26

N.B: I previously used the term "pixel art", which I took to mean "modern, intentionally low-resolution art," though that's an inaccurate definition, so I edit the title

It seems to me that modern low-res art looks better than retro low-res art, in some intangible/hard-to-pin down way.

For example, take a look at this screenshot of Stardew Valley:

Stardew Valley random screenshot

It looks much better than any retro game I can remember, but in a way I can't quite quantify. What is it that makes it look "so much better"? I have some ideas of what might be responsible:

  1. Higher colour depth
  2. Higher resolution (even if it's the same low DPI, it's just larger overall)
  3. Anti-aliasing

Is it these, or something else?

  • 8
    Perhaps the differences you are remembering are down to the capabilities of retro-hardware - i.e lower resolution displays such as CRT monitors/televisions. – Billy Kerr Apr 24 at 23:54
  • 16
    Also may video game systems had limits on the number of possible colours that could be displayed simultaneously. This Wikipedia entry may be of interest to you – Billy Kerr Apr 25 at 0:03
  • 8
    OK, I understand, but I think this is an important aspect of your question. A lot of modern games might appear to be low-res but in fact use the full resolution of the screen by moving objects smoothly across the pixel grid and might even rotate pixels. I believe that "style" is called "Big pixel" and is imo not full-blooded pixelart but actually a kind of vector art where each pixelated sprite is treated as vector objects. Do I make sense? – Wolff Apr 25 at 0:08
  • 3
    I would say it's experience and tools; Forget computers, look at art and paintings in general: the realism, colors, depth, etc have massively increased through the ages. When you were doing pixel art back in the days, unless you had a tablet, which was rare, you'd draw stuff pixel by pixel with the joystick. An accomplished artist today, with proper tools, will get a much better picture on a C64 or an Atari800 than we were able to do back then. – Thomas Apr 25 at 22:07
  • 5
    Since the question doesn't show a sample of retro low-res art for comparison, I'm just going to post a reference for the (innovation of) retro graphic: Gamasutra - As SNES turns 25, devs discuss its 7 greatest graphics innovations – Andrew T. Apr 26 at 7:44
2

I think it is shadows that make a huge difference. If you zoom on the fence, you can see what looks like a grey border that is bridging the gap between the posts and the ground. Other shadows are more obvious. Also:

  • lighting the well and trees are illuminated in appealing ways
  • variation because modern games can have storage space to hold multiple variations of an object or tile, it looks more realistic. So rounded edges, wavy edges, and differing phyllotaxis make a realistic scene. legend of zelda low-res Super Mario Bros., posted in another answer Compare to these pictures, which are obviously tiled and have limited color. You can especially notice that while the tiles seem slightly highlighted, the character is faded compared to the surroundings.
| improve this answer | |
  • 2
    I don't disagree with what you say, but doesn't both anti-aliasing and lighting just come from having more colors to choose from? In pixel art those two features are often added manually, not applied as an automated effect. – Wolff Apr 27 at 22:41
  • Yes, that's definitely true. But in games, which are not manually edited every frame, the individual tiles are still lighted beforehand – Wezl Apr 28 at 13:38
  • 1
    side note: Anti-aliasing works better with more available color gradations and pixels to make transitions. The image presented is twice the size of the original NES resolution. and (without tricks like rewriting the palette mid-screen draw) one could only use about 25 of the NES's 60-something available colors on the screen at one time. – Yorik Apr 28 at 19:33
  • 1
    Uh... there's a lot going on with this answer that's not really quite right. There's no anti-aliasing in Stardew Valley. Pixel art games generally don't have it - each sprite is simply a collection of pixels to be drawn. And the reason there's 'blurring' when you zoom in to the OP's photo is because it's a JPEG. JPEGs don't have crisp edges, because they're optimized for photos. If you don't believe me, fire up the game, take a screenshot, and zoom with Paint/Photoshot/whatever. You'll see that each sprite has a firm area of effect. – Kevin Apr 28 at 23:15
  • @Kevin Thanks for saying that in a nice way. I didn't do much research and so I probably got Anti-aliasing mixed up with something else, like shadows. You can definitely see shadows. But I never said 'blurring' because that's not what I meant. – Wezl Apr 28 at 23:26
26

(I am answering to the body of your question, please split this question into 2 things)

Screens are way better, lets look at things:

  1. You can choose colors as you wish (even if you restrict palette). Your colors aren't smudged across 3 pixels* (Like Amiga) and your palette can actually make sense.

