Actually leveraging the power of palletized graphics has been a lost art for well over a decade. Mark Ferrari is the only person I know of that I can call a master of that art.
When I was working on PlayStation 2 games, it was frustrating that all of the artists I worked with were creating textures that would ultimately end up palletized, but none of them had any concept of the techniques presented here even then. These days GPUs are getting powerful enough that I occasionally see kids reinventing these techniques using tricky shaders from first principles without knowing that they used to be the only way to get things done.
In the HN circles you have a chance to be familiar with Ferrari's work through [1][2]
Mark Ferrari also teamed up with Ron Gilbert and Gary Winnick (et al) to create a new 2D adventure game called Thimbleweed Park[1] (No affiliation!) - awesome backgrounds by Ferrari!
In this talk he goes into detail about the Thimbleweed Park art he made and how he uses Photoshop to pull off pixel art techniques far more efficiently.
Palatte shifting and limited color palates were born from the necessity of effeciency. Now he has to use an emulated DOS program to edit them. He uses an experimental web app to display palate shifting projects he created years ago and looks forward to the upcomming release of a pixel editor from the same author.
Additionally, he expresses extensive configuration necessary, mostly disabling anti-aliasing within multiple tools, which makes his process possible in Photoshop. He also has removed the limitation of color palate to make the final product more pleasing, regardless of weight and efficiency, as the concern is minimal on modern hardware.
His technique in Photoshop makes heavy use of the pencil tool one pixel at a time and even falls back to antiquated software (or manual technique) when he needs to implement dithered
gradients.
I did not get the inpression that he was happy to be using Photoshop, but that it is currently the best option he has.
I took away something different. He emphasized several times how Photoshop enables him to pull off pixel art style but without having to individually drop in pixels. He said that about the lighting templates, the dithering template, and the anti aliasing technique. I'd argue the mere mention of the word "template" supports this. In all cases, "not having to do one pixel at a time" seemed to be the primary advantage.
It's true he also said sometimes he still has to manually put in pixels, but that seemed to exception to me.
The selection wand is nothing revolutionary, and likely works better for his purposes in the obsolete software. PS anti-aliases selections by default, so there's a specific example of a tool he's had to tune for his style of art.
The dithering template only exists because dithering is not a concept in PS; the DOS application has a dedicated dithered gradient tool.
Called paradise lost, and the trick the graphic team are using to fit 8bit pixelated graphics to the modern taste are incredible, like having to tweak animations so no parts ever move half a pixel, which influences walking speed and dozen pther things. Their blog is quite informative.
Absolutely, just the other day I was reminiscing about playing J2ME games in my little phone while I was growing up. There were some true gems in there.
I almost overlooked this link. The presentation is very nice. It starts with an evolution of 8 bit (256 color palette/EGA) images at Lucas Games, from simple to dithered to more advanced (and impressive) animations based on palette-shifting.
It then uses the metaphor of "Renaissance" vs. "Renaissance Faire" to talk about an 8-bit style, which doesn't have to reproduce the exact details of 8-bit colors, but rather the aesthetic - "8-bitish art". Followed by details of how to do it.
I feel you omitted one of the best parts of the talk. When he is explaining the palette techniques used in the X-Men Storm game where a single image can be clouds, trees or a cityscape with just a palette swap. I actually let out a small gasp when he went over that, really impressive.
This is the bit that really got me. As soon as I watched that I immediately paused the talk and emailed anyone I could think of that might be interested and demanded they go watch it! The technique was stunning, really impressive stuff
It seems with graphics and especially with games, there are two uses of "8-bit graphics"
- One, like this, refers to 256 color (or less) graphics
- The other, refers to graphics made using the constraints of real world common 8-bit computers (in the US, the Apple ][ or C64, in Europe, the ZX Spectrum and Amstrad CPC, etc) or 8-bit consoles, such as the NES and Sega Master System.
There are a lot of games which try to emulate these styles, and there are very few true accurate 8-bit (or 16-bit in the case of SNES, Genesis/Mega Drive, Amiga, ST) games. I always appreciated the constraints of having small amounts of memory and so few colors to play with when I was messing around with my Amstrad CPC as a kid.
I think it's no accident that one of the most renown and influential works of classical music ever made — the Well-Tempered Klavier — was mostly written for 3-4 voices, for an instrument with very limited expressiveness, and under the incredibly strict rules of counterpoint. To echo what Mark is saying, this is probably one of the reasons we find classical video game music so memorable. When you only have access to a sine wave, a saw wave, and a noise channel, you get really, really good at inventively utilizing these resources to their fullest. I can't remember most of the over-orchestrated, insipid soundtracks in AAA games today, but music from my favorite NES, SNES, and DOS games has never left my mind.
