All of the above, but also interlacing may have had something to do with it as well? I'm no expert on how interlacing effects image quality, but I have a general feeling it helps TVs punch above their weight a bit, and create an illusion of more fluid movement with less pixels needing to be redrawn.

I suspect it's due to the lower resolution of TV screens coupled with generally lower signal quality via composite or coax. The result is a blurrier image that looks less pixelated.

Computer monitors tend to be higher resolution and therefore produce a sharper pixelated appearance.

I agree it’s probably actually the higher quality of Pc monitors compared to TVs producing a sharper image. I never really thought about PC games as being more pixelated. I think the artists on the NES games were also higher quality, on the PC side they tended to be amateurs.

Perhaps the biggest difference early on was the color modes and scrolling available to the NES vs PC, by the time the graphically comparable games came along it was 1991 or so which was 5 years after the launch of the NES in America, give or take. That was all down to hardware limitations. Keep in mind Windows 3.0 wasn’t even released until 1990 and 3.1 in 92. So I think it comes down to hardware limitations on the PC especially with colors.

I think it's likely that your premise is correct, however I think your question about a possible psychological factor might also be shaping your perception. While some people here have period correct monitors for playing CGA/EGA games, most retro gamers probably haven't seen those graphics on really old hardware. Even those who have are probably going back over 30 years now. I suspect that a lot of our impressions of pre-VGA graphics have been biased by either playing old games on an SVGA CRT monitor or even a more modern LCD. For the latter, this includes people who do all of their DOS gaming in DOSBOX on a modern computer.

Now the NES is a different story. Far more retro gamers had an NES back in the day compared to those who had period correct setups for CGA/EGA gaming. Moreover, I think there are probably far more retrogamers who play console games on CRT TVs compared to those that play DOS games on CRTs. I even think that it's plausible that some people who unintentionally group other 80s computer systems with PCs, so their memory could be shaped by their experiences with the C64 for example.

Altogether, I think it's hard to get a unbiased impression of what you are asking. Nonetheless, I think you are still probably right. In addition to the technical differences between graphics on old consoles versus PCs, another consideration is graphic design choices. A lot of computer games were created by programmers without much training in art and graphic design. This could also account for some of the differences between the NES and PCs at the time that has nothing to do with technology differences.

I also agree a lot of it is psychological, I don’t remember NES games being any more or less pixelated than PC games and haven’t viewed them that way since then either.

Another reason may be the color palette. Especially the 4-color CGA palette(s) do not allow anything close to smooth color transition, so every color change is a sharp high-contrast edge. The NES allowed to choose from 55 colors. In 200-line mode, even EGA allowed only the 16 RGBI colors, which is "not optimized for game graphics", and even if all 64 colors were accessible, you would have to choose 16 of those colors for the whole screen, while the NES allowed choosing color sets per tile/sprite, allowing considerably more than 16 different colors on the screen at the same time.

It’s all because the NES supported a 54 color pallet with 16x15 4 color zones. (240x224 active/visible)

There is Japanese commentary that the nes color pallet required a lot of convoluted logic to generate and wasn’t a simple stepper, so somebody put a lot of extra circuitry to get the nes specific pallet which almost looks hand chosen. Despite the obvious meticulous work the nes is very deficient in certain color types (like you would use in international sports games.)

The main head scratcher is why the nes added the complexity of the PPUs accent bits but didn’t go full Amiga implementing a counter or omission list for sprites. If the moderately complex Ppu would have added programmable functions with a timer it could have added the appearance of filter, transparency and water/snow effects to the console with minimal additional circuitry. As it was the 480 NES colors went virtually unused because there was no timer and no programmability even for color transitions it did little.

CGA was always a bit of a head scratcher like IBM took extra effort to gimp it.

As just one example CGA had 384 bytes free memory. But only used 2 or 3 of these bytes for configuration, boarder and color pallet.
AKA there was a ton of spare memory to setup a correct 16 color pallet
Hell each of the 200 lines could have had their own 4 unique colors palette line by line.

PCJR/Tandy mostly corrected the 4 color palette but almost no one used it sticking with mostly generic CGA bright or CGA red.

