Are pre-render based environments dead?

["Nerve-Damage"]

Does anyone remember how great (IMHO, amazing) Resident Evil Zero looked on the GameCube?

The game had some of the best looking cinematic like pre-rendered backgrounds I have ever seen, especially the caves and tunnels. The environment were so “lively-like” in nature, that the pre-render landscapes, buildings, etc…didn’t stick out as being pre-render.

So I was wondering; are pre-rendered backgrounds or surrounding totally dead (in the vein of Resident Evil Zero, Parasite EVE, etc…)?

Are polygonal based environments more important to gaming than pre-rendered backdrops or surroundings?

Wouldn’t the Xbox 360 & PS3 benefit more (highest IQ settings, 1920x1080 render, etc...) if developers were to use pre-rendered based surroundings?

Wouldn’t the “saved” polygon budget be best spent on more character detail, more characters, objects, or anything else, rather than poly-based environments?

 

[....]

I think they should be used again particularly with BLU-ray, where you could easily fit high quality FMV CG in the backgrounds without much penalty in terms of data size.

Many games used fixed panning cameras, for a cinematic feel, parts of the environment or even the whole environment could easily be substituted by a prerendered 1080p feed. Cell too is quite suited for processing such video.

As it would lead to perfect iq in the environment, higher detail photorealistic player and enemy models with high lvls of AA, and it would be indistinguishable except in terms of quality in some titles(fixed cameras, or controlled fixed panning cameras.). I would indeed consider it ideal for many games.

 

["Nerve-Damage"]

I think they should be used again particularly with BLU-ray, where you could easily fit high quality FMV CG in the backgrounds without much penalty in terms of data size.

Many games used fixed panning cameras, for a cinematic feel, parts of the environment or even the whole environment could easily be substituted by a prerendered 1080p feed. Cell too is quite suited for processing such video.

As it would lead to perfect iq in the environment, higher detail photorealistic player and enemy models with high lvls of AA, and it would be indistinguishable except in terms of quality in some titles(fixed cameras, or controlled fixed panning cameras.). I would indeed consider it ideal for many games.

I was thinking Wardevil; maybe a new integration of HD-Film based backdrops, that aren’t so obvious (paste like look) with characters and objects moving about.

 

[Shifty Geezer]

Pre-'rendered' backdrops can be used ot great effect. See here at 1:12 in
[gt]28204[/gt]

 

["Nerve-Damage"]

^^ WOW for the Wii!!

 

[tongue_of_colicab]

Isnt one of the problems that its alot harder to make the enviroment interactive? Also, would it work well for most game? because you dont actually have a 3d enviroment anymore, just a static image. How would that work with a fps or a platform game?

Also for me it looks less realistic. The enviroment ofcourse looks great, but in the images from RE for example the characters stand out because they are far less detailed and make things look out of place. I'd rather have less detailed enviroments but fit in with everything else.

 

["Nerve-Damage"]

Isnt one of the problems that its alot harder to make the enviroment interactive? Also, would it work well for most game? because you dont actually have a 3d enviroment anymore, just a static image. How would that work with a fps or a platform game?

I wouldn’t say every game type (i.e. first person perspective), should use pre-rendered environments, but certain games types would benefit from it. Most sports games, some 3rd person, or anything that doesn’t require damaging buildings or surrounding, and that the camera is pretty much preset to a cinematic view.

Also for me it looks less realistic. The enviroment ofcourse looks great, but in the images from RE for example the characters stand out because they are far less detailed and make things look out of place. I'd rather have less detailed enviroments but fit in with everything else.

But that’s kind of the point…

Use all the saved polygons for character rendering and detailing. Imagine if a certain area or backdrop consisted of 4-6 million polygons a sec, rather than going that route use the poly budget for charater detailing to match the level of surrealism that pre-rendered environments graphics can achieve.

I’m telling you, play Resident Evil Zero… The lighting and shadowing alone made everything seem congealed.

 

[tongue_of_colicab]

I wouldn’t say every game type (i.e. first person perspective), should use pre-rendered environments, but certain games types would benefit from it. Most sports games, some 3rd person, or anything that doesn’t require damaging buildings or surrounding, and that the camera is pretty much preset to a cinematic view.

Even if you dont have damage to the enviroment how about things like enviromental effects like gas and fire?or shadows from the enviroment, those you want to change if you have moving things like gas, fire, lightsources etc. And things like shade and light would have to change with the player character etc. Wouldnt that be very hard to do with pre renderd images? I know there are some genres that could use it, like sports games. But that just makes me wonder why nobody useses them already in sports games if it really is a advantage.

But that’s kind of the point…

Use all the saved polygons for character rendering and detailing. Imagine if a certain area or backdrop consisted of 4-6 million polygons a sec, rather than going that route use the poly budget for charater detailing to match the level of surrealism that pre-rendered environments graphics can achieve.

I dont know, but isnt things like shaders alot more important for realism these days than pure polypushing power?

 

[....]

Isnt one of the problems that its alot harder to make the enviroment interactive? Also, would it work well for most game? because you dont actually have a 3d enviroment anymore, just a static image. How would that work with a fps or a platform game?

