consistency. You can have 16:9 on the Zs for example, but not on the D850 – which has 1.2x that the Z7 does not.

  2. Otus? 🙂 “shrinking pixel pitches mean that lens maximum apertures and diffraction-impacted apertures are converging. Eventually you’re going to have to buy excellently designed and well-corrected fast lenses and shoot them wide open to absolutely maximize resolution from the sensor.”

  3. parameteres says:

    interesting article… i guess it’s difficult enough for me to take one frame, let alone composing for multiple frames in one shot 🙂

  4. Michael says:

    This is another interesting article. From my view the premise holds especially if you see a scene, have to capture it very quickly, potentially too quick to get composition right the way you envision it in the shot itself. Using wide angle and cropping the picture to your vision of the scene will probably help you to get the shot.

    But generally speaking I feel I would probably try to avoid that. To me it would be similar to me to spending much time on the computer to fine tune pix in terms of exposure etc, but now also adding the dimension of composition. Plus I am not sure if it would lead to a certain sloppiness when composing images.

    • I’d avoid it too. I think it’s sloppy work, and up til this point, most of us have been able to do just fine. I’d rather have a faster, more responsive camera rather than one that’s higher resolution and wider. But I can see popular trends going this way, with their consequent influence on how the public in general perceives photographers – now relegated to button pushers – I can’t say that it’s at all a good thing.

      • yoshi360 says:

        I don’t think composition and framing are the same. This technique would only allow you to frame in PP, but if you didn’t shoot at the right angle then there is nothing you can do afterwards. I actually crop most of my images at least slightly (there is often something to improve in the corners), but I’m trying to get the perspective right on scene.

        • I suppose it depends on how exacting you want to be in your definitions:

          Composition = relative arrangement of objects in a scene
          Framing = where you draw the borders of the scene

          In that framework, we can crop to adjust framing, as you say. Composition and relative positioning, no. But by dramatically changing framing (to the point where you change most of, or all, the contents in the scene) would mean that the composition would also effectively change…

          As for cropping, I try not to and usually don’t. 100% finders help 🙂

  5. Great post. Good food for thought.

    I agree with the principle of what you are saying, I have actually done the same thing with a few images myself. Some images/scenes can be cropped to get multiple images, but it does add an element of randomness to what you get.

    Taking one shot instead of several and relying on cropping later can result in a few of the images not being as good as they would have been if they had been considered as separate images at the time. When considered as separate images at the time you took the photos you could position yourself to optimise the composition of each shot.

    A quick made up example… You take a photo outside a building and there are some people off to the side that you later decide to take as the main subject of a crop of your image. Once you have cropped you realize that if you had just stepped a few steps to the right when taking the image that a more pleasing combination of how the people heads lined up could have been acheived, i.e. a few steps away the people facing you would not have their heads obscured by the people with their back to you. But if you had taken those few steps to the left to take your one and only image then an interesting curve or edge on the building that makes a great crop combined with a nearby tree would have been onstructed by a tree in the other position.

    With the sensor and lense resolution it will be interesting to see where the limit will come. If there was no limit then science fiction like imagery would be possible. An scene that comes to mind is New York city from Brooklyn. I like taking images of the Manhattan skyline from across the river in Brooklyn, imagine being able to take a skyline sunset shot and crop it so that the subject of the shot was a man in a suit smoking a cigarette backlit as the leaned against a post at the base of one of the towers. I am pretty sure things will never go that far but I wonder where the line will be? Will be reach the limits in our lifetime? Will there come a time when no new lenses or cameras are released because the limits have already been released.

    I dont comment much but I’m a frequent visitor here… Keep up the good work 🙂

    • Thanks. I think we’ll hit diffraction and DR limits before resolution ones – unless we start having enormous sensors.

      That said, even though things are going in that direction, it doesn’t mean that I like it at all…

  6. You raise some interesting questions, not least, by deciding on the composition in post, are you de-skilling the place of the photographer, and relegating them to button-pusher? To me, most of the excitement of making a photo is in seeing the finished article in my head before I make the shot.

    • Oh, I completely agree – I was on a job recently where the ‘creative director’ told me just to shoot everything wide so she could just crop it later – said creative director could neither create nor direct (nor communicate); the client wasn’t happy at all, because they expected me to produce images like my portfolio. When I did, the CD complained. Can’t please everybody 🙂

  7. yoshi360 says:

    The first part of your article exactly describes what Nokia’s PureView really does (per default it downsizes the 38MP images to 5MP using pixel binning), while the second part is something you can do when you use the full 38MP output and edit it afterwards. I believe we will see many more cameras that operate like this and sensor resolutions will skyrocket in the next couple of years.

