SYSTEM NOTICE

Auto translation by AI. Be sure, accuracy, nuances and authorial intent may not be fully reflected.
見出し画像

Why I Recommend Horizontal Dithering for Intermediate Color Representation in MSX1-Style CG Illustrations

Points to Note

This post is a discussion based on the assumption of drawing MSX1 rules for large, single-frame CG illustrations. This introduces logic that does not apply to pixel art intended for game production, such as sprites or map chips (background patterns), as game pixel art has entirely different methodologies. Since it ultimately comes down to personal preference, this article is not intended to suggest that other methods are inferior; please be aware of this.


I use horizontal stripes as my main method

In my own MSX1-style CG works (hand-drawn), I often use horizontal stripe intermediate color representation rather than checkerboard patterns.

Left is checkerboard pattern, right is horizontal stripe pattern

This is intentional, as adopting horizontal stripes offers many advantages when working by hand. (The converter I am currently developing also prioritizes reproducing intermediate colors using horizontal stripes.) I think this is a fairly interesting observation, so I am posting it.

Hereinafter, I will refer to checkerboard dithering simply as tiling, and horizontal stripe dithering as horizontal line painting.

Why horizontal line painting is preferable for MSX1-style CG

It is harder to violate the rule of up to 2 colors within 8 pixels

Naturally, this is basically the reason. (I will introduce the advantages derived from this later.) To explain briefly, MSX1 (and game consoles or computers equipped with similar video chips) has a strict restriction regarding screen display of up to 2 colors within 8 horizontal pixels, and in the case of tiling, 2 colors are already consumed within those 8 horizontal pixels. In the case of horizontal line painting, colors are separated vertically, so only 1 color is used within 8 horizontal pixels, which significantly reduces the frequency of violating the restriction even as you add more detail.

You can add a third color!

Let's consider an example of drawing a circle in a different color over a tiled area.

It looks like the following, but there are quite a few parts that violate the 2-colors-per-8-horizontal-pixels rule.

I want to paint it like this, but

The parts highlighted in green violate the rule.

Violations over a wide area

To cover this up and fix it, you have to move the shape of the circle to the 8-pixel boundary to deform it, or deal with it by having the background or the color inside the circle partially glitch.

Example where the outside is color-glitched

There are various ways to fake it, but phenomena where colors bleed out and glitch were common in early MSX CG, weren't they?

Apologies for the bad example: Wingman 1985 © ENIX

Next, let's try the same thing using horizontal line shading.

Perfect representation (the deformation of the circle was applied by mistake, but it is not necessary)

In this state, there areno violations at allof the 2-colors-per-8-pixels rule! There is no need to deform the shape of the circle, and it is extremely convenient when adding new objects.

The boundary edges do not look jagged

This is the example of the circle within the initial tiling, but it is also common to borrow colors from the tiling composition to substitute for the circle's fill color, as shown below. Although the impression changes, no rule violations occur.

It no longer looks like a circle

However, the outlines tend to look jagged. This phenomenon is also often seen in places where shadow colors are painted with intermediate tiling colors. If applied to skin shadows, it can make the skin look rough, so caution is required.

Something imitating skin highlights or shadows

This phenomenon occurs on the Famicom, PC Engine, PC-88, and PC-98 as long as it is pixel art, but it is not much of a problem on machines with higher resolutions because it is less noticeable.

If you do the same thing with horizontal line shading, it looks like this:

The edge shape is relatively clean

You can see that although the shape is slightly distorted, it is not that noticeable.

Intermediate colors can be adjacent to each other

I wrote at the beginning that a third color can be added with horizontal line shading, but even if you place horizontal line shading in the newly added painted area, it does not violate the 2-colors-per-8-pixels rule. With tiling, it is difficult to place intermediate colors adjacent to each other.

In the MSX1 palette, which has few beautiful colors, this is an important point.

Color schemes can be changed later

If you proceed with a drawing by making heavy use of tiling, you may end up with a complex arrangement of surrounding colors, making it difficult to easily change the color tone. This happens when the tiling composition colors are also used in the surrounding areas. You end up in a state where you cannot move freely.

However, with horizontal line shading, you can always replace(*A)the intermediate color fill with a different intermediate color. In the example of the shadow and highlight representation mentioned earlier,

The image above shows the tiling composition colors replaced, and many rule violations can be confirmed, but

this one, where the horizontal line shading composition colors were replaced, has no error locations. If you use horizontal line shading, no problems will occur even if you replace them with completely different colors during the final overall adjustment of the work. This is powerful.

*A This excludes cases where intermediate colors are treated as a single color rather than a composite of two, and where three colors already exist within an 8-pixel horizontal span. You should aim to avoid creating such situations in the first place. I will explain the workflow for that in detail later.

Adjustment work is fast

Depending on the tool you use, when making pixel-by-pixel corrections, adjusting line shading is much easier than adjusting tiling.

MSX1-like

MSX1 game screens often have a look that strongly emphasizes horizontal lines (a result of trying to draw beautiful pictures under the 2-color limit per 8 horizontal pixels). This is especially true for gradients. When drawing MSX1-style CG, you can emphasize the MSX1-ness by actively using horizontal line shading for intermediate color representation.

Maze of Galious ©1987 KONAMI

When drawing MSX1-style CG in the mid-2020s, there is a strong desire to finish it as a beautiful CG that would have been impossible back then, but at the same time, I want to retain that MSX1 feel. While clumsy expression (such as colors bleeding out) can be considered part of the MSX1 style, I don't want to do that on purpose. Within this dilemma, finishing it as a beautiful CG that strongly emphasizes the horizontal direction seems like a good compromise.