Also for me it looks less realistic. The enviroment ofcourse looks great, but in the images from RE for example the characters stand out because they are far less detailed and make things look out of place. I'd rather have less detailed enviroments but fit in with everything else.

THE resident evil titles are pretty good, though If you don't like how well it meshes, it could have to do with the gamecube's limited media size. Blu-ray and cell allow for cg prerendered environments of high quality, the xbox 360 and ps3 also have the ability to far further increase character detail.

Many games don't use interactive environments, and also it does not have to be a picture, it can be a video, as shown on wii on this very thread. On ps3 it could be an HD 1080p high quality video for panning, or to add fluidity to background elements.

We've never before had so much space for HD cg video, most of the time in the past size constraints have caused over compression or compromises in the looping of the cg videos.

Rpgs and many action games like DMC could make use of it, some sports games too.

In the present, I would believe aside from having far higher quality it would be indistinguishable from realtime environments when using fixed or fixed panning cameras. Shadows and lights can be calculated so they interact with characters, especially with all the freed up power, and this will allow them to mesh better and seem more cohesive.

 

[....]

Even if you dont have damage to the enviroment how about things like enviromental effects like gas and fire?or shadows from the enviroment, those you want to change if you have moving things like gas, fire, lightsources etc. And things like shade and light would have to change with the player character etc. Wouldnt that be very hard to do with pre renderd images? I know there are some genres that could use it, like sports games. But that just makes me wonder why nobody useses them already in sports games if it really is a advantage.

There are ways to allow fire, shadows and lights to interact, as well as interactive pieces of the environment*(many times interactive areas in realtime games are limited and precalculated.).

So such things can be handled, also remember we can have realtime elements along with the cg environments. Most games have many areas without movable lights, and the interactive elements in the environment are limited.(Those with highly interactive destructible environments tend to look very bad.)

I dont know, but isnt things like shaders alot more important for realism these days than pure polypushing power?

I believe it should free up a lot of memory and processing capacity on both the cpu and gpu. Allowing for more memory for character textures, polygons, more shaders, along with more AA.

 

[Npl]

I believe it should free up a lot of memory and processing capacity on both the cpu and gpu. Allowing for more memory for character textures, polygons, more shaders, along with more AA.

If youre talking about panning - then no. Having to cache sufficient frames to be able to smoothly move in both directions will quickly eat into memory. Mpeg4 is really only fitted to be played in one direction, if you loose that things get ugly fast.

 

[....]

If youre talking about panning - then no. Having to cache sufficient frames to be able to smoothly move in both directions will quickly eat into memory. Mpeg4 is really only fitted to be played in one direction, if you loose that things get ugly fast.

? You sure? I mean most dvds, and blu-rays can be easily rewinded while playing video without problem, and they don't seem to need 512MB of ram, 8 core cpus, and high-end gpus, to accomplish this as far as I know. A backwards panning on a fixed camera should be similar in terms of resources required, though it can probably be optimized more.

 

[Fafalada]

I believe it should free up a lot of memory and processing capacity on both the cpu and gpu.

It's true you get a fixed size memory footprint, but it's harder to say how much you can expect from CPU/GPU.

Dynamicaly lighting a 'prerendered' bgr is just a special case scenario of deferred shading (ie. attribute layers are prerendered, but you still shade as normal).
Last gen we only needed two attributes - RGB + Z(to handle occlusion and basic projected light/shadowing).
For a current gen app, you'd pretty much want all attributes that apply to standard deferred shading scenario (ie. see the numerous KZ2 and other threads on topic). Which means you aren't really saving in the shading pass, and in attribute pass you are trading off streaming+decompression costs for attribute rendering.

I suppose it could be worthwhile if you want backgrounds that greatly eclipse geometry complexity of foreground actors, assuming you can find compression algorithms that work with non RGB attributes (this could get really ugly if you're after doing stuff with moving cameras).
That said - it's a fair share of extra tech to work on - kind of turning concept of FMV game upside down (from what I understand, it's generally meant to simplify development/costs).

 

[Npl]

? You sure? I mean most dvds, and blu-rays can be easily rewinded while playing video without problem, and they don't seem to need 512MB of ram, 8 core cpus, and high-end gpus, to accomplish this as far as I know. A backwards panning on a fixed camera should be similar in terms of resources required, though it can probably be optimized more.

rewinding is not playing backwards. Mpeg4s efficiency is mainly because it compresses multiple frames together (thats a bit simplified). Say you have sequences of 10 frames, then to play backwards you`d have to either cache whole sequences or decode frame 1-10 and output 10, then decode frame 1-9 and output 9, then decode frame 1-8 and output 8....
The decoder itself cant move backwards in those sequences, and shortening those sequences would mean giving up Mpeg4s efficiency (most tools have a default value of <=300 frames, so 10 was already generous).

 

[....]

rewinding is not playing backwards. Mpeg4s efficiency is mainly because it compresses multiple frames together (thats a bit simplified). Say you have sequences of 10 frames, then to play backwards you`d have to either cache whole sequences or decode frame 1-10 and output 10, then decode frame 1-9 and output 9, then decode frame 1-8 and output 8....
The decoder itself cant move backwards in those sequences, and shortening those sequences would mean giving up Mpeg4s efficiency (most tools have a default value of <=300 frames, so 10 was already generous).

