-
Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.
Do the modern consoles have infinity power for 2d graphics?
Wildebeest
Member
Technically, they have zero power for 2d graphics since people just use 3D scenes and don't use custom 2D graphics chips.
ZoukGalaxy
Member
No, only the emperor have infinite power.
diffusionx
Gold Member
Those old 16-bit systems had graphics chips that could draw a specific, hard limit of sprite and background information that you held in the VRAM along with a specified color palette (NES - 8 sprites to a line, etc.). After that systems wrote to a buffer which held the data on a frame by frame basis. and you are limited by what you can process and write to the buffer. So that limitation still remains.
S0ULZB0URNE
Member
Good question.
I think the best modern use I have witnessed is Dragon's Crown.
I think the best modern use I have witnessed is Dragon's Crown.
DeltaPolarBear
Member
We haven't hit the limits for graphics only because the practical physical limits of what artists can produce is hit first.
Back in the day, artists draw what they want and then try to squeeze what they can into the software. Now, you can throw whatever you wanted into the game and the game would just demand more.
That is the main reason game development ballooned. Historically you stop drawing art assets when the game can't fit any more in. And then you focus on fine tuning gameplay and debugging. But now art assets don't have a ceiling anymore. And you end up with billions spent on one game, not knowing when to stop.
Back in the day, artists draw what they want and then try to squeeze what they can into the software. Now, you can throw whatever you wanted into the game and the game would just demand more.
That is the main reason game development ballooned. Historically you stop drawing art assets when the game can't fit any more in. And then you focus on fine tuning gameplay and debugging. But now art assets don't have a ceiling anymore. And you end up with billions spent on one game, not knowing when to stop.
Heimdall_Xtreme
Hermen Hulst Fanclub's #1 Member
Modern consoles do not have "infinite" power for 2D graphics, but they are extraordinarily capable of handling 2D rendering tasks with ease due to their immense processing power compared to the demands of 2D graphics.
Why 2D Graphics Are Easier for Modern Consoles:
1. Efficient Rendering: Rendering 2D graphics, even with advanced effects like parallax scrolling, particle effects, and high-resolution textures, is far less demanding than rendering complex 3D models with physics, lighting, and shaders.
2. Optimized Hardware: Consoles like the PlayStation 5, Xbox Series X, and even the Nintendo Switch are designed to handle highly demanding 3D games, so they can easily process 2D games at high resolutions (e.g., 4K) and smooth frame rates (e.g., 60fps or 120fps).
3. GPU Power: Modern GPUs can perform billions of calculations per second. Drawing sprites, tiles, and effects in 2D games doesn't come close to pushing their limits.
Are There Limits?
While consoles have vast resources for 2D, they aren't "infinite," and constraints still exist:
High Complexity Effects: If a 2D game adds heavy post-processing effects (e.g., dynamic lighting, shaders, or massive particle systems), it can still tax the system, especially at high resolutions.
CPU Usage: A poorly optimized game with heavy AI, physics, or systems running in the background can bottleneck the performance, even in 2D games.
Developer Optimization: No matter how powerful the console is, poor optimization in game design or programming can lead to performance issues.
Practical Reality
For all practical purposes, though, modern consoles can render even the most visually intense 2D games without breaking a sweat. Developers are generally more limited by artistic design choices or budget than by the hardware's ability to handle 2D graphics.
Why 2D Graphics Are Easier for Modern Consoles:
1. Efficient Rendering: Rendering 2D graphics, even with advanced effects like parallax scrolling, particle effects, and high-resolution textures, is far less demanding than rendering complex 3D models with physics, lighting, and shaders.
2. Optimized Hardware: Consoles like the PlayStation 5, Xbox Series X, and even the Nintendo Switch are designed to handle highly demanding 3D games, so they can easily process 2D games at high resolutions (e.g., 4K) and smooth frame rates (e.g., 60fps or 120fps).
3. GPU Power: Modern GPUs can perform billions of calculations per second. Drawing sprites, tiles, and effects in 2D games doesn't come close to pushing their limits.
Are There Limits?
While consoles have vast resources for 2D, they aren't "infinite," and constraints still exist:
High Complexity Effects: If a 2D game adds heavy post-processing effects (e.g., dynamic lighting, shaders, or massive particle systems), it can still tax the system, especially at high resolutions.
CPU Usage: A poorly optimized game with heavy AI, physics, or systems running in the background can bottleneck the performance, even in 2D games.
Developer Optimization: No matter how powerful the console is, poor optimization in game design or programming can lead to performance issues.