    The only thing I am skeptical about is dynamic range. PureView has an unsually large sensor (1/1.2″), but its dynamic range appears to be on par with other camera phones, i.e. rather low. It seems to me there might be a threshold of pixel density after which dynamic range declines.

    • Yes, that’s where I got the idea from. I’m half curious to pick up one of these phones just to see how good the technology in it really is, though I suspect I’d quickly go back to the iPhone for UI and a proper camera for imaging.

      There’s definitely a relationship between pixel density and dynamic range – it’s limited physically by the size of the pixel well. We’ve been seeing improvement on improvement in recent years because of increases in quantum efficiency and design, but given all other things equal, a sensor with larger pixels will always outdo a smaller one for dynamic range.

      The other elephant in the room is diffraction – I wouldn’t be surprised if the PureView is already diffraction limited at f2.8 and full resolution…which means you scan only ever use a wide lens, otherwise you wouldn’t be able to stop down to increase DOF – the image would just get blurrier. Of course downsizing to 5MP might negate this completely.

      • yoshi360 says:

        On DXO Mark the NEX-7 (24MP APS-C, pixel pitch 3.9) outperforms the full-frame Alpha 900 (pixel pitch 5.9) in dynamic range. Sure, the A900 is a bit older. But similarly, the D800 (pixel pitch 4.7) outperforms the 5DMK3 (pixel pitch 6.08) in dynamic range by almost 3 stops despite the higher pixel density!

        I understand that DXO Mark and especially their score of the D800 aren’t without criticism, but nevertheless it seems that cameras with higher pixel density are still doing pretty okay in dynamic range and there doesn’t seem to be a significant gap. On the other hand, PureView and many other camera phones have a pixel pitch of miniature 1.4 micrometers. That’s way smaller, but still I wonder whether the correlation between pixel pitch and dynamic range is linear or if there is simply a certain pixel pitch that shouldn’t be undercut. Also that makes me wonder if the oversampling idea is really such a good one; after all, dynamic range and high-ISO performance are rather weak on the PureView (on par with other camera phones). At 1/1.2″ and native 5MP it should have been able to perform much better, if the Fuji X10, Canon S95 and the new Sony RX100 are any indication. Of course then it wouldn’t have any digital zoom capability.

        • I almost feel that the sensor manufactures these days aren’t willing to put the best (costliest) technology into low margin products like compacts; it seems like we have to wait for a DSLR sensor generation before seeing what improvements have been made. The DXOmark rating for the D800 is rather impressive, but then again their assessments do not take pixel density or resolution into account. However, in practice, I can confirm that the D800/D800E cameras have the best dynamic range I’ve personally seen to date…

  8. Fishnose says:

    Ming, that first Ion Orchard image – absolutely wonderful.

  9. First of all, very nice images as always. 🙂 Of the construction workers, I prefer the 3rd and 5th images, for me there’s a sense of isolation and loneliness in them that isn’t there with the building to the left in the final image.
    Also, interesting article of course. I come here pretty much every day, hoping to learn and get some new ideas, and I do. 🙂
    I don’t know if I “should have been” further along (10 000 actuations, lifetime total) but I absolutely rely on cropping to get a usable image out of an capture. Guess I’m wasting pixels. :p Or rely on primes too much when I really need a zoom.

    • Thanks for your compliments. One tip: pick your perspective first, then frame. You might enjoy this early article on perspectives, too.

      • Oh yeah, that article basicaly nails it. Anyway your set images in the current post puts your point across very strongly, I love the first architectural series. Even apart from the high megapixel camera idea, the photos are a good example of working on a subject and finding images within images. This might as well been a true multi picture series documenting a search for “the” picture and a peephole to the authors way of thinking.
        I admire your photos all the time and I’ve been visiting the blog daily for the past few weeks (since discovering it)!

  10. Changing crop doesnt change perspective, thats where a multi mpx compact camera would be lacking. I mean you should take multiple shots anyway while changing your position. How about extreme hi-res video shot by keeping the motion camera pointed in the same direction but moving through different posiotions, then choosing single frames, than choosing crop? 🙂

    • No, changing field of view changes perspective. And by cropping, you’re changing your field of view.

      Video would be the next stage, I suppose. But oh boy, the file sizes…

      • Cropping changes field of view, that’s all it does. How does changing crop change subjective spatial relations between objects? Or how can it possibly affect the image to imitate it being shot from eye level vs. waist or floor level? Perspective changes while we move around the camera’s sensor, not so when zooming or cropping.

        • Ah, I see what you mean – I’m sorry, a bit of misinterpretation on my part. You can simulate the effect of a change in perspective (WA foreground emphasis vs telephoto layering) by cropping, but you’re right, it doesn’t change the relative spatial relationship of subjects. My bad.

