I know that pre-rendered games use images of 3D scenes rendered on other hardware; then these images are used as backgrounds. I was wondering if the power of the console is irrelevant to visualize these images. What I mean exactly is: if the console only visualizes these images, and so it must not compute them as it does with games rendered in real time, is it possible that a pre-rendered game, created with the more sophisticated hardware and techniques today, can be visualized correctly with a console like Playstation 1 without problems (resolution and some other factor aside)? For example, to use pre-rendered images like those of Resident Evil 0 (Nintendo Gamecube) on the Playstation 1. Or does the console has its limits regarding the pre-rendered images which it is able to visualize?

Another question which connects to the previous one: what is the difference between pre-rendered images of games and the realistic scenarios used by architects etc? Can those realistic scenarios used as pre-rendered backgrounds? I did this reasoning: if with my dated computer (so with my Nvidia Geforce go 7300 Turbocache) of 2006 I can visualize renderings created with the most advanced hardware today, this should be the same regarding the matter of pre-rendered backgrounds created today with the most advanced hardware, put in old consoles like Playstation 1. Is this reasoning correct?

1 hour ago, orion87 said:

is it possible that a pre-rendered games, created with the more sophisticated hardware and techniques today, can be visualized correctly with a console like Playstation 1 without problems?

Yes. That's really a lot of the point of using a pre-rendered background, that you can expend more time (and technique) in its creation than a background that is rendered from scratch each frame.

1 hour ago, orion87 said:

what is the difference between pre-rendered images of games and the realistic scenarios used by architects etc? Can those realistic scenarios used as pre-rendered backgrounds?

Nothing, and yes.

Pre-rendered backgrounds have fallen out of favour lately for a number of reasons, partly because hardware is now so powerful that the gap between a pre-rendered background and realtime has closed significantly, and the general availability and ease of use of 3d engines with easy import from 3d packages.

In return for the benefits, there are also significant costs, primarily you lose a lot of the dynamic nature of the background. Obviously you can't have a 3d moving camera. If you want to be able to interact with the background, you would need to render it with a depth buffer. And these days, at least on desktop / console, you'd want to be encoding things like normals / materials instead of just colour, much like a deferred renderer.

That said they still have their place, particularly on mobiles and low power devices. A relative of the pre-rendering you are thinking of is a scrolling renderer, which I use in my jungle game (and first used in the 80s on BBC micros). Instead of rendering the whole screen every frame you just render the edges as you move around, and store a revolving 'window' which you then draw onto the screen. This can lead to vastly better performance, but again, the trade off is you are losing the dynamic nature.

Interestingly it is sometimes associated with Carmack with commander keen etc but side scrolling (and hardware side scrolling) was widely used from at least the early 80s, probably earlier in arcade machines. If I remember right on the BBC 6502 you had some registers you could tickle which would change the address used for the start of screen memory. Some other early computers didn't have this hardware ability and weren't so suited to side scrolling (sinclair spectrum I don't think had it).

So, only to give an example, are you virtually saying that the pre-rendered background below of Resident Evil 0 (Nintendo Gamecube) will give the same result when inserted in Playstation 1?

And that the image below is less detailed compared to the image above only because at that time, when Resident Evil 3 Nemesis (Playstation 1) was created, the hardware of the computers where these images were rendered was less powerful than the time when Resident Evil 0 (Nintendo Gamecube) was created?

54 minutes ago, orion87 said:

And that the image below is less detailed compared to the image above only because at that time, when Resident Evil 3 Nemesis (Playstation 1) was created, the hardware of the computers where these images were rendered was less powerful than the time when Resident Evil 0 (Nintendo Gamecube) was created?

Yes, technically you could use the new background on the old playstation (resolution and color limitations aside).

I remember the first offline videos in games were much worse in quality that what we can do now in realtime. Pathtracing and such was too slow even for offline decades ago.

Now I am satisfied because of these answers. Thanks for all the clarifications.

To add to this, it takes the same amount of time (in general) to render a pre-rendered image as it takes to render any other image that's the same resolution and color depth. An image is an image (mostly - things like compression can also play a big effect on performance), whether it's a pre-rendered scene, a sprite, or a real photograph.

Games like Myst (and it's sequel Riven) took advantage of this to make breathtakingly beautiful worlds (for the time) run on such (relative to the present) dramatically weak hardware. The downside is that it's not dynamic.

The reasons why earlier pre-rendered artwork was inferior is for a number of reasons. Part of it is, as others mentioned, lower resolution and lower number of colors available. Other reasons include that pre-rendering images isn't easy either! Toy Story 1 had each frame rendered on a server farm of 300 processors, and even then, every frame took between 45 minutes and 30 hours to render, and Pixar basically had to invent most of the technology and algorithms from scratch, and had brilliant computer scientists like Catmul heading it up. Average videogame studios didn't have the technical expertise in that (at the time) cutting-edge field, nor the massive amount of specialized computer hardware necessary to pull it off within their limited budgets.

Further, there's also the issue of the massive amount of 3D modelling that must be done to create models (to prerender) with very high polygon counts. The 3D modellers' computers need to be able to render the individual 3D models (without lighting) at least almost in real-time, otherwise modelling gets very frustrating fast, so this also puts a soft cap on the quality of the individual objects that will be pre-rendered.

The better option for PS1-era games would be having artists create the 2D backgrounds, perhaps with some basic polygonal scene modelling, rather than trying to 3D model complex scenes.

Legend of Dragoon and Chrono Cross looked very good (about the same quality as your RE3: Nemesis screenshot) when exploring the world, though in both those games the actual characters and all in-combat scenes used 3D models that were low-polygon and suffered in comparison to the non-combat backgrounds.

Diablo 2 pre-rendered all their character models, that way they could have their character models be higher polygons than they'd otherwise be able to do. To pre-render their models, they had to essentially take (automated) screenshots of their models from every rotation, for every keyframe of animation, and do the same for each piece of equipment, which they would layer on top (called "paper-dolling"), because different equipment in Diablo 2 affected your avatar's appearance.