Creating a Fictional Concept Sneaker in Photoshop: Part II

By | Uncategorised | No Comments

Creating a Fictional Concept Sneaker in Photoshop

Part II: Constructing the Composite Foundation

This article is an adaptation of a workshop first presented by M. Seth Jones in 2015.

The Adidas sneaker you’re looking at here doesn’t actually exist. It’s a completely fictional digital composite; a photo-illustration created entirely in Photoshop, combining mostly pre-existing elements from six separate Adidas sneakers.

Combined, the illustration contains 112 layers; 67 of these are pixel layers, 45 of them are adjustment layers. Together these layers are relying on 90 discrete local masks, all arranged within 45 groups. It was completed in two distinct stages; first, the arrangement focused blocking-in process, followed by the refinement heavy finishing stage.

In ‣ the first part of this series, I walked through the approaches and considerations I used to design the initial sneaker concept sketch. In this instalment, I’ll explore the tools and techniques I’ve used to combine discrete photographed surfaces and materials as seamlessly as possible. Along the way, I’ll discuss the necessary requirements for preparing the file for exploring and developing realistic alternative colourways.

Outer silhouette: constructing the concept sneaker upper

My first objective is to lay down the foundations for the NMD City Sock upper, which along with the mid and outsole, form the majority of the overall silhouette of the fictional concept sneaker. The upper needs to be done in a few separate passes. I’m using the Primeknit upper from the mid top NMD City Sock 1, which is a casual sneaker and as such, doesn’t have a silhouette that reflects the overall aesthetic of the EQT range. If I align the toe and the heel to my rough sketch (and even add in a small amount of rotation to attempt to find a more fitting angle), we can see that both the inner and outer guidelines aren’t a great match.

Aside from the height of the sock cuff, the baseline outer edges aren’t too bad. The front silhouette edge aligns relatively well with my draft sketch, running from the front tip of the sock cuff, which is where I want my layered tongue to begin, down to where the toe meets the outsole. The back outer edge of the heel is also fairly kind to my rough plans, and could be corrected with a relatively simple rotation. The height of the sock cuff will need to be dropped dramatically, taking the original silhouette from a mid top to a more EQT-like low top. The midsole seam, where the sock upper meets the Boost midsole is also relatively decent, not taking into account the protruding midsole overlay details of the NMD.

What will really need some work here is successfully creating a cohesive inner pattern to the Primeknit sock texture. The relatively minimal styling on the NMD City Sock is quite front heavy, with its shift in pattern textures dividing the upper into three increasingly larger ‘panels’ from toe to heel. This works well for a mid top light on upper overlays or flourishes, as the slightly larger surface area of the heel works to establish the three striped rhythm in the upper, placing the emphasis on the sock textures shifting directions. This will need to be heavily altered to accommodate the more equal three panel division I’ve laid out in my rough sketch. Not only will I need a shorter heel and logo panel, along with a longer toe panel, but I’ll need to create a more aesthetically appealing three-dimensional shape for my sock upper, as well as work through seamlessly extending and modifying the Primeknit texture pattern in the most transparent manner possible.

Working with the draft sketch guidelines, I’ll begin with the ankle/heel section of the upper. I rotated the base NMD upper to get a better starting point for the direction of the Primeknit texture, focusing only on the rear panel section. A large amount of the early blocking in process involves making edits and laying down foundation surfaces that look terrible in large regions, which is why the early stages of photo illustration can be kind of awful if you have someone watching over your shoulder. It comes down to focusing purely on the region you’re editing, not worrying about how bad the rest of the image might look, and having confidence in your pre-visualisation sensibilities. It might look bad now, but you know precisely what you can do to get it to where you need it to be.

Working with the guidelines of the draft, I can see this is a decent position to begin to get a relatively clean base texture for the heel panel. I’ll likely need to work with the raised section of midsole overlay that’s encroaching around my draft upper, but that won’t be too difficult to clone out. The goal here is to have as clean as possible a section of the knit texture filling out the largest surface area of the focused panel as possible. I won’t worry at all about lighting or even large seams until I get the entire base down. I’ll roughly mask it out so the areas outside the ankle panel aren’t too distracting, and quickly clone in a small section of base knit texture over the protruding area of the rubber midsole overlay.

Now I want to move onto the middle section, which is where I intend for the logo stripes to sit. I want to keep the shifting direction of the knit texture from the City Sock, and make sure the direction of the pattern works with the other inner silhouette lines I’ve established in my mock up.

The base middle panel from the original is too short horizontally, so I’ll ‘stretch’ the pattern by laying down a base section first, then shift two duplicate copies over to the left and right of it. This extends the baseline texture pattern beyond the boundaries of my intended middle panel. At this stage, I’m not really worried about the seams in the knit texture, as the logo stripes will cover some of those. Any other noticeable transitions remaining after that, I’ll clean up along with all the other prominent seams I’ll undoubtedly make. My real focus here is simply making sure that the overall diagonal direction of the new knit texture is relatively well matched across the three sections.

You’ll notice I’m not bothering at all with detail in the top section, where we have some very jagged seams between the texture pattern and the original background edge are very prominent. I’m going to recreate the tongue I need from all new sections of material, that are more aligned with the perspective they should have at the top of a moderately cylindrical shape.

So now I have the heel and middle panels roughly arranged, I need to lay the toe panel down. I’ll duplicate the original base again, position it accordingly within the draft sketch guidelines, aiming to fill as much of the sketched panel area with as clean a knit texture foundation as possible. Once I’ve done that, I’ll quickly mask it out and put a new layer of baseline clone knit texture over any noticeable trouble areas. Again, at this stage I’m not at all concerned with how rough the modifications look. I just want to get the baseline fill down. You can see I’ve removed the prominent construction seam from the texture pattern as well. I’ll need to add something like that as I get closer to the detail adjustment stage, but for now, it’s an unnecessary consideration.

Even though I’m going to try and rebuild the tongue from much smaller, more detailed sections of other sneakers, I’d like to try to get a foundation down for the knit texture here as well, if for no other reason than to stop the large sections of original white background from being so distracting. It also helps me to begin to visualise the intended edits more clearly, by giving me something more closely approximating the desired finished result.

I’ll take the front, topmost section of the middle panel, where the knit texture starts to move away from the lens, to best mimic the perspective shift that should happen naturally at the top of the silhouette. I lay that down at the general toe/middle seam area, then duplicate it and shift the new layer to the left, filling as much of the length of the intended tongue area I can. At this stage, I’m not sure how the layered double tongue will even look, let alone how it will meet the sock cuff, so I’ll leave the natural end of the duplicate section and focus on those details once the cuff is laid out. Now that I have a very rough foundation knit upper to work with, it’s time to see how the cuff will factor into this.

I’m not a fan of the city sock cuff for a fictional EQT sneaker. The City Sock is a little more limp than I’d like for a sturdier EQT aesthetic, so I’m going to take the cuff from the NMD R1 Gum, which has a more rigid elastic snap and weight to it. Aligning the R1 Gum cuff over the top of my in progress upper, I get the angle that I want while only falling short of the desired tongue height by a small amount. That should be easy to extend with some careful copy and paste work. I’ll roughly mask it out so I can start to factor the entire upper into how I progress as a more cohesive single piece.

At this stage it’s apparent how much work will need to go into creating a more seamless finish. The luminance and saturation discrepancies are pretty glaring, which can sometimes make for a large distraction early on. Oftentimes, I’ll encounter students who begin to fixate too early on in the mock up process over fine details. It’s better to disregard them for now, accept that the overall piece will look relatively awful and inconsistent, and continue to lay down the foundations for the other sections. Once you have those in place, then you’ll know which details you’ll need to refine. For now, I might not even be able to see certain sections that are glaring to me as I approach the end of the work. They may be obscured partially, or even entirely, by other areas I’ve added along the way. There’s no point obsessing over small details at this stage if you don’t even know how they’ll factor in to the finish as you near completion.

With the majority of the sock upper laid down, it’s time to move onto the midsole.

Outer silhouette: building the concept sneaker midsole

I’m going to base the majority of the midsole on the Originals EQT Support 93/17. While it’s entirely subjective, I feel like the defining features of the EQTs, which I’d like to retain as my number one influence in my concept, comes from the midsoles. The 93/17s have a more casual/street outsole, which runs flat along the base of the midsole. I’d like to mix the lines of the 93/17 midsole with a more cross trainer influenced outsole. For now though, I’ll try to get the overall base placement of the midsole arranged.

The main shape and lines of the midsole are close to what I want, barring a few small changes. Firstly, to accommodate the more equal division of the three upper panels, I’ll need to make the bridge between the heel and front section of midsole more narrow. I’d also like to extend the silhouette of the outsole of the heel, to shift the weight of the design back slightly, which should also help to even out the effect of the intended cross trainer style outsole I’m hoping to add. The logo stripes will also be shifted back from their original placement to align more accurately with the desired logo strip location on the upper. With the midsole in place, I’ll roughly mask it out along the top edge, then duplicate the front transition into the raised ‘bridge’ and shift the duplicated area back to the left a little more, then roughly mask that out as well.