Practical Reality
For all practical purposes, though, modern consoles can render even the most visually intense 2D games without breaking a sweat. Developers are generally more limited by artistic design choices or budget than by the hardware's ability to handle 2D graphics.
ZehDon
Member
Most modern game engines don't really do 2D rendering akin to the 16-bit 32-bit systems. Instead, they render 3D quads - two polygons layered together to create a rectangular shape - in a flat scene with the sprite simply stuck on it as a texture. Because of this, consoles don't really support 2D rendering in the same way, and so there are practical limitations to what can realistically be done. The biggest thing you're likely to run into is overdraw due to overlapping sprite work with particle systems causing frame time issues - even on modern consoles - because it's essentially throwing around massive amounts of unculled polygons with enormous alpha work due to the way it handles the sprite texturing for the quads. With that said, the limitations aren't something you'd run into if you're trying to emulate the 16-bit or even 32-bit 2D styles - they simply don't try and put enough crap on the screen to be an issue.
kevboard
Member
yes... technically, but a bad dev could still make them struggle through bad optimisation.
see the launch version of Ori 2, which does use 90% 2D graphics, yet couldn't reach 60fps on a One X. a few months later it ran at a perfect 60 on the Switch and was patched on Xbox and PC to also finally run perfectly.
see the launch version of Ori 2, which does use 90% 2D graphics, yet couldn't reach 60fps on a One X. a few months later it ran at a perfect 60 on the Switch and was patched on Xbox and PC to also finally run perfectly.
Papa_Wisdom
Member
No idea but I’m currently enjoying phantom breaker battlegrounds atm
LordOfChaos
Member
ChatGPT?
You can tell by it unrolling words for maximum word count like a middle schoolers essay lol
Heimdall_Xtreme
Hermen Hulst Fanclub's #1 Member
Hoddi
Member
Kinda. It's not 'free' in the sense that there are no limits but it's 'free' in the sense that it's not a concern at 4k and below.
A 4k screen only needs 32MB of pixels at 32bpp on a single layer. Even if your 2D game uses 10 layers at full resolution then that's still only 320MB of raw bitmap data. Contrast this with a modern 3D game that uses thousands of textures with many of them larger than your entire 4k screen and the requirements are not even in the same ballpark.
A 4k screen only needs 32MB of pixels at 32bpp on a single layer. Even if your 2D game uses 10 layers at full resolution then that's still only 320MB of raw bitmap data. Contrast this with a modern 3D game that uses thousands of textures with many of them larger than your entire 4k screen and the requirements are not even in the same ballpark.
Madserb2023
Member
Nothing is free. You are still limited by sheer CPU calculations. Put 10,000 moving interactive sprites on the screen and see most modern CPUs struggle. You could probably get Doon to run at 3 fps if you put enough on the screen.
Parazels
Member
Maybe. But I meant rather quality of 2d animation, than 10000 moving objects.
Can developers add as many animation frames as possible on the modern hardware?
Well, technically it's not quite like that, game looks 2D, but in fact it is 2.5D. It uses a huge number of 3D renders, converted into 2D graphics (including the characters and their animations). If you're interested, you can watch it here, they told how the game was made. Nowadays, quite a lot of 2D games are made this way, it’s easier than drawing animations frame by frame, and it significantly reduces development time.
Last edited:
What do you mean by "way inefficient"? could you please tell us more. I can ensure you that today gpus are much more capable. Back then, ram memory was too expensive and too slow to implement a framebuffer for 60 fps. To overcome that, video game consoles invented tiles, sprites, and color palettes to generate a scan-line on the fly when requested by the monitor. But those have their own limitations: the tilemaps (background) and the sprites have to use the same tiles. a tile has a fixed size and you had a small amount of memory to store those... which is why the graphics of old 2d games are pattern based (same tiles being repeated again and again). Today gpu aren't using these anymore, they are bliting machines with enough memory to hold multiple framebuffers for very high resolutions. Today gpu are able to blit thousands of bitmaps at hundred of fps (You don't need to use polygons btw, both OpenGL and Vulkan have blit procedures but using polygons give you usefull 3d transformations).
amigastar
Member
I miss the days where animations where drawn frame by frame (Symphony of the night, Street fighter 3rd strike etc.)
Parazels
Member
Isn't impossible to create a program (or AI), that could generate those frames?
Nowadays, it's rare for anyone to do this, it's too long and expensive. Usually, they use 3D to 2D conversion, or there's a more complicated method - creating skeletal animation, when the character is drawn in separate parts and animated using a skeleton like a puppet (Vanillaware - they really like this technique). Now there are very few studios, to the old school of 2D games, well, perhaps the most famous is again Vanillaware, they use 3D objects to create backgrounds and fill locations, but all the main art of the game plus characters is handmade by artists. But because of this, their games are released very infrequently, usually every 3-5 years.
