Half-dot technique

Maybe not the right place for this but I was looking at this quote and wondering if it’s a mistranslation:

“ *It’s a technique where by slightly changing the color of surrounding pixels, to the human eye it looks like the pixels move by around 0.5 pixels.* ” explains Kazuhiro Tanaka, graphic designer on Metal Slug (1996). His colleague Yasuyuki Oda adds that “ *Back in the old days, we’d say* [to our artists] *“add 0.5 of a pixel”, and have them draw in the pixels by taking scanlines into account. But with the modern Full HD monitor, the pixels comes out too clearly and too perfectly that you can’t have that same taste.* “

To me it sounds like half-dot technique has more to do with the crt screen’s phosphor structure than the scanlines. By choosing certain colors you can activate “half” (really it’s either 2/3rd or 1/3rd) of a phosphor triad, allowing you to draw “half pixels.” This allows for even finer details. Basically it sounds like “half-dot technique” is actually describing some kind of subpixel rendering (like ClearType).


The translation is fine, scanline bloom and width varies depending on the surrounding colour, it’s a similar technique used with a composite output to blend colours and create perceived detail.


I’m familiar with that aspect of scanlines, however it’s this part in particular that I’m hung up on:

*It’s a technique where by slightly changing the color of surrounding pixels, to the human eye it looks like the pixels move by around 0.5 pixels

the pixels move i.e., shift position by half a pixel.

I’m not sure how blending colors relates to shifting a pixel’s position by half a pixel.

Just having a hard time parsing this.

It’s called sub-pixel animation, some shit like that, I’ve read about it in interviews and in a pixel art how-to thing.

1 Like

Yeah I’ve seen that discussed before, I think that’s what’s being described in the quote. It doesn’t have anything to do with scanlines per se, though. :thinking:

1 Like

It’s straight up what is being discussed in that quote, the pixel art how to book I read, specifically references metal slug and capcom during the sub-pixel animation section of the book.

Can’t really comment on the validity of it being something that directly involved the scanlines

1 Like

Nice, what’s the name of the book?

Here’s an article describing the technique, there’s an example using Metal Slug at some point.

So the next head scratcher is what Yasayuki Oda means when he says “taking scanlines into account” I don’t see what subpixel animation has to do with scanlines per se.


Sorry for the delay!

It’s called Pixel Logic

Ye I don’t really follow what scanlines have to do with this, but I’ve never attempted to work/view my pixel art on a CRT, did it all on LCDs…

If I had to guess, probably something something, color bleed, adjecent pixels, optical illusion, here’s some toast!


This might result in a “duh, yes we know that already” response, but to me it seems to mean that the surrounding pixels being changed had gaps left where they knew the scanlines would be?


Kinda what my color bleed toast comment was about :joy:

Edit: Now I think about it some more, this doesn’t make sense in the context of it being used for Metal Slug tho imho, as Metal Slug would’ve been normally viewed via rgb/vga afaik, which isn’t really known for color bleed afaik. So idfk what is going on with life at this point ROFL



Shameless bump, but I got curious and looked into this myself.

The confusion comes from the Tumblr article compiling a lot of different info together and being a bit unclear on how it fits together. Half-dot is indeed about sub-pixel animation and isn’t inherently tied to CRTs, as evidenced by mentioning GBA games. Most of the devs cited (notably Kazuhiro Tanaka who’s explaining half-dot in that article) don’t even mention CRTs.

But a couple of them do and that’s what we’re interested in. Yoshinori Yamamoto explains the relationship better, I think: “Moving by 0.5 dots feels like a technique unique to the old days when you hit each dot one by one (sometimes it uses the blurring of a cathode ray tube).”

In other words, sub-pixel animation is about how adjusting a sprite’s colors give the illusion of movement. A CRT’s blurring would help that illusion. Yasuyuki Oda’s scanline quote is weird but if you look at the context of the quote, you’ll see it’s from an interview shilling KOF14 and justifying the move to 3D. I think it’s most likely that he’s just over-simplifying things for an audience unfamiliar with CRTs (after all, isn’t calling the blank lines “scanlines” already a case of that? :P)

Arcade games like Metal Slug were generally viewed via RGB, but those CRTs had very high dot pitches (as in, lower image quality), so while they had no color bleed they still had a much blurrier image than LCDs. These tubes didn’t have very noticeable “scanlines” (see below pic), in fact noticeable “scanlines” indicated a less blurry image, so the idea of designing around them would be strange and further suggests that Oda is over-simplifying to make a point about how he thinks LCDs are too clean for pixel art.

I wasn’t correct here. It’s true that US arcade monitors were low quality, but that wasn’t actually the case over in Japan. It’s hard to get exact specifications, but it does appear that these monitors were sharper than TV tubes, at least the monitors used by SNK and Capcom (the devs who’re being quoted here, see below pic for an example). There’s still a bit of blur, particularly in the outlines of the sprites (which is what’d benefit the half-dot effect) but not as much as I thought.

Still don’t think it’s worth thinking too much about Oda’s quote. I’ve recently ordered The Bitmap Brothers book which is actually why the CRT shader “Lottes” was even made, so maybe they’ll provide some detailed insights on how CRTs were used for pixel art.


I don’t think that picture is indicative of much. Seems also very unlikey they’d use vastly different tubes across markets for the same games. Even monitors of the same type can be in different condition.

1 Like

Yeah, these are fair points. In my haste to edit my post I did over-generalize. When I spoke of low quality US monitors I was mainly referring to the Wells-Gardener K7000 from the 80s. Meanwhile the SNK/Capcom games mentioned in the OP would’ve been more likely to be displayed on newer stuff. Can’t necessarily be chalked up to regional differences. I should do more research before I risk spreading more misinfo and generalizations.

Though I’ll add that I’m not just going off one picture but a lot of different sources. Here’s one from a documentary released from 1992, displaying a Nanao MS8. I’ve seen claims of this monitor having lower dot pitch than consumer TVs, but no actual concrete numbers or proof.

You don’t need to be so hard on yourself. :laughing:

I haven’t found any relevant specs in Nanao manuals either, but here’s a relatively good pic of a Nanao MS9.

I would think that at least late consumer sets are easily able to compete. This Loewe Calida pic blow me away when I first saw it:


Yeah, I guess I underestimated consumer sets. I only had crappy CRT TVs so I guess my bias leaked out :stuck_out_tongue:

Researching this is somewhat difficult because, like you pointed out, official specs don’t have a lot of info, so you have to rely on word of mouth of hobbyists who owned and messed with this stuff, and they sometimes contract each other.

Either way, it’s a bit of a tangent on my end, since the main point of my edit is that I over-stated how blurry arcade CRTs were, at least the ones that came out in the 90s.