.

[tritosine2k]

Perhaps this is a better resource :

http://www.vgamuseum.info/index.php/companies/itemlist/category/26-number-nine-visual-technology

They were already at 1024*768 in 1983 and downsample at will.

What you guys propose is using an audio waveform editor on sample by sample basis which is as ridiculous notion as it gets except for outliers.

It's another thing gaming hw cannot handle high quality sprites for a while.

Perhaps look up what Karl Guttag had to do with sprites.

 

[Rincewind]

During VGA days, awesome. In 2021, bandwagon-hopping hipster shit.

Pretty much what this dude says.

Although there are exceptions; when it's done with taste and true to the oldschool spirit, I love it still (e.g. SKALD, or some post-2000 adventures like Gemini Rue).

It's a bit hard to define what counts are hipster shit, though, but I just know it when I see it. Of course, huge pixels are dead giveaways, that shit should be burned with fire. I think at least 640x480 with 2x pixels is the acceptable minimum in 2021, 320x200 looks like pure shit when it fills a typical computer screen.

 

[Rincewind]

The pixel art of Mark Ferrari in the original PC versions of Secret of Monkey Island and Loom are among the best examples of serious technological limitations leading to wonderfully creative results. His masterful use of the 16-colour EGA palette and dither technique is nothing short of amazing. Many (including Mark Ferrari) think that the later VGA paint-overs pale in comparison in terms of artistic merit.

 

[Rincewind]

Some cool pixel art that predates Deluxe Paint by a few millennia Wonder how many colours the artist used; I guess it could be done with a 32-colour palette.

 

[Grauken]

the first time people went to war over whether something is pixel art or not

 

[Nutmeg]

I am firmly on the side that pixels aren't squares.

For one, older displays would stretch their image to fit the display aspect ratio, no matter the dimensions of the incoming image in pixels. If an image was 100 by 100 squares and the screen was 4:3, then each of those pixels was actually 4/3s as wide as it was tall. You can easily see this in e.g. SFC games where if you use square pixels, circles will be ovals.

But even this isn't the whole story.

The best way to think about it, is that pixels are constructed signal samples. Display technology (CRT) of the past was very sophisticated in the way it reconstructed a continuous (or analog) image out of the samples.

It was done by scanning out images line by line, over what amounted to sharpening masks, employing what amounted to gaussian interference across samples on the same line.

The lines themselves were separated by dark areas, with brighter parts occupying more vertical space.

The end effect is very similar to today's advanced software scaling techniques e.g. http://johncostella.com/magic/ but to me even more subjectively pleasing.

 

[Rincewind]

I am firmly on the side that pixels aren't squares.

For one, older displays would stretch their image to fit the display aspect ratio, no matter the dimensions of the incoming image in pixels. If an image was 100 by 100 squares and the screen was 4:3, then each of those pixels was actually 4/3s as wide as it was tall. You can easily see this in e.g. SFC games where if you use square pixels, circles will be ovals.

It depends. 320x200 MCGA (VGA low-res) has a pixel aspect ratio of 8:5, while the Amiga has a perfectly square pixel aspect ratio in low-res modes (e.g. 320x256, or its overscan variants). The Commodore 64 has almost square pixels, but the exact ratio is 0,936:1.

The lines themselves were separated by dark areas, with brighter parts occupying more vertical space.

The end effect is very similar to today's advanced software scaling techniques e.g. http://johncostella.com/magic/ but to me even more subjectively pleasing.

That's certainly true and it gives the image a subtle texture which is very hard to properly replicate with shaders (hence I don't use any shaders in emulators). You can see the effect on these photos taken from a real Commodore monitor (from AmigaLove), and I agree, it's quite nice (except for the 50/60 Hz strobe effect).

 

[Morenatsu.]

The console games I've played often tended to use square pixels, actually, even when the resolution wasn't 4:3.

 

[Serus]

People tend to concentrate too much on the emulation of the pixels as they were done back then. This isn't about emulating any exact method. It is about creating graphics using a a given set of tools and limitations. But it doesn't have to be the exact same set of tools, as back then, as in the 80s and 90s. The goal is good graphics, not a 100% correct emulation. As such there undoubtedly ARE some good pixel graphics made today. Unlike some edgy people claim.
Still, it means that most are shit, done by hipsters who do not understand the history of the medium and the technique they're using. So what? That almost everything is shit? Not all games had good graphics in the 80s and 90 (regardless the tech limitation) too. Even if the % of "good" was higher. But on the other hand back then not everyone and his dog could make a game (and publish it) so it is to be expected.
Basically what Grauken is trying to say - if i understand him correctly.

Spoiler

All this comes from someone who had an Amiga 500 and also played on a 286 and a SNES - all in the early 90s, so i do know how the original pixel graphics looked like. In case someone asks.
And the last thing, without a 1084s you were a pleb, Kev Inkline, I agree.