As expected, there are some significant detail seams along the midsole, both the boost pattern as well as the surface and edge details that form the upper most region of the midsole itself, but I’ll make the necessary corrections to those once I’ve established more detail in the middle panel of the sock upper. I’ll also leave the logo stripes on the midsole where they are for now, until I have a clearer idea of the placement of the upper logo stripes.

Before I move onto the upper details, I’ll focus on getting the more EQT-like outsole blocked in.

Outer silhouette: modifying the concept sneaker outsole

I’m going to use the more textured outsole from the EQT Support 93. It’s got the kind of ribbed edge that I’d like to include to retain a more cross trainer like EQT feel to the silhouette of my concept. It’s not really the correct shape however, and as I’ve started to establish more of the concept at this stage, I’ll need to modify it more carefully than the midsole.

Firstly I’ll cut it out and run it through a little puppet warp to shift and reshape the bottom edges to align to my established outsole in a more aesthetically pleasing manner. Then I’ll need to chop it up and place it in multiple sections, to get it to work with my current in-progress work. The best approach here will be to work with the three panel split notion, and arrange the outsole in a front, middle and rear alignment.

The front toe section extends down to the beginning of the front-most base logo stripe. Although I may shift those later, that’s where the initial curve upwards of the original outsole really begins. Shifting the remainder of the original outsole back to the left allows me to work with it’s natural curve, and appears to be pretty decent at completing the outsole. However I’m not really happy with the understated, flat like curve angle in the heel section, and the additional midsole detail is something I’d like to add. To correct this, I’ll duplicate the outsole and warp the heel section independently of the mid region, getting the rounder curve I was after. This brings the outer edge of the outsole more into line with the dipped curve detail of the midsole.

Disregarding all the glaring seams I’ve now added to the in-progress work, I’ll roughly mask each section out along the upper transition edge to see how it sits. I’m also not happy with the gum outsole, both in terms of colour and luminance, so I’ll need to work on those shortly as well. Add them to the growing list of modifications I’m going to need to make to get this anywhere close to transparent.

Inner silhouette: aligning the upper heel overlay

Before I place the heel overlay, I’m going to position the contrasting, rigid heel ribbing from the City Sock into a more prominent position around my upper silhouette edge. That not only gives me a more realistic looking heel profile, but design wise makes a little more sense for the Primeknit upper, as it would be far too flimsy with no secondary heel support back there, no matter how minimal.

Next I’m going to use the heel overlay from the NMD R1. There are a couple of colours I could use here, to ensure I had a sufficient base region of highlight luminance to work with, however I’ll keep it consistent and use the white variation. I’ll continue to keep the overall foundation of the sneaker a few different values of relatively desaturated white, which will make it easier to work with adding variations of colourway values in the final stage.

I’ll align the heel overlay in the general position I’m after, not taking into account the misalignment of the sock cuff to the newly placed heel ribbing of the upper. Masking away the unneeded elements gives me a better idea of how it fits into the overall design, and shows a few of the pictorial elements I’ll need to clean up in the next pass. I’ll need to spend some time not only making the transition between the new heel overlay and the upper sock texture more seamless, but also between the sock texture, the heel overlay and the sock cuff. The overall finish is obviously looking very roughly cut and paste right now, but that’s completely fine for a first blocking draft.

I’ll likely leave the new double tongue until last, as I’m no professional designer and am not really sure how it’ll either a) work visually, or b) fit into the intended middle section of the design, which I want to be quite busy to work with the EQT aesthetic. With that in mind, I’ll focus now on the the entire middle panel details; the lace overlay, the side logo stripes, the overlay and the midsole logo placement.

Inner silhouette: positioning the Primeknit upper logo stripes

I’ll set the overall placement of all these elements relative to the side logo stripes, as I want to establish a more equal distance between the toe and heel of the silhouette relative to the more centre weighted logo stripes, along with a centred placement of the side logo relative to the width of the middle panel. There’s a natural seam in the upper sock texture between the toe and middle panel right now, but I plan to shift that back a little in my more detailed refining stage.

I’m going to use the logo stripes from the NMD R1 Gum, as they have just enough surface and edge texture to contrast against the sock knit pattern, and I wanted a more minimal logo stripe, one that wasn’t stitched onto the upper, which in this case, likely wouldn’t really work given the material of the upper sock texture itself.

I’m working with these separately, so I can control the distance placement more accurately and to my own specifications. Once I’ve got them down to a placement I’m happy with, I’ll focus on bringing over a new middle overlay that will bridge the midsole and the middle logos. I’m doing this for a couple of reasons; firstly, it’ll keep the design more in line with the established EQT aesthetic, which tends to be busier than the more minimal lines one would find in casual/street silhouettes like the NMD. But further to that, I’m really not a fan of the progressively elongated logo stripes that not having a middle overlay would create. The rear stripe will be longest, the middle becomes a little shorter than that, then the front stripe is shortest. It extends the entire middle region of the sock into an awkward visual finish (I tried, not a fan) and makes the intersecting lines of the logo stripes look quite haphazard, for my personal sensibilities.

Inner silhouette: sourcing a suitable side overlay

At the time of creating these, I couldn’t find a lighter middle overlay that suited my needs. There was a lighter suede overlay on the camo print turbo red EQT Support 93s, but I didn’t want to use suede. I felt like it would be quite jarring up against the plastic heel overlay, so I wanted something that would be a little more subdued and not make too strong of a statement texture-wise.

Instead I went for the overlay from the ultra turbo red Originals EQT Support Ultra. It’s more of a leather styled overlay and as such theres more of a surface consistency between it and the heel overlay. It is, however, black, which doesn’t really work to what I want for preparing the sneaker for multiple colourway variations. I’ll need to modify this to get it in the right spot for that step, but for now, I just care about placement.

I’m going to work with the midsole curve for my best placement, and also keep the diagonal line caused by the direction of the heel overlay in mind, relative to a desired rotation of the overlay. The midsole curve, which creates a straight bridge between the front and rear curved sections of the midsole, will also dictate the seam placement when I’m trying to establish a stronger division between the texture pattern of the rear and middle panels. I shift the position of the overlay relative to both my draft guidelines and the midsole bridge section, but remember any (or all) of this can change once all the pictorial elements are in place. Even though the design is coming along, it’s still very much in its draft phase.

So with a rough placement in mind, I’m going to mask the unwanted elements out, which reveals a a small gap between the natural end of the overlay and the midsole transition. I could remove that entirely by lowering the overlay a bit, and letting the natural end of the overlay meet the midsole, but I really want the downward sloping angle that the overlay creates to kind of parallel the angle caused by the downward slope of the heel overlay. If I drop the middle overlay, it’s lower than the implied line that runs off the heel overlay, so I’ll simply extend the edge with a small section of duplicated middle overlay. Now I can mask the unneeded elements of the logo stripes away to get a better idea of how the stripes sit on the sock knit.

Obviously they look really awkward, as they have no sense of raised height to them, so I know immediately I’ll need to create a sense of shadow around them to convey that height. However, at this stage I’m just interested in if the surface texture qualities of the logo stripes do enough to differentiate themselves form the texture qualities of the sock upper. Next I’ll shift the original midsole logo stripes back towards the heel to bring them into line with the main side logo stripes.

This is a pretty easy edit, I’ll just take each logo stripe, duplicate it to a new layer and shift them all back to align them with the main side logo stripes. In order to do this a little more transparently, I’ll also skew the stripes a small amount horizontally, so they fall in line with the angle of the side stripes. Then I’ll just mask the outer areas away and clone over any noticeable seam edges from the original logo stripes. Normally I’d leave this sort of detail edit for the next stage, however these are super distracting, so I’ll clean them up now. With all of that done, I’m now ready to move onto the lace overlay.

Inner silhouette: positioning the lace overlay

I’m going to use the plastic lace overlay from the NMD R1, which is the same source for the plastic heel overlay. That gives me consistency in surface texture qualities between the two overlays, even though the colour and the luminance values of the lace overlay are a little out of balance for the overall finish. I want to use the overlay from the NMD R1 because with the three lace buckles, it’ll nicely mirror the three individual stripes I have for the side logo, keeping the design elements consistent in the middle panel.

Once I get the rotation sitting right, I’ll mask out the superfluous elements, leaving both the laces and the lace overlay. We get three bridges of lace details to work with, extending the three striped motif from bottom to top. The other advantage of working with the original laces from the NMD R1 is that I can keep the shadows and use them to guide me in creating newer, artificial shadows for all the upper overlay elements, anchoring them more seamlessly.

I’ve got a little stripe and upper detail leftover below the overlay at the top and bottom regions, so I’ll duplicate a new section of the original middle upper knit texture region and mask it in over the top of those areas that are standing out.

With all of that placed and looking decent for a starting point, I feel like I’m ready to move onto the double tongue test.

Inner silhouette: creating a unique, layered upper tongue

The sock tongue is going to be the one truly unique design element in this mockup. Normally with sock style uppers the cuff will link around the ankle joint and the top middle section of the upper will stretch around the foot. This makes for fantastic street and casual style silhouettes, but it doesn’t really sell the EQT aesthetic that I’d like to be the primary design sensibility in the mockup. I came up with the idea of a layered tongue that bridged the three upper panels and gave sense of stability to the ankle region of the design, bringing the silhouette more into line with the EQT style. Whether it’s feasible or practical in real life isn’t really my concern here, as I’m not genuinely designing sneakers, simply coming up with fictional mock ups.

