Creating pixel art

And you can always choose not to move diagonally. Taking reference from SNES and Sega MegaDrive RPGs again, there are a number where you can’t. Also shooters etc. generally don’t move diagonally.

Hi Grauw!

If you feel that you can do a game in screen-10 or screen-12, go ahead and do it. I'm sure it will be great! It's sad that MKid was never finished, because it was very promising. SlotMan also made a demo of Wonder Boy on screen-10 that looked very promising.

My personal experience is that it's perfectly possible to create some beautiful pixel art for the YJK screen modes. Color bleed itself is not a real problem. If it was, there wouldn't have been such beautiful images and games crafted for both the MSX screen-2 or the Amiga HAM. The YJK color encoding is also very easy to learn, as long as you're used to work with other color spaces like YUV or HSL.

I really don't understand the prejudice against the YJK video modes. Amiga users don't have such negative attitude against HAM. In fact, it's the exact opposite: they think that it was a very intelligent trick to hack more colors into a limited amount of RAM. And I personally find HAM much harder to work than the industry-standard 4:1:1 chroma subsampling of the YJK modes.

Since I haven't been coding for the MSX since the surgery, as it requires longer periods of concentration that the rehab is consuming, I had to find some other mental hygiene activity that could be easily interrupted and progress in small chunks of time. That activity ended up being dealing with Pixel Art.

Problem is: I'm just terrible to draw anything from scratch. :) But if I have something to work over (even a line-art), than I can easily edit or post-process it.

So at first, I worked on restoring the damaged Pixel Art for the vol.1 Collection. Restored things like JPEG compression, missing parts/damaged, stretched images and so on. On this way, I learned a lot of tricks the artists used to hide the color bleed on screen-2. And the color illusions artists used to increase the number of perceived colors on screen-5 and 7. Konami, Compile, Bit2, T&E Soft, Micro Cabin etc clearly used such color illusion tricks on the MSX too.

I also analyzed a lot of Pixel Art for the C64, ZX-Spectrum and Amiga, to see how they could achieve such incredible visuals. I noticed that on the C64 they also play a lot with optical illusion tricks to increase the number of perceived simultaneous colors. And on the ZX-Spectrum they use similar tricks workaround the color bleeds. It's all about tricking the brain to see something that it's not really there.

While working on the images, I kept some of my favorite music arrangements from John Hassinck, Wolf, Jorito, Otobeya, Meits, gyabuneko, etc playing on the background, to keep me in the mood. Then I had the insight: "hey, if there are music rearrangements for the MSX, why can't we have graphic rearrangements too?!"

I also noticed that the vast majority of the Pixel Art available for the MSX was targeted at the screens 2, 5 and 7. There was some art for the screen-8, less than a dozen images for the YJK screens, barely nothing for the screens 3 and 6. The screen-7 with interlace was also seldom used.

And the vast majority of what has been produced for the screen-8 and 12 was just some poorly digitized images. To make things worse, even many games that had those for loading/title screens had such low quality work. Probably this is what gave these modes such a bad fame.

So I began working on rearrangements of suitable Pixel Art for the MSX. My main objective was to have some Pixel Art in the nearly-unused modes, and to experience the limits of what could be done with Pixel Art, when they were pushed to their limits. But I also rearranged some images for the screen-2, since I love to work on the colorbleed restrictions of this mode too. Yes, you can call me crazy because I like to work with colorbleed. It's like assembling a puzzle! ;)

And what can I tell about the experience? In a nutshell, that screen-8 becomes really impressive when the C64 style color illusion techniques are applied to it. And as some people already noticed on this thread, when you apply the screen-2 style color-bleed hiding techniques on the screen-12 it becomes like a "super screen-2". It's awesome.

Screen-10 is another league on itself. It's like having a screen-5 overlaid on top of a half-the-colors-screen-12. So both techniques must be combined. The main problem is that I couldn't find any decent image editor to do the post-processing in this mode. So I have very few images for this one.

Until now, this set of rearranged images was meant for personal viewing only. But if there is enough interest, I can publish the images as another Pixel Art Collection (maybe vol.2?) so you can check what I was able to achieve in these modes.