 

[Assisted Living Godzilla]

Done wrong:

I don't remember games looking that shitty at the time when "pixel art" was the only option out there.

But hey, I bet these "retro" lego bricks have dynamic shader lighting. Heck, maybe physics too! How about some ASCII graphics in 3D while we're there?

There most likely was never a time when something quite like that would've been a thing in the past. It's like a mix of shitty looking '80s PC games when Western developers didn't really know what they were doing when it came to pixel art, and later stuff when they had more colors to work with. It's like they're trying to make it look like one of Sierra's AGI Quest games, but with 256 colors.

 

[Bad Sector]

Aren't you supposed to look at these amiga images with some CRT scanline method? As it was intended ?

No, see below:

I am not sure it affects much of their visual appeal, to me Amiga screenshots looks pretty much the same as they did in 320x200 resolution, even on a modern pc screen.

Regarding screens and blurring, the CRT monitors we used for our PC games back in the DOS days were very different from TVs and would accurately display pixels as sharp blocks. These monitors were originally made for business applications so they had to be sharp enough to display readable text and numbers. Amigas and such were also intended to be used with monitors even though a lot of kids had to make do with a TV.

Indeed, monitor CRTs are different to TV CRTs and all the filter, blurry sprites, etc people show are from console games which were meant to be played on a TV.

A monitor CRT is meant to be crisp and you should be able to see the pixels even on a small CRT. For example this is a photo i took from one of my CRTs:

The pixel grid is clearly visible and the image is pretty much like you'd see on an LCD (except with better contrast and black on the CRT because aside from OLED flat panel displays suck there). The glow and smear is not because of the CRT but because my phone is ancient and its camera sucks.

 

[Nutmeg]

For example this is a photo i took from one of my CRTs:

Your set up is wrong. You clearly have more than one line per row of pixels. It looks like you're scaling the image up using nearest neighbor and then displaying it.

See Rincewind's photos for correct use of a CRT. Each line is one row of pixels (rather samples). Lighter colors expand the beam while darker colors contract it. This is an important effect when it comes to image reconstruction from discrete samples (as it corresponds to both a low and high pass filter approximating the sinc function). By prescaling with nearest neighbor you lose that.

 

[Morenatsu.]

For example this is a photo i took from one of my CRTs:

Your set up is wrong. You clearly have more than one line per row of pixels. It looks like you're scaling the image up using nearest neighbor and then displaying it.

See Rincewind's photos for correct use of a CRT. Each line is one row of pixels (rather samples). Lighter colors expand the beam while darker colors contract it. This is an important effect when it comes to image reconstruction from discrete samples and by prescaling with nearest neighbor you lose that.

low-res VGA is double-scanned

 

[Spectacle]

For example this is a photo i took from one of my CRTs:

Your set up is wrong. You clearly have more than one line per row of pixels. It looks like you're scaling the image up using nearest neighbor and then displaying it.

See Rincewind's photos for correct use of a CRT. Each line is one row of pixels (rather samples). Lighter colors expand the beam while darker colors contract it. This is an important effect when it comes to image reconstruction from discrete samples (as it corresponds to both a low and high pass filter approximating the sinc function). By prescaling with nearest neighbor you lose that.

Monitors are not the same. On later, quality models you had to look very carefully to notice the actual scanlines.

 

[Grauken]

For example this is a photo i took from one of my CRTs:

Your set up is wrong. You clearly have more than one line per row of pixels. It looks like you're scaling the image up using nearest neighbor and then displaying it.

See Rincewind's photos for correct use of a CRT. Each line is one row of pixels (rather samples). Lighter colors expand the beam while darker colors contract it. This is an important effect when it comes to image reconstruction from discrete samples (as it corresponds to both a low and high pass filter approximating the sinc function). By prescaling with nearest neighbor you lose that.

Monitors are not the same. On later, quality models you had to look very carefully to notice the actual scanlines.

Yeah, my goldbox games never looked like the pictures Rincewind showed, CRT scanlines was a TV thing mostly, or if something was misaligned with your monitor

 

[Nutmeg]

For example this is a photo i took from one of my CRTs:

Your set up is wrong. You clearly have more than one line per row of pixels. It looks like you're scaling the image up using nearest neighbor and then displaying it.

See Rincewind's photos for correct use of a CRT. Each line is one row of pixels (rather samples). Lighter colors expand the beam while darker colors contract it. This is an important effect when it comes to image reconstruction from discrete samples and by prescaling with nearest neighbor you lose that.

low-res VGA is double-scanned

I know VGA 13h 320 by 200 gets sent as 320 by 400 over the wire and it was wrong then and it's wrong now. From a signal processing point of view.

Monitors are not the same. On later, quality models you had to look very carefully to notice the actual scanlines.