What I need to complete this section are a finished sock cuff around the top front part of the ankle, as well as a differentiated and layered secondary tongue that sits above the sock upper. I’ll work on finishing the sock cuff first.

I’ll use the top section of the cuff from this image of the NMD R1 Gum. This is a pretty poorly retouched image to begin with, you can see all sorts of haphazard cloning, feathered edges and colour and luminance shifts indicating heavy modifications. Once I’ve got the rotation I’m after, I’ll mask out the remaining sections I don’t need, leaving me with a relatively useable foundation base to work with. Next I need to start laying down the details of the double tongue.

I can work with the upper knit texture that’s already there as an initial foundation. However I need to extend the knit texture along a little further to the left, to extend where the tongue would be. I’ll use a section of the same perspective knit texture from earlier, shift it along to the left and mask it out once I’m happy with the length. Next I need to establish the new area as separate to the details underneath it with some PU leather piping.

I’ll use the piping from the NMD R1s, that established a fairly rigid looking frame around the upper and ankle. I’ll need to do some heavy modification to make this work, as the original image has laces intersecting with edges, and obviously the shape of the piping is different to what I’m after. I’ll mask out the noticeable areas I won’t need, going right up to the lace intersection, then lay down a new region of similar material piping from a different colourway of the same sneaker. That gives a longer straight edge to work with, and with a little puppet warp and distortion I have enough to work with to extend the frame along to where the lace overlay meets it. Then I’ll mask it out to give me a more seamless transition, establishing the positional hierarchy of the lace overlay relative to the tongue below it.

Next I’ll take that duplicated section of piping, shift it to the right and do a little extra puppet warp to get the angles correct, to extend the tongue down below the lace overlay closer to the toe. This will bring the tongue ‘panel’ in line with where I intend to create a noticeable seam, creating three panels to the sock upper. I’ll mask that out, leaving only the areas I think I’d like to work with. Next I need to work on the topmost tip of the tongue.

As it currently sits, the area where the sock cuff meets the tip of the tongue looks pretty bad. I intend to add a fabric logo over the top of the piping, as it would sit in real life, and bring that topmost diagonal line of the cuff and the tongue into a more cohesive state.

Inner silhouette: creating a logo label for the layered tongue

I’ll use the fabric logo label from the NMD R1. In it’s original state, the bottom most edge was cut off by the laces, giving it a chevron style lower silhouette. However, once I mask the unneeded parts out, even in it’s current state, you can see how it does wonders at bridging the two regions of the cuff and the tongue together. I’ll take the topmost edge of the label, with it’s straight line, duplicate it and shift it down and to the right to use as a label extension. We obviously have the shadow from the original top edge, but once I mask it out you’ll see the speckled texture of the label aligns relatively well, looking even at this stage like a cohesive single label.

Outer silhouette: refining the outer mask

Now I’ll just mask the outer silhouette edge of the remaining visible elements out, giving me the overall base to work with for the concept. I’ll need to add a lot of detail elements, such as stitching and shadows, alongside the obvious surface and edge fixes that stand out.

Now marvel at the raw, unfinished beauty of your frankenshoe

Or don’t, perhaps, as the case may be with our current cut-and-paste monstrosity. As mentioned countless times throughout this piece – under the guise, absolutely, of useful education, but also motivated significantly by the anxiety-inducing difficulty in presenting such unfinished work to the world – there is still a lot more work to do.

If we compare the current state of our concept sneaker with the finished example from the beginning of the piece, there’s a pretty massive divide. With digital composites of any kind, I always approach the process in two stages. In this piece, we’ve covered the foundation steps of making our concept sketch a partial reality. We’ve filled in the majority of the base we need to work with. We now have the material surfaces in place, and we can turn our attention away from constructing and arranging, toward the more refinement focused process of detail finishing.

In this post, I’ve discussed the methods used to complete the first distinct stage of concept sneaker photo-illustration; namely, the arrangement focused blocking-in process. The second stage is much more focused on traditional digital finishing and retouching concerns. In the next instalment of this series, I’ll present the tools, techniques and approaches I use to not only complete the illustration as seamlessly (and photo-realistically) as possible, but also discuss the methods I use to prepare the file for exploring and creating realistic alternative sneaker colourways.

Image Attribution, Copyright and Intellectual Property

All original illustrations and retouching by M. Seth Jones.
All of the source photographic material, sneakers, logos and trademarks presented here are the sole property of Adidas, and have been used with permission. This article is presented for both educational and editorial use, and as such, I’ve used the source photographic material, sneakers, logos and trademarks here solely for demonstrative, non-profit purposes.

Creating a Fictional Concept Sneaker in Photoshop: Part I

By | Uncategorised | No Comments

Creating a Fictional Concept Sneaker in Photoshop

Part I: Drafting and Refining the Initial Sneaker Concept Design

This article is an adaptation of a workshop first presented by M. Seth Jones in 2015

The Adidas sneaker you’re looking at here doesn’t actually exist. It’s a completely fictional digital composite; a photo-illustration created entirely in Photoshop, combining mostly pre-existing elements from six separate Adidas sneakers.

Finished digital photo-illustration of Adidas NMD EQT concept sneaker by M. Seth Jones, completed entirely in Photoshop.

Combined, the illustration contains 112 layers; 67 of these are pixel layers, 45 of them are adjustment layers. Together these layers are relying on 90 discrete local masks, all arranged within 45 groups. It was completed in two distinct stages; first, the arrangement focused blocking-in process, followed by the refinement heavy finishing stage.

In the first part of this series, I’ll walk through the approaches and considerations I’ve used to design the initial sneaker concept sketch. I’ll use pre-existing design elements from various Adidas NMD and EQT releases for consistency, and introduce an entirely unique feature to create a stronger sense of harmony between the two contrasting aesthetic styles.

The most important concept sneaker design considerations

The main consideration when approaching any form of digital composite, that uses pre-existing raw material from a variety of different sources, is to approach it in the most traditionally sculptural manner possible. Unless you’re incredibly lucky (or able to work with the photographer as they shoot the raw material, and the finished piece has already been through concept and draft development prior to shooting), you’ll likely never find a cohesive, seamless fit occurring when you digitally smash multiple elements together; you’ll encounter colour shifts, disparate surface and edge qualities and luminance, perspective and resolution mismatches. You’re essentially taking pre-existing values for both the outer, defining silhouette and the inner, material surfaces, values that only make sense as discrete entities, and forcing them to play nice together. With this in mind, I like to focus first on simply making sure that all the elements sit together in the frame first, not focusing on how they might align, or errors in if they all snap together, or how noticeable the seams are.

Once I’ve gotten all the defining material features sitting in the frame, I start to think about how I can finish the composite in the most photographically transparent manner possible. By that, I’m essentially thinking “how can I make this look like a photorealistic image”, like something that literally existed in a studio setting and was captured by a real life photographer. That means I need to think about the lighting not only for the overall, global finish, but also how the lighting on every single local element making up the composite sits in relation to the additional local elements that border them, and finally how each of these combines to create a cohesive whole. I need to evaluate the lighting present in actual photorealistic images; even though the majority of the lighting present in the raw material will be optically ‘truthful’, they will be independently so, and in order to make all of these disharmonious elements sit right together as a new, combined whole, I’ll need to extend those qualities found in the raw material to the new, recombined final piece.

Creating unique sneaker concepts from pre-existing designs

So let’s have a look at all the sneakers I’m going to use for this walkthrough.

I tend to approach evaluating the sneakers at this stage in terms of two considerations: outer and inner silhouettes, or perhaps outer silhouettes and inner materials. There’s an overall outer edge to each of these individual pieces, which generally constitutes the sneakers upper, midsole and outsole. Then you have the design elements that exist within the boundaries of the outer silhouette, that contribute equally to defining the overall design of the shoe.

What I’d like to do here is take the upper from the NMD City Sock 1 Primeknit, the midsole from the Originals EQT Support 93/17, and the outsole from the EQT Support 93, as shown above. Let’s break these three sneakers down into just their outlines.

This gives me a cleaner, less distracting foundation to work with for draft sketches. Next, I’ll combine those three elements to see what I’m left with.

What we have, with just these three elements combined together, is pretty minimal. This makes sense, as the majority of the silhouette is dominated by the NMD City Sock Primeknit upper, which is a very stripped back casual/street style release. I’d like to take the aesthetic of the NMD City Sock and combine it with the design language of the EQT range, which tends to be busier in construction, being more of a traditional cross trainer. What I’d like to do is introduce some more design elements to the upper, in an attempt to bring that general EQT aesthetic to more prominence.

To fill the inner silhouette in, I’m going to combine pre-exisiting elements from other NMD and EQT releases. I’ll use the sock cuff and the lace overlay from the NMD R1 Gum, along with the heel overlay and the tongue label from the NMD R1.

So let’s break down each of the already established elements into their outlines first.

With those sneakers broken down into their silhouettes only, I’ll take the inner details from them and combine them with the previous draft mockup sketch, to see how they all sit together.

