GR#034 - Death Korps - Portrait

[OriginalAdric]

I've got a bit of a backburner project I had some time to work on tonight, and I figured I could use some C+C to get a feel for how I'm going. I have what is probably considered an unorthodox workflow for pixel art, but I've got a background in animation and CG that leads me to favor things like shadow masks and multilayer compositing. I've split the process into multiple images to help clarify how I think about my work.



L1 is a down-rez'd scan of a sketch from my sketchbook. L2 is a cleanup done using a standard round brush tool and a Wacom tablet. In L3, I've broken the forms down into planes to use as reference for working out the shading. L4 is a WIP snapshot of doing a full clean on the lineart.



C1 is a continuation of the work from L3, working out the shading of the faces relative to their angle from the light source, irrespective of any shadowing or ambient occlusion. C2 is a lighting mask showing cast shadows, ambient occlusion, and ambient/reflected light. C3 is a flat color layer to block in the basic forms for painting. The colors in C3 are actually significantly darker and duller-looking than this image indicates because the layer uses the "Hard Light" blend mode.



This shows how the layers composite to create a flat-shaded image. The C2 light mask is set as a "Multiply" layer over top the C3 color map. (In this instance, C3 could be converted to a straight color layer without the Hard Light blend mode, but the reason for retaining that setting will be apparent in the full shaded image.) Finally, the L4 line art layer is multiplied over top of the color layers to finish the look. In the final artwork, there will be another color layer added over the line art as a clipping mask so that I can color the line art w/o having to go back with a single pixel brush.



This is the same image as the flat-shaded art, except that I have added the C1 shade map underneath the C2 color map. This adds an extra level of detail to the overall lighting effect of the image.

I realize that this setup is overkill for basic pixel art, but my reasoning is that I'm still trying to work out how I want the elements of the image to look, so by modularizing the different aspects of the image, I can alter some parts without having to re-do whole sections. Additionally, if I want to start animating stuff, this makes it very easy to flesh out the overall motion of a piece, then add in the light and shade without risking the original art.

The C+C I'm looking for mainly is about the aesthetics, but I'd love to hear what you think about *how* I'm working as well.

[EvilEye]

Why so much work

Most people just do lines ( or no lines ) then basic colors then shadows + highlights followed by AA ( anti-aliasing ).

Also your method of blending several layers together is not considered traditional pixel art. It's unnecessary and leaves dirty looking stray shades of color. Pixel art is clean and precise.

You should take C1 and replace the greys with your chosen colors and work from there.

[OriginalAdric]

Thanks for the feedback.

I had hoped that I'd made myself clear as to why I (unconsciously) chose this way of working. Breaking the image into render layers, so to speak, gives a flexibility in editing that you can't get otherwise. I have to admit, the whole one-paint-layer business freaks me out. I can see why it works for doing single images, but for animation (which is the mindset with which i approach my work) it seems suicidal in comparison to breaking elements into layers. The actual movement part of animation doesn't take all that long in comparison to the time it takes to clean up and paint the image. To me, any means of making that easier and faster is going to more than make up for the initial setup cost.

The real "problem" I'm facing, though, is the color. I think the full shaded version is too muddy, but I'm relatively happy with the direction in which the flat-shade is going. However, color is a weak spot for me, and picking palette colors is difficult. Any advice or tips on punching up the color would be appreciated.

[PixelPiledriver]

I can dig it. Layers are totally useful. Changes happen all the time. Especially when on the job and layering each element will save you tons of time without destroying stuff. What you refer to as 'alot of work' is merely a form of organization. When you practice any method it becomes natural. Honestly this will save your ass over and over if you are being directed by a client or boss.

I can agree that in some forms 1 layer is neccessary or prefered. But most of the artists I know whine ALOT when they have a 1 layer painting and are asked to change this and that. Cause its hella work and hella messy to change. I often find that artists don't want to use layering methods because they have really bad hotkey setups and they find it annoying to click on stuff. Try setting it up like a 1st-person shooter.Use the WASD QE ZC X RFV keys and then all the combinations of ctrl alt and shift with those keys. keep what you use most on the left side of the keyboard and what you use least on the right side. Example: in photoshop 'ctrl-Q' by default is Quit. This is completely useless on the left side of the keyboard (and at all as a hotkey). I put Make New Layer on ctrl-Q. It is very easy to push and without even thinking about it I type in a name (sometimes just 1,2,3,4,5) and hit enter and continue drawing. Theres a ton of other hotkeys that can be removed and replaced with much more useful actions. I suggest everyone read thru the list of every program you use and make a very nice compact custom set.