    Old games were designed with shoestring memory budgets** and with terrible color palette limitations. Even if you were to open it in a emulator it would still have this limitation as it is built into the assets themselves.

  2. Authoring software is better, and there is more authors. Back in the day there wasn't all that many people doing this. There is still a lot of mediocre pixel art in the world yet you select for the better ones. So combination of less restrictions and some sort of selection bias
  3. Your screen is better, like way better suited to this task. The images in old CRT's were allays more or less smudged due to the method they used to make the image. Modern monitors are way sharper and the expectation of having that quality is given.

    On the other hand the graphics were designed with this in mind. As that was what the artist was seeing and adjusting for. So a small amount of smudge may make the images more authentic and possibly better looking.

  4. Game engines have fer less restrictions. So the artist can concentrate on art not optimizing things.
  5. We know better how to make pixel art. Don't underestimate experience of the community.

* I really hated this feature when making graphics. Loved the feature when consuming computer art.

** Consider that a simple modern webpage can have more data even if you exclude images. So just the CSS, javascipt and html would have broken the memory budget artist had in late 1980's.

| improve this answer | |
  • 1
    #1 and #3 aren't that relevant, as old games can be run in modern emulators on modern screens, and many still don't have the quality of the best current pixel arts – vsz Apr 25 at 20:29
  • 5
    @Vzr 1 still holds game graphics palettes are still made with the limitation of the time. The OP himself concedes to 3 in the comments – joojaa Apr 25 at 21:33
  • 2
    Your 3. is incomplete; you're missing that CRTs displayed graphics differently, so modern renditions of retro pixel art will look different to how they were intended; some, especially many of the more famous ones, were designed to exploit that smudging to get better colours, pseudo-higher resolution etc, and display much worse on a modern LCD display – especially blown up to show the jagged edges. – wizzwizz4 Apr 25 at 21:54
  • @wizzwizz4 good point ill add it when i have a computer around – joojaa Apr 25 at 22:31
  • 3
    Some art actually looks worse in emulators because it was designed for blurry pixels. – user253751 Apr 26 at 11:07
21

I suspect that you are seeing realism by chaos. If you take a good look at, for instance, the radishes in the lower right, you will notice that they differ. In this case, by mirror imaging, and position variation. The two rows of beans sprouts on the left also have two distinct images to start with (but didn't mirror image one of them). One needed trick in this is to ensure (where appropriate) that the same random variation happens with every rendering.

I suspect that the older games would have restricted themselves to one image without variations, that would make them look somewhat artificial. This would presumably be because of lack of storage and compute power to do the mutations. (And maybe because nobody thought of it.)

Edit: There is also some layering going on here. The poles with the bean sprouts are not part of the garden bed or the cell behind it, but occludes both.

| improve this answer | |
  • 1
    I didn't even realize, but yeah, you're totally right! – Alexander - Reinstate Monica Apr 25 at 20:08
  • Related to your edit - layering takes more RAM, but also fancy bounding boxes do; whether for clicking, collision detection or whatever they now match the shape of the sprite much more closely. This (and more physically realistic paths) make motion graphics look better - and more is moving independently at a time now. – Chris H Apr 28 at 10:28
7

Edit: This answers only to the original question: What makes pixel art look better than simply low resolution images? Later edits of the question have changed the subject.

Pixel art objects are created to be presented exactly with zero error on low resolution screens. Convert another image to the same low resolution and so much details can be lost that nobody can guess what was the original image. It can be originally very well done but that's lost. Low resolution hides nothing of an pixel art image, the result contains every detail and relation the artist was able to create.

If I have a low pixel dimension canvas in a program and I paint with a brush the result is crap, no matter would it be crap or not in higher resolution. That's because I'm not able to invent nor put down shapes nor relations that look pleasing in low resolution media. A pixel artist is less limited in that sense and that does not depend on do we talk about pixel artists of today or 40 years ago.

| improve this answer | |
4

The screen shot would reproduce nicely in 256 palletised graphics of many vintage platforms. But (as has been said) there are technical reasons why vintage low-res games don’t look as good as modern ones. I’d agree it’s mainly memory and remembering that the palette was probably shared throughout the game. Plus colour cycling animation techniques and text may have meant the artist had additional limits on the palette they had to work with.