It’s not too bad really, and a huge testament to how consistent the design standards of both the NMDs and the EQTs actually are, in that they don’t really look that out of place when haphazardly mashed together. I haven’t really presented them with any kind of placement consideration, instead opting to simply align elements as close as possible to where they were on the originals. There’s a bit of a weight issue to me, however, in terms of how the upper and middle areas are sectioned off as well as the overall distribution of a cleaner sense of white space, for lack of a better term. The rear section of the sketch feels a little too long horizontally to me, which in turn compresses the front toe section. I’m also not a huge fan of either the angle, or the length, of the logo stripes, which are pretty integral, as they serve as a sort of right angle intersection for the converging lines of the top and bottom angles. The height and placement of both the sock cuff and the the heel overlay are kind of clumsy and not working together all that well. The downward angle of the heel overlay leads the eye toward the toe, but the length of the side stripes kind of abruptly interrupts the implied line. Then there’s this huge open section at the top of the upper that runs all the way along the cuff seam, which definitely amplifies the expanse of white space in the rear heel area.

Refining the concept sketch and comparing it to our raw mockup

What I might do is use a three panel construction on the City Sock upper itself, with defined seams that trigger a knit texture direction change. Then I’ll use the side overlay from the Originals EQT Support Ultra to create a lace overlay/silhouette top parallel, right angle intersection of the bottom of the side logo stripes, which will also extend the implied line running from the heel overlay. I’ll rework the height and angle of the sock cuff and shift the placement of the heel overlay. Finally, to address the top of the sock issue, I’ll attempt to implement an original design element for the upper: a tongue overlay, or layered tongue, that sits above the City Sock Primeknit upper.

Here’s the final refined sketch, modified with all of these considerations in mind. Compare the subtle changes between our earlier raw mockup, which was arranged with no real placement or design considerations in mind, and our more considered final concept sketch. Even though the difference between the two seems minor, compare the overall weight of the finished design with the original mock up. The seams are more cohesive, and the shifted silhouette lines create a stronger sense of cohesion between each discrete section of the design.

Now what I need to do is start filling in the silhouette with materials. This will be a two stage process, that will first just kind of lay down raw materials and textures to align to a general placement. They’ll likely look incredibly rough and not at all harmonious, as they’re coming from a range of different sources all lit slightly differently, or they exist in an arrangement or perspective not intended for my final concept design. Getting all of these elements to sit cohesively in a finished, seamlessly transparent finish will be the main goal for the second stage, where they will all need to be reshaped, refined and recoloured, lighting added and contoured, and prepared for recolouring.

In ‣ the next part of this series, I’ll walk through the tools and techniques I used to achieve the first stage of these edits in the most seamless manner, and discuss more insights into why we all need to adopt similar approaches for photo illustration.

Image Attribution, Copyright and Intellectual Property

All original illustrations by M. Seth Jones.
All of the photographs, sneakers, logos and trademarks presented here are the sole property of Adidas, and have been used with permission. This article is presented for both educational and editorial use, and as such, I’ve used the photographs, sneakers, logos and trademarks here solely for demonstrative, non-profit purposes.

Deconstructing High-End Beauty Retouching: Part I

By | Uncategorised | No Comments

Deconstructing High-End Beauty Retouching

Part I: The Problem With Representational Post-Production

15 minute read | Insight, Advanced, Beauty & Portraiture, Retouching, Photoshop

It’s a misconception that the largest barrier to simply getting better at beauty retouching is purely technical. Beauty retouching belongs to that group of unique artistic processes that are equal parts technical, mechanical and creative. So what really holds retouchers back when it comes to advancing their skills?

Having worked closely with thousands of students over the last decade, the most common obstacle I’ve witnessed practitioners encounter in the advancement of their retouching skills hasn’t been caused by a limited understanding of software tools, commands or functions.

The most significant factor in holding students back has been a fundamental lack of appreciation for, and comprehension of, the impact both cognitive and mechanical procedures play in determining the quality of any pictorial finish. This is caused, in no small part, by the very nature of photography itself and largely dictated by how we view images.

Why photography prevents us from better understanding retouching

We tend to evaluate photography in terms of its surface qualities, as a finished pictorial artefact. Students of retouching commonly approach the act of visual problem-solving in a similar manner, emphasising the representational above all else. Elevating the peripheral effect to this status generally doesn’t compel retouchers to refine their understanding of the complete retouching process.

It affords them little insight into the ideal Tiered Retouching Model, which emphasises an equal balance of quality, speed and flexibility; heightens the retouchers value sensitivities; and promotes a more cohesive awareness of the process beyond the technical. A tiered approach is not only concerned with the finish. It factors in which edits to make at each particular stage, how subtle or varied each edit needs to be, how one edit interacts with another, and how all of this impacts upon both the finished image and the editing work-in-progress, the process itself. Understanding this process is vital for both post-production practitioners and photographers alike.

The real secret to becoming a better retoucher isn’t so much that you’ll continually discover newer, superior techniques being revealed to you as your skills increase. The reality is that, at it’s core, one could make a fairly persuasive argument that there are a fixed number of scalable, fundamental principles to photographic post-production and, once you’ve learned these, you’ll be able to confidently approach and overcome any number of pictorial problems of varying degrees of complexity.

It’s a given that modern technologies will continue to grant us seemingly newer editing abilities, access to higher resolutions, more granular control, more intelligent algorithms and vastly more complex colour models. Until photography however, at least photography in so far as post-production is concerned, evolves beyond the act of mark-making on surfaces both chemical and digital, the only editing factors that truly matter will be edges, surfaces, tones and colours. Four considerations. Make that three if you’re working in black and white.

The real secret to becoming a better retoucher is encouraging the development and refinement of your ability to confidently visualise, interpret and analyse values, and understanding that it’s these skills, above all else, that will both inform and advance your mechanical efforts, not the other way around.

The real challenge is in finding a way to achieve this.

How photographic representation affects retouching presentation

Ironically, it’s actually more difficult to learn these insights, despite the wealth of learning options currently available to us. We now have access to more information, both in terms of detail and volume, regarding other practitioners working methods than we’ve ever had. However it’s commonly presented in a manner that emphasises the representational above all else. Essentially, we’ve never been so well equipped with knowledge of how we can overcome any number of pictorial obstacles with our software while simultaneously being so unaware of why we should. Seasoned practitioners know that once a retoucher develops a heightened comprehension of the why, it feeds back into the how. Without that understanding however, ones working process often becomes ill informed. This is how bad habits begin.

So how can we strengthen our comprehension into the why of retouching? The most common approach is to turn to evaluating finished work, as it’s available to us, for further insight into the retouching process. The problem is, we really only get to evaluate finished work these days through a very narrow lens. It’s most commonly presented to us in it’s final state, stripped of all context to the un-retouched original. Occasionally, we get a chance to view it in a before and after comparative state, which is a deceptive method of display.

If we’re lucky to have any understanding of the process ourself, we can also attempt to evaluate the work in a slightly more insightful manner. The problem with this is we’re only going to be able to interpret our insight as defined by the comprehension we already have. This can be useful for those who have been practicing for years, but it’s value as a learning aide plummets for those who have already started building their workflows on top of those frustratingly common bad habits.

Additionally, these methods of evaluating retouching don’t afford an insight to anyone beyond the domain of aspiring, enthusiast or professional practitioners. We now commonly interact with hundreds, in some cases thousands, of images every day. The benefits of understanding and interpreting retouching, in terms of how it interacts with photography, how it modifies the message communicated by photographic imagery, how it feeds into the very act of looking and seeing, even at a basic level, are significant.

Imagine if we could only understand or appreciate the act of reading if we were writers ourselves. Writers may have a heightened awareness of the complexities of a given text, but anybody that can read can consume, understand and appreciate it. Even with a generally solitary act like writing, where the creative process occurs separate from the consumption of the finished piece, the manner in which the text reveals itself to the reader can grant non-writers the opportunity to witness the creators thought process unfurl as they progressively engage with the narrative as it’s presented to them. Photographic image consumption doesn’t present us with the same opportunities regarding retouching. Because of it’s very pictorial nature, we engage with it in a radically different manner. It would be like if we condensed every narrative rhythm, beat, cadence and dynamic from an entire text into a single word, then encountered that word in an instantaneous impulse, with all of the subjective experiences that word conjured reverberating after.

That nature in which we engage with photographic images, and by association, the manner in which we are most commonly able to evaluate photographic post-production, is sadly the biggest barrier to becoming more aware of the process of post-processing. We never really get to see the process of editing stripped of it’s context to the original un-retouched image. It’s not common to present retouching, by it’s very nature a photographic act in a way that allows viewers to evaluate the work on it’s own merits as a photographic act, and to reflect on how much of the final presented image has been modified, in what manner it’s been modified and to what extent it’s been modified. We evaluate beauty retouching on the final merged composite state of the image, or in regards to before and afters, which are a misleading way to assess beauty retouching.

When presented with a before and after comparison, the initial foundation value we’re presented with tends to set the standard of how we respond to the subsequent value. I’ve explored this very tendency in hundreds of lectures and presentations. When presented with the retouched after first, viewers tend to be more critical of the un-retouched before state.

With the same image and presented first with the un-retouched before photograph, audiences become more critical of the retouched after. Either way, a before and after comparison of retouching is generally presented for shock status more than anything else, and does not represent the process of modifying pre-exisiting optical gradients in an accurate or thorough manner.