But no offense to anyone. Smiles all around.    In honor of good old fashioned education OriginalAdric should try Evileye's process. And Evileye should try OriginalAdric's process. If you don't see why something is useful at least give it a couple of gos.

So I played around with your art a little.



I typically use Overlay layer mode to do shadows and lighting. It mixes colors well with others without making them too muddy and white and black up and down the luminance in a (usually) pleasing way. When ever I feel like my photoshop art is too mid range I throw on a couple of Overlays and it gives me a lot more key to work with.

Your drawing could probably stand 1 or 2 more levels of highlight. And maybe a little bit stronger bounce light. The final mix almost completely erases it. I drew in a little of both but you could do it cleaner and more precise to your liking. Multi color shadows never hurt anyone, but if its not your thing its cool. You can base the multi coloring off of reflected light, opposite light and shadow colors, or just colors that 'do something' together. Depending on what I feel like I use 2-4 colors in a shadow. Comparison is always the best way to go. just shove all the layers into a group, clone it to the right and then change the colors and see what happens. Clipping masks and pixel lock are the best ways to do this like you stated in your post.

I also really like the original drawing so I was curious to see if that would add anything to the composite. The bright version uses the sketch with Overlay and the dark version with Multiply. I also used Adjust-Hue/Saturation and checked 'Colorize' to play with the sketches line color. The black pixel lines might like some color as well even if it is very close to 0,0,0. I didn't try that in my edit. oops.

Your art is educated. You can do much better than my quick edit. Just try some stuff and go for it.

[OriginalAdric]

Man, that looks a whole lot better with your edits. I always get blank-canvas syndrome when choosing colors, your paintover's given me a pretty good idea of how to get started on it now. I like the feel what adding the sketch on top does, even if it is starting to get away from that clean, controlled pixel art look.

I'll have to go thru the keyboards hotlist and take a look at what things I do most often. I like your suggestion of using the standard gaming controls for the main hotkeys. For people with tablets, putting your main controls on the tablet itself is a huge timesaver. I've got my Intuos' ExpressKeys set to Brush Size Up/Down, and History Step Fwd/Back; the pen's rocker switch is set to Alt for the forward switch (eyedropper with the brush tool), and the Hand/Pan tool with the back switch

I'll take a whack at cleaning this thing up a bit more and trying some new color combos, then post the results when I get the chance (today, I hope). Thanks for the advice.

[EvilEye]

Both of your are misunderstanding what I am saying. I didn't say you couldn't work in multiple layers. Most of us ( myself included ) work in multiple layers. I am saying the act of blending several layers together at less then 100% opacity ( or with some kind of modifier ) is not considered pixel art on this forum. It's considered regular digital art, which is much less precise and controlled ( and usually larger in resolution ). If your going to do regular digital art, why not just paint everything at 10x the size then shrink it down? It's much quicker.

Also your layer scheme is not going to make animation any faster. When people use multiple layers for animation they are splitting up the moving parts over several layers. Like if you wanted the head to nod, you would have the head on a different layer then the shoulders. But you are just splitting up the shadows and lines on different layers, unless you want to animate him changing colors or his outlining there is really no point in doing that.

[Helm]

Again, instead of focusing on 'is this pixel art or not?' I think you should be discussing whether the methods on show on this thread are helping the art. The only reason I'd suggest you do more pixel-purist work is if it would arrive at a better result. And I think, in this case, that simplifying your methods and doing some of that preliminary work *in your head* would shave off time and save you effort and arrive at a better result simply by virtue of methodological simplicity. It's great that you're exploring this 'rendering pass' method, there's much to learn about construction like this, but if you want my advice, once you've internalized the method, just go from the volumetric grayscale (which should be merged with what you tagged 'L2' very easily) to choosing colors and doing the rendering work pixel by pixel. It's much faster and the result will look more cohesive.

edit: here's a partial edit working from your flat, which I color-cleaned and used as a base



It's very incomplete, but do look at it in your pixel program and see the sort of choices it's indicative from and consider: are you equipping yourself best to make these choices by using disparate layers and 'rendering passes'?