Another example of how constraints drive creativity, interesting ideas and algorithms.
It's always interesting to me where this appears - you see a ton of it from Carmack in "Masters of Doom" and it's happening again with current hardware in VR.
There's probably something to be said for purposefully setting up constraints that otherwise wouldn't need to exist in order come up with something interesting - and learn more faster.
Deluxe Paint II enhanced had a demo of two images in one with palette shifting that shipped with it.
It really is a phenomenal program. I had to do some pixel-art drawing c.a. 2000 and found nothing remotely as good; fortunately I still had a machine that would boot to dos and my copy of DPII.
He mentions Deluxe Paint II; I was fortunate among my PC owning friends to have that rather than PC Paintbrush. It was one of those rare tools that was so well made for its purpose as to make it hard to describe what was so good about it. Plus it shipped with several tutorials and a very complete manual.
My favourite feature, and I suspect of many others given how often I saw this at demo parties etc.:
The split screen, pixel perfect zoom.
I've been looking for paint apps for Android for years, and one of my pet hates is that finding one that didn't do bilinear filtering was incredibly hard (I finally found one I like reasonably well called "Artflow").
But even without filtering, having a quick and easy (one keypress) zoom that keeps the un-zoomed image visible on one side, and shows the zoomed region on the other, was vital to me, because it was one of the most frequent operations. Pretty much every paint app I see either complicates this with multiple separate views in separate windows, or by showing only a zoomed in view.
I find one of the biggest problem with modern paint applications is that they crowd the screen and don't surface the most important paint tools very well - they try to be too many things.
When I used Deluxe Paint, I wanted as little clutter as possible. I'd paint with everything off except when I wanted to change a tool, or change colour - and this to was one keypress away.
Though (relatively) low-res pixel art also tends to lend itself to using far fewer tools. I tended to mostly use the "spray can", smooth, and zoom in and draw individual pixels.
$ youtube-dl http://www.gdcvault.com/play/1023586/8-Bit-8 Bitish-Graphics
[GDCVault] 8-Bit-8-Bitish-Graphics: Downloading webpage
[GDCVault] 8-Bit-8-Bitish-Graphics: Downloading XML
[download] Destination: 8 Bit & '8 Bitish' Graphics-Outside the Box-1023586.mp4
[download] 61.0% of 714.86MiB at 12.99MiB/s ETA 00:21
Thanks for the submission. One thing I'm confused about. So, he said that in the animated scenes, the animation was done with palette shifting, and not separate components/layers. I understand that bit. However, how would the animation be done in the game? Does he mean that the palette shifting is implemented within the game, or is the palette shifting technique to easily and quickly produce art assets?
> Back then video cards could only render 256 colors at a time, so a palette of selected colors was used. But the programmer could change this palette at will, and all the onscreen colors would instantly change to match. It was fast, and took virtually no memory. Thus began the era of color cycling.
The technique is even older than that -- most graphics modes on the 8-bit Ataris only supported a 4-color palette. To get more colors on the screen, you had to take an interrupt partway through the frame to swap out some of the colors and then put them back during the vertical blank to be ready for the next frame.
At the time, the video cards worked by having a 4 or 8 bit image resident in video RAM and also a 16 or 256 entry pallete of colors as well (18 bit was common in VGA, you didn't even get 24bit palletes!). The card would do the pallete transform every frame pixel by pixel as it scanned out the image to the CRT. The point being: at no point in time did a 18 bit, full color image exist anywhere in RAM.
That meant even on a machine that was too slow to even just upload a whole frame of 320x200 8 bit pixels 30 times/second, you could still easily upload a pallete of 256 colors every frame. Just doing that would change how the whole image displayed without actually doing the work of touching any of the pixels.
It bothers me when popular culture uses the term 8-bit graphics for low-resolution no matter what the pallette. They see pixelization and automatically use the term.
When I was working on PlayStation 2 games, it was frustrating that all of the artists I worked with were creating textures that would ultimately end up palletized, but none of them had any concept of the techniques presented here even then. These days GPUs are getting powerful enough that I occasionally see kids reinventing these techniques using tricky shaders from first principles without knowing that they used to be the only way to get things done.
In the HN circles you have a chance to be familiar with Ferrari's work through [1][2]
[1] http://www.effectgames.com/effect/article.psp.html/joe/Old_S...
[2] https://news.ycombinator.com/item?id=3890267
But, in gaming circles, you might be more familiar with his work in Loom, Secret of Monkey Isle and Indiana Jones and the Fate of Atlantis.