While I’ve previously discussed both the detective and depictive qualities of more traditional photographic post-processing, and how they relate to the act of photography itself, one would do well to become more aware of these, in relation to the modification heavy approaches of beauty retouching, to gain a more thorough insight into the extent of the entire image editing process.

Evaluating the beauty retouching process with finished beauty retouches

Let’s use two finished retouches to explore how these qualities impact upon both the final image itself, and to evaluate how accurately they convey the beauty retouching process on it’s own merits. Firstly, its important to highlight that both of these images were completed for the sole purpose of being used as instructional aides. Their finishes are not necessarily representative of any kind of ‘universal’ or specific commercial standard, however the approaches used to finish them are. In their finished state, they should not be viewed as depictions of the aesthetic standard anyone else should set out to replicate, or the only kind of pictorial finish other retouchers should be able to recreate. If the un-retouched raw files were given to 100 different Creative Directors, they would likely be completed to 100 vastly different aesthetic and graphic standards. Additionally, they have not had any kind of global or grade-dominant colour adjustments applied.

Beneath the finished, merged composite surface, what we’re looking at here are two discrete retouching approaches being applied to their respective un-retouched raw files. While both of the images display a similar range of depictive qualities, and both fall firmly into a relatively well established beauty aesthetic, the approaches I implemented in regards to retouching them varied significantly. Despite their fundamental differences, the two approaches have been applied with the purpose of achieving the same end goal: simply, a standard high-end beauty finish. The final images share an aesthetic consistency and, with all the modifications merged and visible, I’ve attempted to complete them to a similar pictorial standard while utilising the different approaches, tools and methods.

Both approaches utilised a Tiered Retouching Model. In so far as just the retouching is concerned, the first image utilised an approach that favoured micro-adjustments, with a heavy emphasis on spot alterations, micro-transition modifications, relatively small brush radii and surface value sampling. Surface value sampling relies on the separation of surfaces into positive and negative values, and emphasises the replacement of negative values with pre-existing positive values. In the hands of a skilled practitioner, micro-adjustment retouching affords the retoucher a much more natural finish due to it’s re-use of natively occurring variation. The second image utilises a significantly more heavy-handed macro-adjustment approach, which leans more prominently on whole surface adjustments, macro-transition substitutes, comparatively larger brush radii and surface value replacements. Replacing native surface variation foregoes any pretence of reusing positive surface values at all, and simply substitutes entire pictorial elements with artificially constructed values, resulting in a much less natural finish. Due to this, a retoucher utilising the surface value replacement method is generally required to spend more time refining and balancing their edits, in order to achieve the most transparent finish possible.

There are equal amount of pros and cons with both approaches, and the methods favoured by retouchers generally come down to personal preference. Some practitioners will not only prefer the finished pictorial qualities of one over the other, but find they can utilise the tools and methods faster; that the value shifts one approach makes over the other align more closely to their own unique value assessment, interpretation and intuition; or one approach affords them more flexibility during the revision process relative to their professional outcomes. As always, there isn’t a universally better or correct approach, and most professional retouchers will seek to master variations of as many different working methods as possible, so that they are best able to overcome the broadest amount of workflow obstacles for the largest given set of circumstances. A modular workflow serves this purpose.

While I’ve gone to great lengths to stress that there truthfully is no strict Universal Standard when it comes to the finished qualities of beauty portraits, universally there are thousands of individual interpretations regarding the desired pictorial outcomes. These interpretations will shift due to hundreds of varying social, cultural and professional variables, but I am a firm believer that there are significant amounts of identifiable overlap across the many varying points of view. For example, it’s not a stretch to claim that the main focal point for beauty portraiture is generally always going to be the surface of the subjects face. Depending on the brief at hand, other elements such as hair or accessories such as jewellery or glasses may nudge their way into the frame of view as well, but very rarely will they become the dominant feature, relegating skin surface to a lower priority.

To many non-practitioners, the end goal for beauty retouching is commonly classified along the lines of simply smoothing the skin, but this is overly reductive. While on the surface (no pun intended), the pictorial motivation of retouching can often result in smoother looking skin (amongst many other qualities, ideally), for practitioners the main focus of the beauty retouching process should always be on both macro- and micro-variations in the pictorial surfaces and edges of the file. Generally the goal is going to be either:

  1. reducing the contrast of micro-variations in the edges of the file, and/or
  2. creating more flattering transitions in the surface gradients of the file.

What constitutes more flattering is relative to the file at hand, and will depend largely on a range of factors, including:

  • Image aesthetics
  • Commercial purpose of the image
  • Subject gender and/or age
  • Depictive optical qualities (lighting, composition, pose, makeup)
  • Social/cultural considerations
  • Client/subject input

So for example, the approach required for a photograph of a subject with a very matte looking skin complexion is going to be different for a subject with a very glossy, high-key reflective complexion. Retouchers are still dealing with both the edge micro- variations and the surface macro- and micro-variations, but the pictorial qualities of both matte and glossy surfaces and edges require the retoucher to both think through, and work through, two very different sets of problems. Further to that, surface and edge qualities shift between positions of the same element, so the surface and edge qualities of the skin directly under the eyes are very different to the surface and edge qualities of the skin on the forehead, which differs greatly to the jaw, and again shift when contending with the cheeks, and so on. If a retouchers goal is transparency above all else, then the main consideration of any tool, method or approach used to achieve it should be that it mimics as closely as possible the pre-existing detective and depictive qualities of the pictorial element being edited.

Keeping all of this in mind, we can return to the two images from the beginning of this section, to see just how few, if any, of these considerations are apparent when viewing a beauty photograph in it’s final state. A photograph, no matter the subject or intended aesthetic style, is representational above all else, as both a pictorial artefact and in regards to the manner in which we view it, and that includes all post-production baked into it’s final form. We really have no way of understanding any more about the photographic process of photographic post-processing, and how that might affect the finished image, the work-in-progress image and the photographic act of capturing the image itself, before any digital post-modification even begins. Most importantly, without any further insight into both the modification process that runs parallel to the finished modifications we can gain little insight into how to identify, read and interpret photographic values and the impact they have on our own aesthetic standards and goals.

All of these shortcomings are a barrier to developing a more sophisticated insight into the process of retouching, in turn obstructing both practitioners from learning more flexible, modular or tiered post-production approaches, and non-practitioners from developing a more relevant pictorial vocabulary. Additionally, modern beauty photography as it’s most commonly presented is the biggest obstacle for aspiring modern beauty photographers to refine their craft. Before and afters don’t help either, as they are simply a testament to the pictorial division between two vastly different states. They’re like being given the first instruction and the final instruction from a set of navigational directions; it’s of alarmingly little use to us to when all we can see is where we begin, and where we end.

So how can we move forward? How can we overcome these obstacles and become better viewers, better translators, better interpreters and better practitioners of contemporary digital photography? And how can we adapt any insights we might uncover into the modern digital process to aide us in revisiting and reevaluating photography, as both practise and artefact, from it’s analogue opitcal and chemical origins? In an upcoming part II of this series, I’ll present the first of two methods we can use to analyse modern beauty retouching, on its own merits, to help us overcome these barriers and along the way, uncover a surprising insight into the very heart of photography itself.

Image Attributions

All retouching and illustrations by M. Seth Jones.
Unretouched photographs provided by: Garazi Gardner (1-3); Hannah Khymych (4-7); Emily Abay (8); Garazi Gardner (9); Ivan Monge (10); Jocke Jonsson (11); Garazi Gardner (12-19); Ivan Monge (20); Hannah Khymych (21); Grant Thomas (22); Garazi Gardner (23-29); Ivan Monge (30).
Used with permission. All photographs copyright their respective owners.

Expressive Black and White Post-Processing: Part II

By | Uncategorised | No Comments

Expressive Black and White Post-Processing for Fine Art Photography

Part II: Second Pass Approaches Explained

13 minute read | Walk-through, Advanced, Black and White, Fine-Art, Photoshop

In order to complete the black and white photograph I began in Part I, I need to adopt a more hyper-local, value based approach in my Second Pass if I wish to convey the refined, high-end fine-art finish I set out to achieve. Let’s go through these approaches in more detail.

Screenshot of work in progress comparison of First and Second Pass stages of black & white post-processing by M. Seth Jones, completed in Adobe Photoshop. Unretouched image: Johnston, Frances Benjamin, photographer. Carver's Old Mill, Gordonsville vic., Albemarle County, Virginia. 1935. Image. Retrieved from the Library of Congress.

The image on the left is the complete, finished file. The image on the right is the in-progress merged composite as it currently appears after applying the First Pass editing techniques I covered in Part I of this series. If you haven’t read the first pass overview yet, I’d encourage you to ‣ go here first, then come back when you’re ready to find out how I approached finishing this photograph.

In the First Pass of processing this black and white image, the approach I take is much more in line with traditional darkroom techniques; there are digital equivalents for basic proofing, test prints and basic darkroom based dodge and burn. This helps to establish Photoshop as a feature rich complement to fine-art printing methods. In the Second Pass stage of editing, we take the capabilities of Photoshop to a much more precise, hyper-localised tier, affording us editing methods, approaches and options we could only dream of with an optical enlarger.