[OriginalAdric]

Also your layer scheme is not going to make animation any faster. When people use multiple layers for animation they are splitting up the moving parts over several layers. Like if you wanted the head to nod, you would have the head on a different layer then the shoulders. But you are just splitting up the shadows and lines on different layers, unless you want to animate him changing colors or his outlining there is really no point in doing that.

Sure, splitting out absolutely every element to its own layer would make animation slower, not faster, but a happy medium would make things more efficient overall. Painting shadows is one thing, but animating them is a different beast. If you look at traditional (ie. cel-type) animation, shadows are almost always done with a mask layer because of how much of a pain it is to try to go back and fix the animation on a flattened, composited image. Personally, I wouldn't split the different moving parts of a single character onto multiple layers b/c it's faster for me to block things in traditionally. Heck, if I were doing some serious animation, I'd do it all in pencil on paper so I could roll the drawings and nail down the movement, then scan, downrez, and clean up, because it gives me a greater degree of precision and control of the motion before the pixel stage, regardless of the "extra" work.

I agree with Helm's take of not focusing on whether or not the process is helping the image instead of if it's pixel art. Part of my goal is to be able to internalize the process, but for my own understanding, I want to break it into its constituent parts to see how they interact with each other. One of the reasons I decided to start doing pixel art is because it gives me a chance to develop my painting skills in a smaller, more controlled environment where I can more closely examine how light and colors interact.

I find the suggestion that what I'm doing isn't considered pixel art because of my use of layer techniques somewhat shortsighted. I was under the impression that pixel art was defined by the conscious and precise planning of where and how the image is put together, pixel-by-pixel; that's exactly what I've done here. I don't see why pixel art and so called "digital art" (which I think pixel art could be considered a sub-set of) cannot inform and enrich each other. I think the adherence to purely traditional techniques certainly has its place, but to limit the definition of pixel art to those techniques has the potential to cause the form to stagnate by discouraging people from experimenting with new or different methodologies.

[Helm]

I want to break it into its constituent parts to see how they interact with each other.

That's commendable, but keep in mind one of the strengths of pixel art is that you can work on such an opaque and easy-to-edit medium very fast and do things on a whim. I wouldn't paint shadows on the frame if I were doing hand-drawn animation, but I'd certainly do it in pixel art, where fixing a frame's lighting takes 2 minutes of work. Part of discovering the strengths of this medium lies in diving in. The analytical approach wields great results when it's tempered with that impulsive energy I think.

One of the reasons I decided to start doing pixel art is because it gives me a chance to develop my painting skills in a smaller, more controlled environment where I can more closely examine how light and colors interact.

That's great. However, the art you've posted here has little to no attention paid to the shapes of the pixel groups (or clusters). I realize you're ways off from the finish render, but keep in mind that until you sit down and make pixel clusters look as beautiful as possible and make them work together as well as you can, you're effectively just working with light and colors in a small resolution. Which is fine, but I'm eager to see what'll happen when the pixels enter the picture, so to speak.

Let's leave the discussion on whether this or that view on pixel art purism is short-sighted until after you've submerged yourself in pixels for some time, yes?

[OriginalAdric]

Thanks for the awesome paintover, Helm! After reading over the comments, I decided to take this back to square one. Using the paintover as reference, I roughed out a quick structure exploration, then began to block in my tones. I'm making this a greyscale study so that I can get a better handle on how to use tone to create the appearance of different types of materials without having to worry about color interactions on top of that.



I'm a long way from done, but I figured I'd post my WIP to get some C+C on the new direction I'm heading. I think I may just continue doing greyscale works for a while until I'm confident enough with my tonal control to move on to color.

[Helm]

Right off the bat, you have really dark shades, and very bright shades, but you don't have enough middle shades. Color identity is in the middle values, how do you solve this?