Unless the signal was doubled (see above) higher resolution screens increased the dark areas between lines as lines were thinner.

 

[Nutmeg]

Yeah, my goldbox games never looked like the pictures Rincewind showed

You missed out tbh!

 

[Grauken]

Yeah, my goldbox games never looked like the pictures Rincewind showed

You missed out tbh!

Nobody had massive visible scanlines on their PC monitors at the time. You sound like someone who learned about CRT monitors from a Wikipedia article and never actually used one

 

[Nutmeg]

ou sound like someone who learned about CRT monitors from a Wikipedia article and never actually used one

I've owned 10s of CRTs tyvm.

Anyway, just to explain why signals with lower line counts have thicker scan lines on higher vertical resolution screens. Say you have a screen capable of resolving 960 distinct black and white lines across its height, and you make the whole screen white using a 960 line signal. Then 4 lines worth of screen might look like

W
W
W
W

Where W is for white. But if you display a 480 line signal you'd get

W
B
W
B

where B is for Black. and if you display a 240 line signal you'd get

W
B
B
W

i.e. more black. For whatever reason, people call these black lines "scan lines". So the higher resolution the screen, the thicker the scan lines.

 

[newtmonkey]

DOS VGA games simply do not have prominent scanlines. I have an actual DOS machine in my office that I use for playing games on a CRT monitor. The doublescan mode means that you have a single black line in between every two rows of horizontal pixels displayed. In effect, the black line becomes so faint that it contributes nothing to the image. No one designed their art with TV-like scanlines in mind, once VGA became mainstream.

EGA on an EGA card with EGA monitor is a different story (as is CGA composite on a TV). Likewise, probably, for game art for the C64, Apple II, Atari 8-bit computers, and arguably the Amiga/Atari ST as well.

Scanline "blooming" is a consumer TV thing. It barely occurs on a PC VGA monitor or even on a professional monitor such as a Sony PVM or BVM.

Having said that, I love the scanline effect of a console plugged into a TV (or an upscaler or emulator with an excellent scanline filter). The blooming causes brighter colors to have thicker scanlines and darker colors to have thinner scanlines, which results in lines narrowing smoothly to black rather than with a stair-step pattern like you get with the raw image. The scanline effect also provides a degree of free anti-aliasing. It makes low res games look quite a bit sharper, especially on a nice CRT TV.

 

[Nutmeg]

In effect, the black line becomes so faint that it contributes nothing to the image

It does contribute, but you're right very little on your average 19" consumer monitor. Look at a 480 line signal image on a large 24" 1440p CRT PC monitor (e.g. the Sony FW900), they will be quite prominent.

 

[Nutmeg]

Scanline "blooming" is a consumer TV thing. It barely occurs on a PC VGA monitor or even on a professional monitor such as a Sony PVM or BVM.

Oh you edited your post. Oh and again! haha. Well this part is wrong.

This is a macro photo of an extremely high res BVM. Notice how the darker colors are much thinner than the lighter colors.

 

[newtmonkey]

Licorice
I will take your word for it as I have not played any PC games on a 1440 res PC monitor, but what I'd like to know is how a DOS 13h mode game looks on that kind of monitor. I think with doublescan taken into consideration the black lines would be more noticeable but still wouldn't contribute much at all to the image.

As for scanline blooming on the BVM, I stand corrected. It's not as noticeable on my PVM.

 

[Bad Sector]

Your set up is wrong. You clearly have more than one line per row of pixels. It looks like you're scaling the image up using nearest neighbor and then displaying it.

What setup are you talking about, this is a Samtron CRT (i do not remember the exact model) connected to a Pentium 3 PC with a VGA cable, nothing special about it. It has more than one line per row of pixels because VGA monitors did double scan for lower resolutions - this is one of the major differences between monitor CRTs and TV CRTs and the entire point of my message that these two are not the same. Monitor CRTs on PCs look pretty much the same as you'd run them on an LCD with nearest neighbor scaling. You can see some scanlines in the larger CRTs but those do not affect the crispiness of the image - you can see the "square pixels" perfectly fine and you should be able to see them on a monitor CRT because they are supposed to be crisp for displaying high resolution text.

 

[Nutmeg]

this is one of the major differences between monitor CRTs

It's a difference between PC graphics cards adhering to the VGA standard and other graphics output standards. It's not a difference between PC monitor and TV displays.

What double scanning amounts to in image processing terms is just nearest neighbor doubling vertically. That paired with the fact that most PC monitors had horizontal resolutions 800+ means horizontally you get sharp borders between pixels too.

I concede that VGA games were made with chunky pixels in mind (still not squares like on an LCD tho).

So from the point of view of reproducing what the game artists saw, your set up is fine for VGA games. From the point of view of reproducing an image in the mathematical sense, VGA itself is bad and you should use a hardware line halver. From the point of view of subjective preference, it's up to you, of course.