Always use the values in your images to accurately finish your photographs

Let’s have a look at the before and after transition of the current in-progress file, and the complete finished file in motion first, to compare the difference between the two states. For the purpose of this video transition, the completed finished image appears on the left, and the in-progress merged composite containing the First Pass stage of edits is on the right.

All of the base composite adjustments were made with either first or second order curve response modifications. They were masked locally, but in a near binary manner, similar to how I would make test exposures in the darkroom; select a general area of the frame, cut or boost its luma values. There was a little gradation on the local masks, but only to minimise seams. It has been, until this point, rudimentary depictive masking.

Now I need to turn my attention to a more detective form of modification, in order to transparently finish the file. This is done in four discrete stages, into a final targeted mastering pass, before sending the image off for a spotting and output sharpening pass. These are progressive modifications, meaning that I will refine successive passes based on the adjustments I’ve applied in previous steps.

Balance the primary and secondary global elements of your file

The subject and it’s immediate surrounding environment is my initial focal point. The subject itself is too flat individually, and too muddy in relation to the overall finish, so I need to bring it out in the mix. It’s still essentially the brightest pictorial element in the file, and it’s compositional focus ensures that, even when lacking some amplified light quarter tones for additional luma planes and compositional impact, it’s not in danger of being lost entirely in the finish, but it needs some attention.

There is a little luma clash going on between the subject’s camera left arm and the barrels immediately behind, but there’s also a subtle, natural gradient from plus-right to minus-left on the barrel itself. The rigid structural horizontal and vertical right angle repeating lines on the barrel do establish enough of a marked difference to separate it from the softer, cascading fabric folds and creases of the subjects jacket, however I’ll add a few new planes of luma values to the arm in this step to further differentiate between the two elements, creating a stronger sense of perceived three-dimensional depth within the two-dimensional frame.

I’ll clip two channel modifiers together here for this adjustment: the modifier itself, a second order response curve targeting the upper transition between the mids and the light quarter tones, with a broad bandwidth rolling off into the lower mids and highlights. This response curve boosts everything below roughly 190 with a luma rolloff, into a narrow shelf cut around 210. This needs to be mixed into it’s detective surroundings, so transparency is cut to a little above 50%. This is being controlled by a manually added mask that allows 100% of the partially transparent modifier to show through from the crossed knee upwards – the jacket, the shirt, the subject’s face and hat – then rolls off gradually from that targeted focal point down the legs, along the two lightest grain sacks (top and bottom only) and spilling outside along the wooden steps. This helps to introduce a few new planes of luma values to the file. They’re closer to traditional mid tones than anything else, but because the foundation values of the file were established at such heavy, low densities, they have the aesthetic impact of behaving closer to light quarter tones and highlights in this particular boost.

Continuously review your previous adjustments as you progress

The difference the initial subject refinement made to the overall finish made me want to apply a similar modification to the front steps. The exterior of the cabin should be naturally brighter, but it’s looking a little dull against the previous modification. The contrast between the luma values of the front grain sack and the wooden step surfaces is jarring to me as well. Ideally, I’d like to create a gradual scanning roll off for the viewer from the subject/interior to the bottom edge/exterior of the frame, and it’s just too stepped for my liking at this point. The bottom shoe is a bit too dark as well, and gets quite lost in the transition from the dark quarter tones of the step edge to the dense blacks of the shadow under the chair and around the darker middle grain sack.

The process here is the same as the previous step, simply targeting different values. I clip two channel modifiers together: the modifier with a targeted luma mask, and the control mask which I’ll paint in manually. The luma modifier essentially boosts the mids and the lower range of light quarter tones, and has the same effect as the previous step. Even though it’s only a subtle boost to some fairly middle-histogram values, because the surrounding tones in the file are so dark, it impacts on the finish in much the same way as a generous boost to the light quarters and highlights would on a flat looking file. It adds just enough contrasting (relative to it’s initial baseline) luma values to the exterior steps, and because the steps have a natural tonal variance between horizontal and vertical edges, we get a much more pleasant, naturally stepped luma gradient to guide the viewer into the centre of the composition (or centre out), with accurate luma transitions.

It also brings the bottom edge of the composition into line, as far as mirrored luma values go, with the left and right exterior edges of the barn itself. This creates value harmony between the discrete elements of the barn exterior without losing anything in the overall finish; the surface and edge qualities of the wood in the steps is naturally different to the equivalent qualities of the left and right wooden planks. Structurally we already have clarity between the two elements, so we can safely have them share overlapping luma values without worrying about creating a muddy exterior and subject frame.

The control mask will target the front and top facing wooden floor and step planks, avoiding the grain sack and the subjects legs while partially including the bottom shoe. I’ll paint in a partial target on the camera right side darker horizontal plank that extends right beyond the door frame, and allow a gradual rolloff of the control mask to spill onto the ground at the bottom of the frame.

There are countless final truths, and none of them are correct

The process of gradually adjusting local values one element at a time often reveals the biggest caveat with post-production; modifying one value often reveals, or creates, the need to make subsequent modifications, throwing the balance of values out with even minor shifts. The initial base composite showed I veered off-tangent somewhat, in regards to the overall luminosity of the finished image; by setting a very dark initial baseline, and building up the environmental elements before the subject, I was perhaps too conservative with my mids and light quarter tones, having had the overall balance of the finish skewed towards the heavier shadows.

At this stage, I’m looking to push the subject, along with the pictorial elements within the frame of the doorway, out into the front of the mix a little more, for enhanced luma definition. It’s still a little too dark for my taste, but I need to be cautious of creating an obviously masked two-dimensional diorama by amplifying the subject too prominently.

Nested modifiers control one more local adjustment to the subject and grain sacks. Inner and outer control groups are active here; the outer targets the pictorial contents within the doorframe, as well as the first outer step down, while the inner stacks on this by applying a discrete modifier to the subject and grain sacks in a chain.

The outer modifier boosts the luma curve response just a little higher than 127. It’s a first order, single point lift that rolls off evenly at the upper and lower ends. The targeted luma mask applies that adjustment to the upper highlights only; it’s not an even luma rolloff here. The lower and upper shelf of the response ramp in a little to the left and right of the target value, which is roughly around 190 or so. This ensures that the fairly rudimentary global curve modifier deals most effectively with the values I’m after here, namely: boosting the light quarter tones relative to their current luma values. The base adjustment is exactly what I want, but it’s too loud in the overall finish and needs to be dialled back to around 30% opacity.

The inner modifier works on the same principle, but targets a slightly narrower band of highlights, with an overwhelmingly subtle control mask, reducing the intensity of the adjustment. I’m boosting the same range of tones as in the previous step, and halved the opacity reduction here, to stack two relatively minor adjustments on top of each other to achieve the overall finish. It’s a small shift, but plays a significant role in pushing the subject to the front of the finish, just enough to stand on it’s own in regards to it’s luma values. There’s enough luma definition between the inner compositional elements now to create pictorial clarity between them, as the inner centre of the frame is quite compositionally busy, which is at odds both with the overall pictorial qualities of the scene, and also the slowly transitioning diffuse environmental detective qualities.

Evaluate your finish after you’ve taken a long break from the screen

I have to admit, I actually thought I’d finished the image at that stage, ready for spotting and output sharpening. As is always the case, I let the file sit overnight, resting my eyes (and my sensibilities) before opening it up with a renewed perspective.

It was still too dark.

Tracing back through the process here, I can see I made a significant impact on the overall finishing process by establishing such a dense, heavy initial baseline.

The final finishing process is relatively straightforward: I want to brighten it up, basically. To do this, I drop in two modifiers targeting non-linear luma ranges, one focusing on the inner compositional elements – the subject, the doorway, etc – while the second focus on the compositional frame, so to speak – the exterior surfaces of the barn itself, as well as the exterior elements of the composition.

The inner adjustment rolls in around 140, targets around 160 and rolls off around 180, with a second order, narrow bandwidth minor boost around the 130 range. The base curve response is a standard single point, first order function that tapers off to 0 and 127 evenly. The outer adjustment is skewed more heavily to the right of the histogram, targeting the highlights, of which there aren’t many. In the purest sense, we don’t even really have true highlights. The brightest values in the file at this point probably aren’t even higher than around 200 or so. The lower shelf for the second target starts around 180 and rolls off around 220 or thereabouts. The brightest sections on the control mask probably don’t even push what we would consider to be lower mids, so it’s a very subtle modification.

What I get here is an optically accurate modification to the two main elements of the image; inner and outer pictorial elements, and an overall balancing of the final mix of luma values. While not quite as dramatic as it sounds, the overall finish does fall apart a little with only one of these modifiers applied. With just the inner adjustment, attention becomes focused on the unnatural compositional qualities of the subject, looking a little too poorly-collaged in it’s awkward balance in regards to it’s surrounding surfaces. Without the inner adjustment, the external frame of the file becomes the main luma focal point, pushing itself out to the front of the finish just enough to shift the overall compositional balance of the file. The weight of the composition skews towards the outside, away from the obvious pictorial focal point towards the centre of the frame. With both modifiers applied, each section gets luma accurate lifts relative to their current values, as well as each other, ensuring the overall balance doesn’t get thrown off.