[OriginalAdric]

Right off the bat, you have really dark shades, and very bright shades, but you don't have enough middle shades. Color identity is in the middle values, how do you solve this?

A weird problem I've run into is that the image looks one way in Photoshop, but when I save it out to PNG, the colors all seem to darken up, so the contrast in the dark areas is significantly less than when I'm working on it.

I'm not sure I understand the question. How do I solve color identity, or how do I solve the lack of middle shades?

[blumunkee]

PS can biff png gamma.

http://morris-photographics.com/photoshop/articles/png-gamma.html

The grayscale is good, but watch out for gradiants and banding. Helm's paint over uses color and texture to keep things from looking too cylindrical.

[Helm]

You solve the issue by introducing more middle-shades and evening out your lightness spectrum on the shades you already have! The middle of the range is where you need to work most, not the extremes. Leave the top of the range for small highlights and the bottom of the range for dark shadows. Right now, whereas you've got the right amount of dark in your picture, when you need to bridge the gaps. You have 33 colors in your image, I am not sure how, I see about 11 handpicked in that little palette you have on the top-left. Also they're not really grey, they have small saturation and hue values.

Crash course on pixel art:

1.You'll never be in control of your palette until you can select every index and change its HSL values in real time in a real pixel art program. If you're using photoshop, you'll never control your palette like a person that uses graphics gale or pro motion or any other program where you can edit every index entry in real time on the fly. Even if there are workarounds, they'll never be as fast as a real pixel art program does it, and what is not fast, whatever you have to do extra labour to achieve, your artistic reflex will be to do less. And pixel art is control, you need to control more, not less.

2. Small palettes are manageable, this is why pixel artists use small palettes, not to look 'retro' (well, not all of them). Even the palette in my edit is too big, your image can be done with 16 colors or even 8 colors if someone wants to accentuate the planes and doesn't care about AA so much. If you don't know how many colors you've got in a picture at any given time, you won't be able to control the palette.

3. An optimized palette is one where you can convey a lot of different hues and gradations using interchangeable 'linking' colors on a small linear index. That stuff can only be done right in a deluxe paint-like program. You won't be able to map it out on the little right corner of the screen.

4. conveying small shapes that flow into each other well and avoid banding and mudded up volumes is what pixel art is all about. If you have very large flats and too many colors to convey them with, take away colors! Reach a balance between what you are drawing and how much you need to use to draw it.

Are you starting to see why pixel artists work like pixel artists and not like photoshop-in-small-resolution-artists? There is wisdom in taking the aesthetic form to its logical workflow conclusion. Pixel art isn't 'pure' because some guy on the internet made up some rules and marketed them, art made of pixels looked best when done in a certain methodology set and that's what informed people's opinion of what a pure version of the nascent medium would look like.

Here's an edit



And yes, try to avoid just doing a gradient that follows the shape of the mask from point a to point b. Instead think of the thing as if it's a low-poly flatshade 3d mesh at first. Just use straight lines and cut through the complexity of the real thing.

[Demon11777]

I really like where this is going but the left eye was bugin me a bit. I think it should be more oval like

Uploaded with ImageShack.us

[OriginalAdric]

@blumunkee

I think I figured out what was wrong with the color darkening. It wasn't the PNG issue, it's actually a Photoshop color profile problem. Basically, the target sRGB colorspace was not the colorspace PS was using to display. Setting the View/Proof Setup options fixed that.

@Helm

Thanks for the thorough writeup, as well as the second paintover. I appreciate the time you're putting in on helping me out. I was expecting simply for a clarification on the question with the thought that I'd have a chance to take a whack at an answer before being told, but this works well, too. I find it odd that you're seeing hue/sat values for the greys. Even when importing the version I flattened/uploaded, my color picker is showing H/S as zeroed out. Oddly enough, your paintover *does* have some low H/S values in its color table.

A few questions on the notes you had:

1) As I deconstructed the palette you used on the color paintover, I saw that there were a number of colors reused in different areas which appeared to have different colors. Is that what you mean by "linking colors"?

2) In both paintovers, you have some very precise shadowing areas with multiple color/tone transitions (the greyscale collar is a good example). Are those being done with individual strokes lined up manually, or are you using some sort of transparency or other brush techniques to do broad shade areas w/o having to line up individual strokes?