Well, and perhaps the most striking example when such commitment to the old school can kill a studio if you are not careful is SNK, they already had problems with finances, and they first made KOF12, which turned out to be unsuccessful, but showed what their artists are capable of, and they finished KOF13 - which is probably still the most beautiful 2D fighting game, drawn in 2D. But the game, due to the huge costs of 2D graphics, never paid off, which drove SNK into a tailspin from which they have only recently begun to somehow get out =)
RoboFu
One of the green rats
You basically just argued why old consoles were way more efficient. lol Those tiles sprite and palettes were very efficient.
To even mimic palletized graphics on a modern gpu you have to use shaders that check and change each pixel and at least two triangle polygons and one texture with a resolution in the power of 2 for each sprite. You need an entire graphics / input api loaded into an os.
So it takes way more resources today to display one sprite than it did to run an entire nes game. A nes / snes run at 17 watts.
Try to run even a nes emulator on a system with the wattage lowered to 17 watts.
Modern consoles are orders of magnitude more powerful than 2D consoles, so technically, yes, they could render almost infinite 2D graphics. However, their architecture is also totally different. Older consoles were slow or unable to draw every pixel of the screen, so they relied on raster rendering to draw things faster, basically racing the beam to draw things. These consoles had a hard limit on how many things (like sprites) they could draw every frame, and because they were racing the beam, you couldn't delay a frame to draw more things if needed.
On the other hand, modern consoles draw things on the screen all at once from a frame buffer in the VRAM. Their limit is basically how fast the GPU can draw to that frame buffer in a given period of time. If you need more time, you can delay when to draw that frame. However, for modern GPUs, the frame buffer is so small in terms of pixel count that they can draw almost infinite sprites in one frame. The problem is that no developer is asking the GPU to draw things directly to the frame buffer; they use engines, and these engines add a lot of overhead to the GPU, such as using polygons to draw sprites or calculating every object collision every frame. So, even if the machine is powerful, a bad engine or developer can still mess up a 2D game.
On the other hand, modern consoles draw things on the screen all at once from a frame buffer in the VRAM. Their limit is basically how fast the GPU can draw to that frame buffer in a given period of time. If you need more time, you can delay when to draw that frame. However, for modern GPUs, the frame buffer is so small in terms of pixel count that they can draw almost infinite sprites in one frame. The problem is that no developer is asking the GPU to draw things directly to the frame buffer; they use engines, and these engines add a lot of overhead to the GPU, such as using polygons to draw sprites or calculating every object collision every frame. So, even if the machine is powerful, a bad engine or developer can still mess up a 2D game.
Last edited:
Holammer
Member
We've had order of magnitude more powerful 2d graphics since the PS2 compared with old 8 & 16-bit systems.
If you have enough CPU you can do anything, it's been true since the 8-bit days. The spectrum despite harsh limitations still managed to have games with rudimentary music, sound effects, scrolling and sprites by throwing its 3.5 MHz CPU at it.
If you have enough CPU you can do anything, it's been true since the 8-bit days. The spectrum despite harsh limitations still managed to have games with rudimentary music, sound effects, scrolling and sprites by throwing its 3.5 MHz CPU at it.
Why would you want to mimic palletized graphics when today gpu are able to do more and do it faster? color palette graphics is an inferior technique. But if you want that, you can implement it with a ten lines shader. Shaders are the best and most efficient thing that happened to computer graphics. You neither need polygons nor power-of-two textures to do 2d graphics; you can blit a bitmap to a framebuffer with both Vulkan or Opengl. The super nes has great limitations: a tile size is 8x8, a tilemap (background) 32x32 tiles, 128 sprites at most, 32 sprites on the same scanline, 60 fps at most, one video resolution (the others are buggy), ... An apple mac mini with the cpu at %100 is consuming ~20 watts at the plug (entire system), you could probably emulate a super nes while staying under ... ten watts!You basically just argued why old consoles were way more efficient. lol Those tiles sprite and palettes were very efficient.
To even mimic palletized graphics on a modern gpu you have to use shaders that check and change each pixel and at least two triangle polygons and one texture with a resolution in the power of 2 for each sprite. You need an entire graphics / input api loaded into an os.
So it takes way more resources today to display one sprite than it did to run an entire nes game. A nes / snes run at 17 watts.