Make your final surface and edge corrections and enhancements

Spotting is applied to a few distracting marks in the upper left of the heavy black background region. Some might prefer they were left as is, but the pure isolation of the central dust spot/scanner artifact against the heavy blacks just drew my eye too often. It’s a subjective thing, that’s the decision I made at the time. Any other day I might have chosen differently.

Output sharpening here is applied with a narrow band second order contrast increase to a finely tapered band of higher frequency edges, and dropped back to around 30% opacity to sit it right in the overall finish. In order to retain the sense of optical transparency that should be at the forefront of our post-production decisions, it’s being applied relative to the luma values of the merged composite. We want the sharpness to roll off relatively to the overall luminance of the image and it’s component pictorial elements and shelve back in as those values spike.

Trusting your expressive instincts against the lure of infinite aesthetic control

It’s output preparation time. Or is it? Honestly, the one thing that most of my students seem to stumble over the most is “how do I know when an image is finished?” It’s a bit easier to give a definitive answer for that when it’s client work; it’s the deadline, or when the client says its finished. That’s how you know. It won’t matter if you finish it a week in advance of the initial turnaround time, they’ll make sure you have enough revisions to see you through.

For personal work, however, it’s simply a matter of saying it’s done. Which is actually not simple at all. Everyones values are different. I could say a personal image I worked on is done when it reads correctly. However, as I’ve (hopefully) demonstrated in these two articles, what reads correctly means to me is, much like the 350 layers in your latest Photoshop masterpiece, a relatively convoluted mix of discrete interpretations of isolated imaging considerations. And every one of those considerations is entirely subjective, based on my own experiences and sensibilities. It’s not as clear cut as saying ‘my black point needs to be set to a value of 15 to avoid shadow clipping for outputting to that specific printer’. Not even technical imaging interpretations can lay claim to being a singular standard for truth, although they do have more rigid boundaries to keep their translators in place.

To paraphrase Ansel Adams once again, post-processing is a unique combination of technical execution and creative activity. When it comes to aesthetics, imaging interpretations are generational. They’re social. They’re cultural. They’re physical. They’re genetic. You can read other’s interpretations, but don’t take them as a definitive universal standard. It isn’t possible to tell another person what to see. The best I can do is give insight into my own unique interpretations of the process of first visualising a desired finish image, and the craft of achieving those results.

Image Attribution

Johnston, Frances Benjamin, photographer. Carver’s Old Mill, Gordonsville vic., Albemarle County, Virginia. 1935. Image. Retrieved from the Library of Congress, https://www.loc.gov/item/csas200904177/.

Expressive Black and White Post-Processing: Part I

By | Uncategorised

Expressive Black and White Post-Processing for Fine Art Photography

Part I: First Pass Approaches Explained

16 minute read | Walk-through, Advanced, Black and White, Fine-Art, Photoshop

To achieve a refined, high-end finish for our black and white photography, we should aim to adapt traditional fine-art darkroom approaches to our digital post-processing methods. This is most efficiently achieved in two distinct stages. Let’s go through them both in more detail.

Screenshot of before and after black & white post-processing by M. Seth Jones, completed in Adobe Photoshop. Unretouched image: Johnston, Frances Benjamin, photographer. Carver's Old Mill, Gordonsville vic., Albemarle County, Virginia. 1935. Image. Retrieved from the Library of Congress.

The image on the left is the raw, unprocessed file. The image on the right is the complete, finished file. The final file contained 28 separate layers, combining both global and localised adjustments. As with all of my higher-end post-processing work, this image was completed in two distinct passes. In the First Pass of editing, I restrict my modifications to global adjustments, made to discrete regions of the image. In the Second Pass of editing, I use localised adjustments to hyper-local elements of the image.

In this process walkthrough, I’ll cover the First Pass approach of post-processing. In ‣ Part II of this series, I cover the Second Pass approaches in detail, so be sure to read that once you’re done here.

In expressive post-processing, there is no photographic truth

Let’s have a look at the before and after transition in motion first, to compare the difference between the two states.

As you can see, the finished image is significantly darker overall, however pay attention to the way the light quarter tones and darker highlights shift in relation to the dark quarter tones and lighter shadows. There’s some interesting compression of tonal balance happening here. What’s even more interesting is that this final image is not correct in any sort of universal way. This is simply the finish I wanted to pursue. You might see it and think it’s too dark, or too flat in the exterior ambience, or maybe it’s overly sharpened. The reality is, all of those lines of thought are correct.

Now let’s have a look at the individual stages of the complete, finished image, step-by-step. This will give you a chance to see how each stage of editing modifies a single element in the image, while establishing a foundation for each successive adjustment, along with the overall aesthetic direction of the image. This is essentially a direct line into my personal train of thought throughout the process of working on the image, as each stage is presented in the order in which I applied it.

At every step of the post-processing workflow, we should always seek to take the considerations and motivation behind each discrete adjustment into account. We should always attempt to ensure the final image is a delicate balance of precisely varied single modifications, each playing an essential role in contributing to and achieving the overall finish. The challenge with expressive post-processing is knowing what the finish should actually be.

Anyone who tries to tell you that the craft of making a fine file (or print) is in someway connected to an objective photographic truth is lying. This is controlled expressive manipulation, plain and simple. The hand of the photographic post-processor is intended to be present as both a depictive and detective artifact in the finished work, however the goal at the front of our modifications should always be entirely transparent; that is, we should always seek to use only the values in the capture to modify the values in the capture.

To paraphrase Ansel Adams then (because I’m sure he’d just love that), finish quality is essentially a matter of sensitivity to values, which varies greatly between individuals. Rather than attempting to achieve a predefined standard, we should strive to ensure that the values of the final image suit the pictorial and optical qualities of the capture itself, and contribute to our intended visual effect. To better understand our visual intentions, we should always remain cognisant of the wholly subjective nature of our own instinctive visualisation sensibilities.

Separate the image into its main pictorial elements

Screenshot of an unretouched linear raw file with and without pre-processing pictorial separation. The coloured regions define the discrete sections of the image that the post-processor will modify independently of one another.

Before I begin marking the raw file up, I’ll establish the baseline pictorial structure for local adjustments. This is simply separating the file, conceptually, into multiple discrete pictorial elements that I’ll alter independently of each other. It’s pretty common that as I progress through the post-processing, I’ll refine these to either split an element into multiple segments, or merge one into another. For now though, I can see four elements I’d like to treat separately as a starting point:

  1. Subject
  2. Grain Sacks
  3. Interior distance/background plane(s)
  4. Barn exterior

I’ll begin to literally separate these elements not long after I begin editing, but for now, they will simply serve as a framework for my markup.

Always mark up the linear file to guide your vision

Screenshot of an unretouched linear raw file with and without post-processing markup applied. The markup defines the aesthetic direction the post-processor will apply to the image.

I’ll always start from a linear baseline file for both comparisons sake as I progress through the post-processing, and to guide my (entirely subjective) markup decisions. While marking up black and white photographs, we account for both luminance and surface considerations. The overall balance of the linear baseline is front-heavy; that is, there’s a muddy cluster of light quarter tone details sitting nearest the viewer, with not a lot of clarity between them, followed by a jarring wall of heavier shadow and dark quarter tone details, that feel like they take up less than a quarter of the composition.

What I’d like to do is create a little more harmony between the interior and exterior of the barn, enabling the subject to stand on their own, luma wise. I want to do this while retaining the density heavy interior distance, but have a more gradual transition from farthest interior plane to exterior environmental light. I want the image to invite the viewer not only inside the barn, but around the frame of the composition, to more slowly reveal optical details and create a more weighted sense of three dimensional depth.

The approach for turning the markup into a finished file isn’t all that different from taking a contact print to a fine print in the darkroom; I’ll find a baseline density value to act as a luminance anchor for the entire image, then locally adjust discrete pictorial sections of the image to establish a guide baseline file.

The importance of anchor values for expressive visualisation

Generally, for my initial adjustment I’m looking to set a baseline anchor value, which is similar in principle to a base exposure in the darkroom, however I’m not thinking in terms of dMax here. Digitally it’s a different approach to simply setting the black or white points. An anchor value, for me, is a broader, more global aesthetic quality that the finished file might aim to achieve.

An anchor value is a detective quality; that is, it’s an overall characteristic of the photograph, but not necessarily a trait that is photographed. When we look at a finished photograph of a human subject posing by a window, with a full expressive range of values, we implicitly notice the qualities of the environmental light. Even when we lack the insight, vocabulary or inclination to understand it more thoroughly, it continues to shape the manner in which we read and interpret the depictive elements of the fine print.

In this particular example, I’d like the image to read heavy in density. Not simply dark, but for the values left of 127 to feature more prominently in the finished image, regardless of output intentions, and to sit with clarity in the overall finish, rather than muddied together. This is an entirely subjective decision, divorced completely from any notion of a universal standard that one should adopt absolutely. It’s simply what I like to look at, what I’d like to look as I work through this file, and what I think would work best for this particular photograph. It’s also what I think will best guide me throughout the multiple passes of adjustments to achieve that particular finish, and will help work with me to reach that goal, rather than against me.