I've seen fluid video rewinding, iirc, especially in DVRs, basically on the fly. How would playing backwards differ from this? Even if it demands more resources, it seems rewinding can be done quite rapidly and smoothly on demand.

 

[Npl]

I've seen fluid video rewinding, iirc, especially in DVRs, basically on the fly. How would playing backwards differ from this? Even if it demands more resources, it seems rewinding can be done quite rapidly and smoothly on demand.

rewinding can get away by just displaying keyframes (the first frame in sequences). If you have a big solid RAR-Archive with many files and extract a single file (which happens to be at the end of the archive), you will see that the Programm has to decompress all preceeding files too - and therefore taking way longer than decompressing the first file in the archive. Its quite similar to Mpeg4 in that respect, getting the first frame is easy (which is usually done with rewinding, you dont need all frames for that), getting out the last frame will take way longer (decompressing the whole sequence).
Of course, if you use Mpeg2 on the DVR you dont have that problem.

 

[....]

rewinding can get away by just displaying keyframes (the first frame in sequences). If you have a big solid RAR-Archive with many files and extract a single file (which happens to be at the end of the archive), you will see that the Programm has to decompress all preceeding files too - and therefore taking way longer than decompressing the first file in the archive. Its quite similar to Mpeg4 in that respect, getting the first frame is easy (which is usually done with rewinding, you dont need all frames for that), getting out the last frame will take way longer (decompressing the whole sequence).
Of course, if you use Mpeg2 on the DVR you dont have that problem.

What about the myst series, there's user controlled panning of prerendered cg video loops there. I've not personally played the latest versions(at least not the non-real-time 3d more recent ones), but they did use that in the older ones I played. The quality can only have gotten higher[ in the more recent versions with prerendered cg video loops not the realtime ones], and they obviously wouldn't heavily tax a ps3's or xbox 360's hardware.

Superimpossing a few realtime 3d models with some creative lightning to give it cohesion, can't possibly be out of this world taxing on the hardware. It would require creativity to pull it off, but it seems viable, and could lead to spectacular results.

 

[jlippo]

Why wouldn't one just use prerendered images as a screen projected texture on a object?
It should make small movements possible and if merged several images/objects we could wider areas with visibility to behind objects.

 

[user542745831]

hello,

i was thinking about the same as the user which started this thread.

before making an own, i found this one.

i was thinking of the resident evil 1 remake for the nintendo gamecube.

especially games like resident evil 1 and 2 and final fantasy 7 are one of the best games ever made.

i think if they wouldn't have been done with preredendered backgrounds they wouldn't have been that good. alot of the atmosphere and good gamestyle comes from the very detailed and good looking prerendered backgrounds.

i was thinking about the comparison between the original resident evil 1 on psx and the resident evil remake 1 on gamecube.

the game looked very well on psx when it was released and the remake looked so much better on the gamecube and still looks very well. the prerendered backgrounds in the remake allready have been a bit more animated than in the original. one example would be the weather outside and the lightning effects from it in the main hall of the mansion at the beginning of the game. they lit up some parts of the hall. another would be moving candlelights. and so on.

i thought about how great games like that would look when thinking about a comparison between the gamecube and ps3 or xbox 360.

the detail and overall graphic would be great i think and it would be possible to render everything at 1920 x 1080 at 60 fps, wouldn't it?

furthermore i think it would look better than resident evil 4 and 5 how they are today, even if they aren't looking that bad.

gameplay and atmospherwise (in fact storywise, too ) the earlier resident evil games were better with prerendered backgrounds, fixed cameras and so on, too .

of course, as the thread starter mentioned, resident evil 0 was very good looking, too.

i think there would be alot people who would like to see new games with prerendered backgrounds and 1080p graphiocs with a similar or even bigger increase in overall graphics as resident evil 1 remake when compared to the original resident evil.

likely not every game genre would benefit from prerendered backgounds, but i think some genres allready have suffered from not beeing made with prerendered backgrounds , don't you think, too?

but, to state again, realtime rendering is likely better for many games.

en.wikipedia.org states, that the resident evil remake and resident evil zero on the nintendo gamecube are both based on the unreal engine 2, is that really correct?

would like to see more info on this thread. games with prerendered backgrouns would certainly look absolutely awesome on ps3 or xbox 360 .

the implemented videos within the prerendered backgrounds or something like that which someone mentioned here sounded interesting, too .

 

[Freak'n Big Panda]

FF8 benefited hugely from it's pre-rendered BGs. Even if you play that game today it doesn't look so bad because of how well the pre-rendered backgrounds were done.

However from what Fafalada said it seems that pre-rendered backgrounds aren't so simple anymore so that's probably one of the main reasons we're not seeing them in games as frequently as we once did.

I think folklore for PS3 might have used partially pre-rendered backgrounds. The camera couldn't pan when you were in the town or in houses etc. and the level of detail was quite impressive (although no doubt a lot more could have been done on that front).