There are also many practical reasons why vintage low-res didn’t achieve the standard of modern low-res.

Higher lowest-spec targets

Not only did old machines have different graphic and memory constraints but they were more diverse. Sometimes assets would have been quickly/crudely reworked to fit a less popular/capable platform.

Ease of workflows

Meaning artists are more efficient and can pay more attention to details.

  • Loading times from 3.5in floppy drives vs modern HD. A problem developing your assets but also affecting the run-time of the game.
  • Modern big monitors allow many assets to be edited/reviewed side-by-side.
  • Distributed version control means assets can be shared and bounced back and forth between many artists more effectively.

More sources

  • Scanners, digital cameras, Internet images all provide ample sources for textures and reference images already in digital form for reworking/tracing.
  • 3D models which can be rotated and rendered to 2D with consistent lighting and texturing is a massive time saver even if you hand-retouch each image.

It’s incremental

(standing on the shoulders of big pixels)

  • Artists these days probably aren’t working from a clean slate - so assets are always being refined.
  • Pixel art is here to stay, so studios may see more incentive to invest in and reusing good low-res assets.
| improve this answer | |
4

Technology's one aspect, but I'd like to focus on what I think is an even bigger one: experience.

The original games you're comparing this to? What experience did its designers have in creating computer graphics gaming - at all? Pretty much everything they were doing was groundbreaking - they didn't have any existing knowledge set to go off on it.

But as time went on, people starting learning how to do things well. And as an example, let me present two games: Super Mario Bros and Super Mario Bros 3.

Super Mario 1 pic Super Mario 3 pic

These are two games for the exact same hardware platform. Yet Mario 3 had much more pleasant graphics. Why? Because it came out 3 years later, and was able to capitalize on the experience learned in creating video game graphics.

It's the same for pretty much any creative enterprise. The people doing the groundbreaking - while important - are never going to create as good of works as the people that build on those groundbreakers' knowledge.

Stardew Valley's creator did an amazing job. But they weren't starting from a vacuum.

| improve this answer | |
3

Let's assume that you're comparing your image (24-bit color) to an earlier game image done in 8-bit color. If you're working with 8-bit color (max of 256 colors) to start with, you could, in theory, come up with an image very close to the one you've posted, but it'd take an incredible amount of work to fine-tune all of the edges to reduce the appearance of jaggies.

If you have a system that allows for 24-bit color (ie, every computer out there today and for many years past), you could create the image at higher rez then color-reduce to 8-bit, letting the software do all the hard work of antialiasing and so on. As a test, I converted your image to 8-bit color using a free program called IrfanView. The conversion was all but instant and the results practically indistinguishable from your original.

| improve this answer | |
3

As Billy Kerr pointed out in a comment, a reason sprites of the same size and resolution look better in modern pixel art than in classic pixel art is that a modern desktop computer generally has at least a 16-bit colour rendering ability. Roughly speaking, according to the capabilities of the hardware, any pixel on the screen can be turned to any of these 216 colours at any time. This means that artists have much more freedom of colour selection.

On older hardware, the hardware's rendering ability was much more limited. Depending on the mode of operation, a limited subset of colours (known as a palette) could be used to render the screen. For example, the SNES had a 15-bit master colour palette, and based on that, you might think that the SNES is capable of outputting graphics of a similar colour depth to modern desktop computers, but not all of them could be used at once:

The Picture Processing Unit (PPU) used in the Super NES has a 15-bit RGB (32,768 color) palette, with up to 256 simultaneous colors at once.

However, while the hardware palette can only contain 256 entries, in most display modes the graphics are arranged into between 2 and 4 layers, and these layers can be combined using additive or subtractive color blending. Because these blended colors are calculated by the hardware itself, and do not have to be represented by any of the existing palette entries, the actual number of visible colors onscreen at any one time can be much higher.

The exact number depends on the number of layers, and the combination of colors used by these layers, as well as what blending mode and graphical effects are in use. In theory it can show the entire 32,768 colors, but in practice this is rarely the case for reasons such as memory use. Most games use 256-color mode, with 15-color palettes assigned to 8x8 pixel areas of the background.

If you want to know more, I recommend Retro Game Mechanics Explained video series on SNES graphics.

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.