Establishing a global baseline anchor value

The first step I take is establishing the foundation for the darkest detail preserving values in the file. At the start of a file, I’ll adopt the position of considering the file in terms of seven distinct luminance targets: blacks, shadows, dark quarter tones, mid tones, light quarter tones, highlights and whites. In the case of a density heavy finish, I’ll try to globally establish the blacks, in addition to all of the shadows and around 30% of the dark quarter tones, with a luminosity ramp graduated roll off starting around 50 or so.

To ensure I don’t throw my upper mid and light quarter interpretation baseline off, I’ll also add a couple of subtle, targeted luma lifts around 140 and 190. Even though I’m focusing on those heavier density values to the relative left of the histogram, if I allow linear global adjustments to cut these lighter values as well, it can shift how I read the values and alter the foundation for further modification. Kind of like looking at a bright yellow screen for 10 minutes before attempting to colour correct a photograph.

Further to this, the modifications here are pretty significant: the global, near linear cut to a large chunk of left of 127 values is showing up without the upper mid/early light quarter values as seen here with them all zeroed out. It takes the surface establishing qualities of the outer left and right frames of the barn and completely washes it out. Again, there’s no universal truth to this process; I don’t need to retain those values (and they’re relatively easy to re-establish) but a) I’d just like to, personally, and b) I need to always keep in mind how one modification sets a foundation for successive modifications. There are literally thousands of possible finished variations of this file, and the difference in both pictorial and aesthetic qualities conveyed between 10 different values can be immense.

Reducing the dynamic range of discrete elements for a better mix

Turning my attention to the left and right outer frames of the barn, I’ve got a slight difference in global luminance between the two. This makes sense; the light source is about 55/45 in favour of camera-right, and the front length edge of the barn skews from the viewer along the same plane. Removing any semblance of optical truth from the equation, I’m focused here on creating a smaller dynamic range between the luminosity values of the left and right outer barn edges. There’s a natural contrast increase in higher frequency edge variation on the wooden surfaces camera-right, plus two or three naturally darker planks, which initially feels a little like throwing off the transparency of the current file. In much the same way a rough, blocked in mask in the early passes of a studio-to-outdoor composite might look, this subtle shift in overall luma values between left and right exterior surfaces draws attention to itself which, in turn, highlights the heavy handed baseline adjustments.

Interestingly, it also adds the first of many initially undetected luminance compositional elements; namely, the structural angle of the light as it cuts diagonally through the frame, shifting from exterior to interior guideline almost perfectly from lower camera-right door frame to upper camera-left door frame. The scattered luminance variation present in the camera-right side planks helps to break up the linear luma gradient inside the barn that progressively rolls off the main subject plane-by-plane to the pure black rear of the interior. There’s a relatively full range of light values on the currently visible right side of the main subject plane – chair, clothing, shirt, hat, cuffs, jacket – which almost perfectly mirrors the variance the outer left barn displays with it’s matching pictorial elements, the lower contrast and significantly less luma-variable fertiliser sacks.

Always favour luminance based lighting modifications

Next I apply a luma-relative lift to the 120-220 values on the subject. Initially this gives me the distinct impression of a poorly composited modification; it’s a well-worn, aesthetically dated approach and looks as if we threw a heavy negative vignette around the edge of the canvas and applied a casually feathered inverted oval mask over the subject. Post-production should be a gradual, step-by-step process. As I’m treating each pictorial element in the file discretely, I have to look past the rough edges of the current merged composite and overall lack of transparency in my edits. I’m simply trying to establish a luma baseline for each independent element within the frame.

There’s not much in the way of complex technique going on at this stage. I’ve applied a relatively simple cut to the baseline shadows, to a) align the darkest surfaces on the subject to the same black levels as it’s surrounding areas, and b) create a stronger dynamic range on the subject itself, to begin to increase the pictorial contrast on the subject’s clothing, amplifying the optical capture. It’s a wide-band shadow and dark quarter tone cut that runs into a broad, targeted boost around 190, with a gradual rolloff either side and a slight parallel cut from 127 down with a luma-relative ramp at both ends. It also includes the first plane behind the subject, as there is a natural gradient transition from the luma values present on the subject to the luma values immediately behind the subject, so the added modification rolls off nicely.

Don’t modify the values on a single pictorial element only

The grain sacks need a little recovery here, for a couple of reasons. Firstly, I’d like to shift the lighter values here to be more in line with the equivalent, neighbouring values on the subject. Secondly, I also want to settle them into the intensity of light we expect to see hitting the front edge of the doorway, in turn making the roughly composited look of the subject against it’s surroundings look a little less cut-and-paste. This in turn helps to align the sacks with the causal qualities of the foreground, creating pictorial definition between the exterior surrounding doorframe. This has been achieved with two second order, narrow bandwidth curve responses at the far left and far right of the histogram.

Don’t accept the raw values as the final expressive truth

We don’t really have a true background here, but we do have a few planes of interior distance to work with. The furthest visible plane from the viewer needs to be recovered now, and established within the black, shadow and dark quarter tone framework I’ve already established.

This is done in a few separate passes, with discrete modifications made to:

  1. the far distance, which is the plane in the image farthest from the viewer with visible detail before the interior drops to pure black;
  2. the upper half of the middle distance interior divider, which is an extension of the original middle background that was boosted with the subject, which needs to be rolled off into a more pleasing transition before it hits pure black; and finally
  3. the entire middle background section, modified as a whole, to create a stepped luminance rolloff from the front of the barn entrance to the rear.

While not entirely essential, it creates a less jarring visual transition from subject focus to pure black distance, ensuring the subject sits more accurately within the currently applied lighting modifications. This helps to ensure the current adjustments convey to the viewer in a more photographically transparent manner.

In the process of making these edits, I also realised I had some crushed blacks in the distance that were sitting around 12-15 or so, creating a less than desirable clash between the established pure blacks of the furthest interior plane and the solid area of (intended) non-detail.

Create added weight, depth and visual rhythm within the frame

In all of the interior background adjustments, I was simply looking to adjust the surface qualities of all pictorial elements until I found replica qualities; that is, depictive details that mirrored the exterior surfaces, to establish a sense of aesthetic harmony between the discrete edits. The interior surfaces were a fairly inconsistent mix of both high and low contrast structural elements, so I wanted to extend the horizontal rhythm of the exterior wooden planks to within the barn itself, while still giving the interior surfaces their own unique identity. Achieving this was simply a case of establishing some heavier blacks, while retaining the gradual luma rolloff from exterior to interior.

This also helps set a depictive rhythm for transitioning into the farthest background plane, with it vertically lower horizontal dividing lines. There is a wonderful pictorial gradation from:

  1. exterior barn planks, horizontal dividing marks, to;
  2. interior middle distance background planks, horizontal dividing marks, ending with;
  3. furthest distance background plane, interior ladder steps.

These horizontal dividing marks continue plane by plane from upper frame right to lower middle frame left, parallel with the compositional rhythm established by the pose of the subject. Even the transition from middle interior background plane to furthest interior background plane repeats the pictorial metre of the subjects crossed leg in a concurrent manner.

Always refine your initial pictorial elements as you progress

The final modification I made in the first pass is the chair, which ended up being one of my favourite depictive elements in the entire image. By this stage however, I’d lost most of it to the shadow it was surrounded by. With no target modification applied, all I’m looking at is the initial base darks adjustment I made to establish the overall luma baseline anchor.

There’s some beautiful detail in the chair, and the slightly imperfect right angles of the chair frame not only help to establish how the subject is grounded in the overall composition, but create a point of interest in the shadow area underneath the subject, breaking up the heavy blacks that, at this stage, make it almost seem that the subject is floating. The entire chair frame is lifted in the light quarter tones and highlights with a fairly wide bandwidth, and a first order mid range contrast boost adds a small amount of structural variance to the frame, rolling off around 160 or so. Finally, the upper horizontal leg is boosted with the same adjustment again and balanced in around 40%, independently of the initial adjustments.

This becomes my initial base composite, establishing the foundation for more refined, targeted local adjustments in the next stage of edits. It’s absolutely too dark here for output. The density-heavy finish does, in it’s own subjective way, enhance the global aesthetic qualities of the composition. It’s not unsuitable or working against the depictive qualities of the scene, but it does tell me I need to push some more refined depictive and detective qualities of the photograph to the front of the finish before it’s ready for export.

Now rest your eyes, you need them fresh for the second pass

Now that the first pass of processing is complete, I need to turn my attention to a more intricate value based approach to refine my initial adjustments before the file is ready for output.

In the video above, you can compare the before and after transitions of the final, complete file on the left, and the current in progress merged composite on the right. While I’ve managed to transform the image significantly, creating a pictorial depth and aesthetic mood that wasn’t as present in the linear raw file, you can see that I’ve skewed a little too dark in the image on the right. And by ‘a little’, I mean ‘a lot’. So how does the Second Pass of editing help me achieve this more compressed, luma-accurate finish?

In ‣ the next part of this series, I detail my approach to achieving these edits in the most photographically transparent manner, and discuss more insights into why I need to implement them.

Image Attribution

Johnston, Frances Benjamin, photographer. Carver’s Old Mill, Gordonsville vic., Albemarle County, Virginia. 1935. Image. Retrieved from the Library of Congress, https://www.loc.gov/item/csas200904177/.