It looks like a scanline effect and is cool (perhaps

Scanlines are often applied to pixel art as a CRT-style effect. I feel that by making heavy use of horizontal line shading in CG, a cool, scanline-like atmosphere is created.

Example of scanlines: YS2 1988 © Nihon Falcom

Challenging black outlines

A typical drawing method in early microcomputer adventure games involves painting intermediate colors within black outlines. This is fatally incompatible with the MSX1's 2-color limit per 8 pixels. This is because a third color, black, is forced to intrude between the two colors used for shading.

Dezeni World 1983 © Hudson

The reason why the MSX versions of Salad Kingdom and Dezeni Land were 2-color screens of blue and white was likely because it was difficult to perform so-called bucket fills using the same methods as other models. The sequel, Dezeni World, was never even released (though the difficulty of drawing the art might not be the reason for it not being released. Perhaps it was the capacity limit of the Bee Card or profitability?)

I will try to tackle this difficult enemy of black outlines + shading by making full use of horizontal line shading. Is this a reckless battle?

First of all, in the shading method of the image above, there are many places where the 2-color rule within 8 horizontal pixels is violated,

When adjusted like this, there are no rule violations. By having a common color in the constituent elements of the two intermediate colors, placing a black dot at the boundary, and placing one of the colors (the darker one is better) at the boundary where the constituent elements differ. When you look at this picture from a distance, it feels like the outlines are placed normally. Since the color palette is limited, this method cannot always be used, but it can be applied as a trick in small areas quite often.

Now, the MSX1 rewrite CG of Mirika from XZR that I introduced at the beginning was finished with the pretense that sprites were placed where I wanted to draw black outlines.

The parts enclosed in frames are sprites

If you use sprite mobilization in this way, it seems that black outline + shading ADV art can be reproduced on the MSX1 to some extent. I want to draw Salad Kingdom.

Weaknesses of Tiling Patterns

As I have already mentioned slightly, horizontal line shading cannot be perceived as an intermediate color unless it is assigned to a somewhat large area, and instead just feels like a horizontal striped pattern. In contrast, with tiling, it looks relatively like an intermediate color even if the area is small.

Which is more beautiful as an intermediate color: checkerboard, vertical stripes, or horizontal stripes?

I had written part of this topic as well, but since it is a story that also applies to models other than the MSX1 (Famicom, PC Engine, Mega Drive, Super Famicom...), I decided to split it into a separate post. It is quite profound.

When Tiling Representation is Preferable

I have been pushing for horizontal line shading up to this point, but there are many cases where it is better not to use horizontal stripes. Here are a few examples.

When drawing small pictures

To repeat, when drawing small pixel art such as game assets, it is better to avoid horizontal line shading—or rather, it is ineffective and makes it difficult to express changes in fine shapes. Since pixel art requires placing each pixel based on skillful calculation, it is no longer a question of which intermediate color dithering method is better.

When representing natural objects

When representing natural scenery such as rocks or the ground, tiling, which can produce a sense of randomness and ruggedness, is more suitable. In the case of horizontal lines, it ends up looking artificial.

When you want to express a rough surface

Similar to the above, if you want to include weathering effects, tiling is still more suitable. I would like to use tiling mainly for expressions such as creature illustrations or the shadow colors of skin on a rugged old man.

When you want to draw a picture with warmth

In the case of expressions where light is spreading, it is desirable to blur it with graduated tiling. Horizontal line shading has a strong sense of direction and feels hard and cold.

When you want to add contrast to the entire image

If the entire screen is represented by horizontal line shading, it becomes monotonous, so I think it is good to actively insert tiling where possible to create contrast. For the purpose of creating contrast, it is better to also use solid base colors, even if they deviate from the original color scheme.

Workflow for Applying Dither Patterns

Listing so many situations where tiling is appropriate gives the feeling that maybe it would be better to make tiling the main focus, but as I have already stated, horizontal line shading is easier to handle and more convenient given the MSX1's 8px horizontal limit.

As a workflow, if you adopt the procedure of proceeding with solid colors and horizontal line shading from the beginning to the middle of CG creation, and then replacing areas with tiling in the final adjustment phase, there will be less rework and failure, and corrections will be faster. The quality will also be consistently high. If you decide to adopt a style that boldly avoids tiling, you can even handle commissioned work with short deadlines (though such work doesn't exist)!

Let's proceed without dithering

Having explained all that, I actually recommend that in the early stages of creating MSX1-style CG, you paint lines without tiling and intentionally use intermediate colors that cannot be produced on the MSX1 as solid colors (while keeping the rule of two colors per 8 horizontal pixels in mind). What is that, you ask?

This is the reason why the MSX1-style CG creation toolkit for Photoshop that I am currently developing (though it is already at a practical stage) is designed to allow the selection of intermediate colors that cannot be used on the MSX1. I will discuss the reasoning behind this another time.

Gradius II in the header image

This is a speculative MSX1 version of the Gradius II title screen, drawn with a strong application of horizontal line painting. It is not Gradius 2.

I created the base image by retouching, cutting, pasting, and adjusting the colors of an existing image in Photoshop, then converted it to MSX1 colors and horizontal line painting using my own converter. I finished it by manually adding the title logo and the blown-out white areas. It took less than a day of work. I feel like 30% of that time was spent drawing the logo (lol). Even images with unreasonable compositions end up looking quite clean.

Next time, I plan to summarize some examples or how to use color deception.

いいなと思ったら応援しよう!