I find the psychological factor really interesting. Most people are unaware of how easily their perceptions can be biased and the inherent subjectivity of regular experiences. A good example is wine tasting. Research has shown that the first reviews to come out have a big influence on subsequent reviews. The earliest reviews are key to a wine's reputation. You can even convince people that tap water is actually a premium bottled water by lying to them about what they are drinking. Think about that. Water is something most people drink everyday and you can convince people that they are drinking something fancier than what they get out of their tap at home.

I think that a similar thing probably occurs in retro PC gaming. I really like showdowns between old soundcards and midi devices. However, I think that our opinions on what sounds best is shaped by other people's opinions. Similarly, when people say that there's a noticeable difference between midi music recorded from old hardware and emulation, I often wonder if telling people what they are listening to affects their perception. I prefer videos that don't tell you which one you were listening to until the end of that reason.

Coincidentally, I have been cleaning several devices from the time these days, an Amstrad CPC 464 and a 6128, a SNES and today finally an NES, I tested them all via SCART connection with adapters on a 19" LCD TV and there is a big difference in what the Amstrad computers and consoles look like. The computers have more defined edges, you can clearly see where one pixel ends and where the other begins;
On the consoles you have to pay attention to notice the edges of the sprites, it seems to me that the consoles make some type of mixture between one pixel and the other, which on a CRT screen I think would disguise the pixels more. The pixels of the consoles are not square, they seem rectangular, in the case of the Amstrad, in mode 2 (higher res) the pixel is rectangular vertically, in mode 1 (medium res) it is square and in mode 0 (lower res) it is rectangular horizontally. But in all modes the pixels are more clear than in the consoles.
On the NES and SNES they appear rectangular horizontally.
I think it's the same with the PC, more definition in the pixels, which makes it look more pixelated, while the consoles use some method to blur the edges of the sprites and make them look less defined and look less pixelated.
This is how I see it.

The interlacing on TVs during the heydays of CRTs really make a difference to these (g)olden console games.

I still use the interlacing shaders in my emulations, be it Kega, Nestopia, Zsnes, Mame or RetroArch.

Malik wrote on 2023-09-24, 23:00:

The interlacing on TVs during the heydays of CRTs really make a difference to these (g)olden console games.

I still use the interlacing shaders in my emulations, be it Kega, Nestopia, Zsnes, Mame or RetroArch.

I thought that most consoles from the CRT tv era used progressive scan rather than interlaced? I still have a lot to learn about TV formats, but I've read that old NES and SNES games often look wrong through many converters because the converters output 480i instead of 240p. Similarly, laptops with s-video outputs don't work great for CRT tv gaming because they only output 480i.

Ensign Nemo wrote on 2023-09-24, 23:49:

Malik wrote on 2023-09-24, 23:00:

The interlacing on TVs during the heydays of CRTs really make a difference to these (g)olden console games.

I still use the interlacing shaders in my emulations, be it Kega, Nestopia, Zsnes, Mame or RetroArch.

I thought that most consoles from the CRT tv era used progressive scan rather than interlaced? I still have a lot to learn about TV formats, but I've read that old NES and SNES games often look wrong through many converters because the converters output 480i instead of 240p. Similarly, laptops with s-video outputs don't work great for CRT tv gaming because they only output 480i.

Progressive scan on a tv just results in skipped lines

there's also ntsc color bleed and fuzzies, and the fact the nes doesn't really output rgb and there's a lot of varying NES experiences due to that. even emulators couldn't get the palette down for a long time, so there's options for raw color output for specialized ntsc shaders to decode with for the most accurate color reproduction possible (in software).

Even the worst VGA CRT have much better dot pitch quality than a consumer CRT, plus doublescan, and bad quality composite on the other side, is a big difference, but this is now, back on time, when we are young we did not notice difference, even i remember did not realice much about differences in arcade Operation Wolf and PC port, except obvious the machine gun jajaja

I remember, maybe 87 or 88, i was a young kid with my MSX, and a old friend of my dad, make me a RGB mod, and i realize how much better was quality compared to RF. Sadly same mod was impossible on the C64! i was dissapointed i remember