Please keep in mind that this collection is not just a "throw in the converter and were done" kind of work. While I used converters to speed up the work, their results for Pixel Art conversion are usually poor and serve only as a draft. I used the converters in the same way an artist would have used a scanner. To achieve the desired quality, a lot of pre-processing has to be done on the images, and also a lot of post-editing/cleanup. Besides MIFui for the "scanning" part, my main tools in fact where the GIMP, Graph Saurus 2.0 and the awesome Pixel Polizei.

Really nice stuff, FRS!

I find that pixel art itself is not so difficult, just a lot of learning of (interesting) techniques, thinking about physical properties like how light behaves (not something I expected to learn initially), and finding how tweaking a few pixels can make a huge difference.

But for me the difficult part is to design interesting art. Maybe that’s why I’m a coder and not an artist. It’s pretty easy to make boring generic-looking characters and environments. The hard part is to go beyond that. When I look at some characters from games or anime series, I really admire what details they added to make something look really cool or outlandish, yet fitting.

I take a lot of screenshots to serve as inspiration, when I see something cool or cute or unexpectedly awesome. Sometimes I feel like all I can do is to mix-and-match parts I like, make a variation on it to make it more unique, and then end up with hopefully a good original result. I guess nothing is truly unique so you shouldn’t feel too bad about getting inspired by the work of others, but I wonder if that’s how real good artists work as well, or if it just really comes straight from their imagination.

It really made me understand why in games development concept artists are really valuable. If I had concept art for my game, it seems like it would be a lot easier since I just have to bring that to the MSX in whatever size I need. I hope with practice I can get better at the concepting, I’m not giving up .

Totally agree! That’s also why I was trying to probe if people knew some techniques to create graphics for these MSX2+ modes in a better way. Similar to the topic of how to best use the colours of a limited 16-colour palette. It seems there’s not many people to learn from, so it’s a bit of pioneering. I hope to find a nice conclusion at some point, like if you do it like this and this, it’s going to work out.

Having done a bit of graphics in screen 4 for SMW, I can definitely appreciate that. Once you learn a few techniques, it’s not even very difficult I think. It was a really good feeling when I managed to get extremely close to the SNES graphics by just making smart choices about the tile alignment and the thickness of bands of colour.

I really like the dragon in screen 5, because it looks so varied in colour, as if there’s no 16 colour limit at all. It’s pretty hard to create scenes where it’s not obvious that you only have a select number of base gradients.

The screen 7 picture is nice as well, though I wouldn’t anti-alias black lines with red in the shadowy areas, it looks a bit converter-y to me still. Like it decided, well this is the closest colour I have available. Using the dark blue there would look better there I think.

That screen 10/11 pic looks really, really nice. That’s pretty much exactly what I had in mind when I was talking about pixel art in screen 11 earlier.

From the first screen 12 pic I think I can conclude that even if you don’t draw around the 4-pixel grid, but more freeform, if you use pastel-y colours (nearly grayscale with some light colour information), it’s going to be a pretty safe conversion without too much artefacts.

The second screen 12 picture I think shows that you can have pretty nice edges off the grid, but it does introduce a bit of Composite/S-Video-looking visuals (like around the hair), which are fairly subtle and acceptable in consumer electronics for years, though it’s better avoided.

By the way, as for the screen 10-12 pictures, did you generate them as they are shown in emulators and on RGB screens, or did you use the S-Video output filter that was discussed in an earlier thread?

One realisation I had today is that you can actually flip that thought around; don’t have screen 12 with a few patches of palette colours overlaid to fix it up. Rather, make a screen 5 image and “poke holes” in it in places where you want more details or colours.

If you want some flowers in a grassy patch but don’t have the colours, no problem! Just pick colours outside your palette and draw them. Just don’t put flowers of different colours too close to each other. Want a nice gradient on that wall, or details in the water? Just draw it. You get way more colourful graphics, and barely have to worry about attribute clash.

As you continue to do this, I think you will also slowly approach the point where the YJK pixels become the majority, and surpass it if you persist. You reach the same goal, but approaching it from the other direction, and maybe the result will be better integrated with the palette, and easier to achieve.

Definitely something worth trying.

Wow...

Definitely my favourite...