Try to run even a nes emulator on a system with the wattage lowered to 17 watts.
cartman414
Member
Forgive the necrobump, but this is one type of conversation I've been looking for.
Haha.
The PS1 and N64, the first two major consoles to make a clean break from the sprite paradigm of graphical processing, lagged significantly behind the Saturn in 2d graphics.
Aside from the Cave CV1000 I don't think there's been anything since that had sprite/tile specific hardware functions.
Haha.
The PS1 and N64, the first two major consoles to make a clean break from the sprite paradigm of graphical processing, lagged significantly behind the Saturn in 2d graphics.
Aside from the Cave CV1000 I don't think there's been anything since that had sprite/tile specific hardware functions.
You are pretty much spot on, everything now would be done via software or like the PS1 where 2D were often 1 sided 3D polygons. The Cave board might just be the last hardware example. Not too sure on the PGM2 and it's chips.
I mean these new consoles with all that RAM and power can do a lot and get close. But no matter what, there is still loading times and limitiations. No specific hardware for scaling, rotating or tile based graphics. Nothing similar to Saturn with the VDP1 and 2. 2D efforts on N64 were pretty limited, but it could fake it.
To me, nothing beats the Saturn with 4MB ram expansion for CD 2D consoles games. Just shows the importance of RAM.
But cartrifge based systems with special chips focused on sprites, still more capable. Just need massive cartridges.
Parazels
Member
What is this statement based on?
Where are these "giant" cartridges with the best 2d games in the Universe?
Last edited:
I would think Neo Geo and some modern Genesis games showed what can be done with large carts and bank switching, almost unlimited animation, sprites and instant loading of new assets. Maybe it's ancient techs pecifically built for tiles and sprites, but I think the theory is sound. Even recent 2D efforts on consoles require loading of levels and creens. Maybe I wrong, entirely possible.
a'la mode
Member
This is absolutely nonsensical.
Nothing is done in software these days; the GPU handles transformations, all hardware accelerated. There are no limitations. Saturn's edge at the time was that it was optimized for quads, which 2D sprites generally are, but a quad for any other system is really just two triangles. There is absolutely no point today to that kind of architecture.
Cartridge-based games loaded fast partially because games were very small back then and it was just ROM chips on a board, literal extensions to the console directly accessible on the address bus. That had nothing to do with sprites or tiles, and they're not some magical devices that have zero access times.
Putting a modern game on ROM chips would be extremely cost prohibitive, and really, really slow.
Last edited:
Hudo
Gold Member
It really depends on the implementation.
polybius80
Member
but what means having no limitation for 2d graphics? a sprite as big as the screen? having a sprite for each pixel?
Last edited:
Spiral1407
Member
What about phones? I can run a Nes emulator with full CRT shaders on and still use under 3w.You basically just argued why old consoles were way more efficient. lol Those tiles sprite and palettes were very efficient.
To even mimic palletized graphics on a modern gpu you have to use shaders that check and change each pixel and at least two triangle polygons and one texture with a resolution in the power of 2 for each sprite. You need an entire graphics / input api loaded into an os.
So it takes way more resources today to display one sprite than it did to run an entire nes game. A nes / snes run at 17 watts.
Try to run even a nes emulator on a system with the wattage lowered to 17 watts.
I dunno man. It runs fine on my laptop's shitty 4-core CPU + iGPU. Now granted, I haven't been able to get past the 4th level yet, but how much worse could it really get?
Last edited:
a'la mode
Member
It's just an another clueless argument that makes no sense. Whooptydoo, a NES emulator under 17 watts? That's crazy talk! But if you look at power consumption vs capability as a metric of efficiency, a RPI would consume less than 17 watts and emulate up to a PS1 well.
GymWolf
Member
Much worse, i had my 13600k slowing down a little in extreme casesWhat about phones? I can run a Nes emulator with full CRT shaders on and still use under 3w.
I dunno man. It runs fine on my laptop's shitty 4-core CPU + iGPU. Now granted, I haven't been able to get past the 4th level yet, but how much worse could it really get?
It was years ago so maybe they optimized a bit, the game has by far the most complex physics system in gaming.
GigaBowser
The bear of bad news
infinirty?? NOTHIN IS INFINITY
Hyet
Member
That's a fast turnaroud in today's climate, actually. Releasing more that 1 game a generation is quite a feat. I've been able 2 play 2 excellent very different games on my switch and I didn't ever bought everytrhing they released. I think theres a good oportunity for studios like Vanilla to develop a good 2D pipeline and have fairly consistent releases. But I don't think its easy to find people to fill such a studio.