I hate admitting I'm wrong, especially since I'm generally used to being the go-to guy for production workflow solutions, but I see why this way of working is better for doing pixel art. Still not totally sold on using it for animation, but I'll withhold judgment til I give it a go.

[Helm]

Don't worry I haven't told you anything because being told means nothing. If you put the effort into this medium, like any other, it will teach you all you need to know first-hand. I'm just hoping words act as a vague catalyst to build interest in doing the work yourself. If I had even the smallest fear that instruction can rob someone of the journey of self-discovery, I'd shut up forever. I've read some philosophy and if it worked like that I'd have made right choices in my life from 16 on up and instead I've made so many mistakes I can feel Plato and Nietzsche together face-palming when they watch me from within. Whatever I know I know, it had nothing to do with being told.

The hue/sat values might be from when I imported the image into pro motion, I might have mistakenly checked 'use current palette' instead of 'adaptive palette'. Anyway, it doesn't matter. Even if something is wrong with your photoshop that's tinting your grayscales, I hope eventually you'll stop using photoshop for pixel art anyway, heh. Solve a maze by not walking in it.

1) Yes, exactly. As Tsugumo used to say [I paraphrase] 'look, I used the same dark brown from the skins of the shadow as the main hair color, like Capcom would!'. We may have reverse-engineered a few solutions from video-games, but if you look at real painting, many of the masters had small and controlled & reusable selections of color too. It makes sense if you're aiming for control.

2) There is a draw mode in pro motion called 'Brighten/Darken' to which you feed a value x from 0 to 255. When you click down with the mouse button, it checks the lightness value of the pixel(s) below it and replaces them with +/- (left click/right click) x by approximation. If you are working with a 4 color palette, this is pretty useless because you can do it by hand anyway, but if you're trying to mock up something fast with 22 colors, then it's very useful for selecting a stencil of a whole area (as I did here) and pressing the right mouse button down twice to make the whole selection move down two index slots in the linear palette.

Pro Motion has a lot of other very useful things for pixel art in specific. You should check out a demo or something. If you're serious about pixel art, it's the go-to tool.

I'm looking forward to seeing you grow as a pixel artist

[OriginalAdric]

Thanks for all the advice, Helm. I've downloaded the demo for ProMotion so I can see whether I want to buy it, tho' for $80, it's a steal (in terms of graphics software prices). Getting used to its mindset and workflow means I'm working slower than I'd like, but putting down some custom hotkeys is helping to alleviate that.

I took this back down to scratch again to get the feel of doing a piece in ProMotion from beginning to end. I've only blocked in the basic tone shapes, but I think I'm getting the hang of it.



I've found that taking this same piece and re-doing two or three different times has given me a good way to focus on understanding the process, since I'm not having to work out so many details of the structure and form as I go.

[OriginalAdric]

I've started to notice how all of my work is pretty plain b/c it lacks any sort of significant story-telling details, so I decided to add some damage/claw marks onto the helmet and face mask while refining the overall lighting a little bit more.

[Cure]

This is coming along nicely

One thing I think could be worked on that helm covered in his edit-
Yours seems like it's flooded with light, over all very bright so it doesn't fit the mood of the battle-weary, post-apocalyptic soldier. it also creates huge lit plains that compete for focus, you'll notice in helm's image that there are a few key highlighted areas that make the image pop where it needs to- drawing the focus up to the face. Introducing some darker greys should help fix this.

[kriss]

The last one is very good, i'm waiting for the color edit ^^

[OriginalAdric]

Taking a mental break from the Flame Trooper portrait by working on a portrait of a Death Korps officer. Still figuring out tonal control in the block-in stage. C+C more than welcome.



EDIT: Hurr, forgot to upload the last iteration of the Flame Trooper.

[kriss]

about flame trooper, honestly i prefer

In the last one, there is too much dark, but I think it will be better with face in metal color and other with green/maroon color

Death corps seams less details than flame trooper and it's lack of contrast : I think it's lack of very light color

[OriginalAdric]

Did a bit of a palette edit, adding some saturation to the darker tones and bringing the overall tonal range closer together.