  11. Brilliant thoughts! Thank you

  12. HI MING

    I am from India. I really like your b&W images. Can you do a writeup /tutorial on b&w processing? In case you do make a trip to India specifically Chennai / Madras I would be glad top take you around.



  13. The Lytro camera may be an early tool for infinite composition as you describe it.

    • I’ve got some pretty strong thoughts on that one, too – but more around the company’s control of your IP via them having you by the balls with their proprietary software for viewing the images – you can’t even output a finished JPEG, and even if you could, the poor resolution is of limited use. The products are shipping, but interestingly the buzz has completely died off.

      • yoshi360 says:

        That’s not true, you can absolutely export to standard JPEG (at 1080×1080). This is what makes the technology so powerful in theory because it gives you even more control in post production; the “let the viewer set the focus” aspect is little more than a gimmick IMO.

        • Hmm, that changes things a bit. Everything I’ve seen up til this point says that you have to use their software to view the images, and when you do, they retain rights or something similar because you’re using their code…

          1mp square is pretty useless for anything other than the web. And if you’ve got only 8 focus zones (another spec I see repeatedly), that’s pretty pointless too – especially at longer focal lengths where you need more fine control over the focal point.

      • yoshi360 says:

        You do have to use their proprietary software to open those lightfield RAW-kinda files, but then you can set the focus and export them as standard JPEG files. I don’t see any difference to something like Sigma’s Foveon technology where you are basically dependent upon Sigma’s software, too (there are some apps that can read Foveon files, but most can’t). I don’t know anything about licensing, but I imagine that’s only if you use their HTML service to let the users set focus afterwards. I believe that is running on their servers, I am not sure though.

        Also, as far as I know this is the real deal with “true” light field technology, theoretically giving you an unlimited number of focus spots. Maybe the software is limited for now? But the sensor is so small that usually everything is in focus anyway, which kind of defeats the purpose. I totally agree with you, the current performance (1MP, very weak low light performance) is not really acceptable. I like to think of it as a demonstration of what’s to come; just like the first digital cameras in the early nineties.

        • That would make sense, but then again it’s like Google saying they own all of your email.

          As for ‘true’ light field technology – it uses an interference pattern to back-calculate the origin point of the light rays (I’m a theoretical physicist by training). What it requires is one set of pixels to capture the light from each relative position – so I suspect the reason we only get 8 zones to choose from is because it’s a small 8MP sensor, trying to make the most of both small sensor DOF and light fields, but balancing that out against the inherent limitations of small sensors.

          I don’t think we need infinite adjustment though; I’d be happy with just being able to tweak front/back focus a little around the actual focal plane to ensure 100% critical sharpness every time. I know what I want the camera to focus on when I shoot, but sometimes either it or I have a minor lapse and we don’t quite get there 🙂


  1. […] lot about cropping in the past, when I think it’s justified, and even a little bit about the proto-wimmelbild interpretation of recursion in composition. Bottom line: good/acceptable cropping is when the composition and restriction of edges is done […]

  2. […] – no matter how small – is relevant to the overall scene and idea. It is the idea of infinite sub-compositions in a frame taken further – recursion in extremis Even though the idea may appear to be one of chaos, […]

  3. […] After that comes the ability to create a composition using available subjects, elements and light; I like to think of it as the ability to find a frame, or create some semblance of order out of chaos. The photographer isn’t so much creating something as isolating and presenting something that’s already there in a way that makes it stand out to the audience. This in itself is not an easy thing to accomplish, and most photographers never pass this stage; our world has so many potential images from so many perspectives which change as the light illuminating it changes, which means that it’s more than possible to never exhaust this source of composition. It’s related to but not the same as the idea of a strong image being constructed of a near infinite number of smaller sub-compositions – and you can read about that in my article on the subject here. […]

  4. […] After that comes the ability to create a composition using available subjects, elements and light; I like to think of it as the ability to find a frame, or create some semblance of order out of chaos. The photographer isn’t so much creating something as isolating and presenting something that’s already there in a way that makes it stand out to the audience. This in itself is not an easy thing to accomplish, and most photographers never pass this stage; our world has so many potential images from so many perspectives which change as the light illuminating it changes, which means that it’s more than possible to never exhaust this source of composition. It’s related to but not the same as the idea of a strong image being constructed of a near infinite number of smaller sub-compositions – and you can read about that in my article on the subject here. […]

  5. […] cropping is bad. I’ve touched on the reason in previous articles – notably these two on compositional building blocks, and proper perspective practice – But I don’t think I’ve really explained why. […]

  6. […] article and this article both enter into more detail on compositional […]

  7. […] Link: Infinite frames in a crop: compositional building blocks (or, the future of photography?) – … how many images are there in an image? Posted by Trent Nelson Posted in Photography […]

%d bloggers like this: