I count 11 distinct colors, so strictly speaking one could call it 4 bit graphics (thus, 8 bit is closer than 16 bit).

Those images have way too many colors (11 colors each) to be practical on any 8-bit console, and the colors used are not appropriate to an 8-bit console's color gambit.

They do, in fact, very roughly approximate the "low-fi" style of 16-bit console games like Earthbound on the SNES, which include certain hallmarks of 8-bit-style art, like hyper-saturated colors and pure black outlines, but within the greater color depth afforded by a 16-bit console.

So, to be clear, does 8/16-bit refer to color depth (256-colors vs highcolor) or word size (NES vs SNES)?

It can mean anything (in context) and this discussion has gone way off track...

I count 11 distinct colors, so strictly speaking one could call it 4 bit graphics (thus, 8 bit is closer than 16 bit).

Those images have way too many colors (11 colors each) to be practical on any 8-bit console, and the colors used are not appropriate to an 8-bit console's color gambit.

They do, in fact, very roughly approximate the "low-fi" style of 16-bit console games like Earthbound on the SNES, which include certain hallmarks of 8-bit-style art, like hyper-saturated colors and pure black outlines, but within the greater color depth afforded by a 16-bit console.

So, to be clear, does 8/16-bit refer to color depth (256-colors vs highcolor) or word size (NES vs SNES)?

Its almost always word size. CPU word size.

Most people who say "8-bit console" or "16-bit console" don't even know what a bit is. They just find it easier to categorize the consoles that way.

I'd put them as late-gen 8-bit sprites, remember the Master System guys.

I'd say they're too wide to be 8-bit, regardless. I agree it *could* be done on the SMS, but you'd consume too many of the sprites-per-scanline at that width. In the West, most people think NES when they think "8-bit graphics" -- It was the last popular 8-bit console, the SMS never gained much traction here (I've actually got two, myself, and enjoy the games I have, which definately look better than NES contemporaries, but often gameplay isn't as good) and the Atari was very, very limited.

In other parts of the world -- South and latin america, and Europe to a lesser extent, the SMS had a lot more traction (I think they made games, game-gear ports often, until 1996 or so. I know you can get PAL versions of Sonic games, Street Fighter II, Mortal Combat for the SMS) so people there would more readily recognize SMS-caliber art as "8-bit".

I'll also go further to say that art is only one aspect of making a game that feels 8-bit. If the sprites are too large it looses the feel, if the physics are too realistic or smooth it looses the feel, if things are able to move with total freedom it looses the feel, if color is used too freely it uses the feel, if backgrounds are highly animated (except with palette-swapping techniques) it can loose the feel. A retro game isn't just about 8-bit-style tiles and sprites, but the overall package.

But remember the NES and SMS can do 8x16 sprites, so those characters would only consume 2 sprites per scanline.

Regarding the off-topic conversation about high bit-depths -- yes, most Dx9 cards can render with 32bit float channels (and/or 16bit float channels), for 96bit colour (if you ignore the alpha channel because it's not typically "color"). The PS3/360 are included in this club

I'd say that it's likely very common for PS3/360 games to use 30bit (10bit x RGB), 32bit (16bit luminance + 8bit x 2 chrominance) or 48bit (16bit x RGB) colour as their internal rendering format. This will then get crunched down to 8bit x RGB or 10bit x RGB for output to the TV.

The situation is likely still the same on the next-gen consoles and PC's -- there's not much benefit in going up to 96bit colour from 48bit.

But remember the NES and SMS can do 8x16 sprites, so those characters would only consume 2 sprites per scanline.

No, that's 8 wide by 16 tall. OP's sprites consume 3 hardware sprites per scanline on either machine -- neither of them supported sprite widths of anything other than 8 pixels. Sprite width is typically the limiting factor in real sprite-based hardware -- its the only way for simple hardware to maintain a deterministic speed of execution, different hardware might support wider sprites or more sprites, but there's a fixed budget of processing power available per scanline, and results need to be available in hard real-time.

Even the NEO-GEO, which had the most capable sprite hardware of any console (Saturn was more capable, but supported sprites with hybrid sprite-polygon hardware) had sprites of a fixed width of 16 pixels (although they could be 16 to 512 pixels tall) and the hardware could "only" handle 96 sprites per scanline; interestingly, the NEO-GEO had no dedicated concept of tiles or background layers -- the entire screen is drawn using z-ordered sprites.

Some contemporary computers of the day had more-flexible blitter-based graphics (like the Amiga, and its consolized version, the CD32), but that's different.

As wise as I am, even I sometimes forget that its X-Y not Y-X.

I count 11 distinct colors, so strictly speaking one could call it 4 bit graphics (thus, 8 bit is closer than 16 bit).

Those images have way too many colors (11 colors each) to be practical on any 8-bit console, and the colors used are not appropriate to an 8-bit console's color gambit.

They do, in fact, very roughly approximate the "low-fi" style of 16-bit console games like Earthbound on the SNES, which include certain hallmarks of 8-bit-style art, like hyper-saturated colors and pure black outlines, but within the greater color depth afforded by a 16-bit console.

So, to be clear, does 8/16-bit refer to color depth (256-colors vs highcolor) or word size (NES vs SNES)?

I've been thinking about this recently... people say look at this 8bit style gfx...

I dont think they mean bit depth I think they classify based on consoles / computers cpu bits

On top of that most people say 8bit for nearly anything pixeled that looks like it has a limited pallete.

So if it was more like 16bit, someone will still say 8bit... some of that might even be they forgot the limitations

of 8bit machines.

It seems there is some confusion here ?

The "bit" aspect of consoles that was so popular back in the days wasn't about the bitrate of Still Images,but a quality of hardware,specifically processors.

NES had an 8bit processor,SNES a 16 bit processor,ps1 a 32bit processor,and N64 a 64bit processor.

There is a reason why companies stopped marketing 'bits' for their consoles,and that is because up until PS4 and Xbox One,nothing more than 32 bit CPU was actually needed.

Even the high end computers today,still use either 32bits or 64bits. 32 bit processors are good enough until you need more of 4gb of RAM. 64bit processing is only now starting to become mainstream,as more and more heavy applications are made,and users tend to multitask more. 128 bit processing isn't coming any time soon,as 64 bit processors can use up to 16 exabytes of memory.

That basically means that no matter the pixel count of each individual art piece you make,if you want let's say to make a true 16bit game what you should do is to make a game that is limited only to what 16bit processors could do.

A limitation of that,is that the game shouldn't need more of 16 mb of RAM to run...