Retro graphics are coming back in a big way, and it's surprisingly easy to do. Buy Me A Coffee: ko-fi.com/garbaj Discord: The discord has been shut down, sorry! Twitter: / garbaj2 Website: garbaj.net Github: github.com/GarbajYT
@@nicolaslindken9468 Yeah but it's a little more complicated than that, you I import my 3D model to unity, some faces disappear, some bones function wrong and so much more. I have no problem with creating a 3D model, nor with unity, but when these teo gotta br mixed shit always goes wrong
@@rodolf5941 Me neither, but if you put in your time and willingness, you can learn all these things. Especially with the amount of resources and tutorials available in today's internet.
Skeletal rigging wasn’t a thing on the ps1 or n64 until late in their lifespan. The polygons for arms and legs were separated due to the animations having to be done by rotating the polys by hand without rigging.
The Crash Bandicoot games used per-vertex animation, but using a custom compression algorithm to reduce the filesize of the animations to get it working on actual hardware.
Funny story, the devs at Valve (or maybe prior) during their pitch for Half-Life got turned down by a few studios because they wanted to try skeletal rigging. I'm trying to recall which interview it was, but nobody believed they could do it.
@@DesertPunks Skeletal rigging actually requires non-euclidian four dimensional math. Quaternions and matrices. I remember having to crash course that subject and I thought my brain was going to melt before I understood.
@@Fooxoul thats true. It used somehow pre-written lookup table with trigonometrical values. Some DOS games were using that stuff too, for people that didnt have Voodoo or such, back then. Funny times. Today engineering students learn about such stuff, and try to implement it by themselves, as a project for a subject. It used to be a real technology to earn money with, lol.
to be fair, most indies didn't even go for any kind of 3D graphics until relatively recently. For a long time, most indie games were 2D with very thick pixels for a "retro" look due to the limitations of the one-man-army making them, then people started slowly upgrading to PS1 and N64 looking simple 3D graphics, again because it's usually just one person working on the games who typically doesn't have as much of a background in graphics as they do in programming and game design
"It's pretty hard to look at" 1:58 - I think something that people miss about retrogames is that they would be played on those CRT televisions, which had the effect of "smoothing out" the blockiness and pixels, and so when you try to re-create the look, you need to cheat a little in order to recreate the "feel"
@@ZairotOne No, low poly modelling was never particularly hard, they had a hard time because they had sub-100Mhz processors and 4MB or fewer RAM to work with. The software engineering, level design, animations, characters, all had to be carefully designed so they hardware of the time could handle it. An universal engine like Unity was simply not possible; each game had to have a dedicated engine specifically designed for the game itself to extract the most performance.
@@jackmcslay The DS has pretty decent graphics, comparing it to the PS2's graphics, which came out 4 years earlier. The Nintendo 64 had graphics that look more like the ones in this video Edit: decemy is my new favorite word
Nice one! For a full PS1-like effect you should figure out how to implement affine texture mapping and how to disable the z-buffer depth, but I think you already knew that.
@@garbaj Also, by the way, the PS1 does support 640x480, it’s just locked to interlaced rendering. Not very many games used it for regular gameplay due to the hardware demands though, but some 480i titles did crop up later into the PS1’s lifespan. Tekken 3, Dead or Alive, and Ridge Racer Turbo Mode are a few examples. Most just stuck with 240p since it was easier to optimize a game’s assets for that resolution. Also, instead of skeletal animation, PS1 animators used different techniques. Namely, the Crash Bandicoot games preferred to manually animate by each vertex, but the animation data was compressed at an extreme ratio: something like 50,000:1.
Affine texturing can be done with just a shader, but the lack of z-buffer depth is not something that anyone is going to replicate. PS1 devs had to MANUALLY define the depths of vertices so that back parts of the model wouldn't show through to the front which created occasionally glitchy looking graphics because it wasn't perfect. No way an indie dev is going to put in that much effort for a small effect that most people wont even notice
The part I struggle with the most is the shapes, and getting that to feel authentic. A lot of PS1 games were still aiming for "realism" despite the limitations. And the main difference between low poly and PS1 aesthetic is that the PS1 used tris, not quads. Which gave them more geometry to work with, under the same vertex count. That's why stuff like the Harry Mason model in Silent Hill 1 looks super detailed even though it's super low poly. And that's... kinda hard to go for intentionally. Like, you're not aiming for stylized low poly. You're aiming for "how can I make this look as high fidelity as possible within these limitations?" Because that's what developers back then were aiming for. It's a weird balance.
Alternatively, there were games even at the time (Mega Man Legends comes to mind) that embraced the low-poly aesthetic and were stylized accordingly. So it all depends on what look you want to go for!
@@sebastianaliandkulche crash bandicoot did not look like a next generation videogame... you could literally count the polygons on his model. i think the ps1 game that pushed the ps1 to the limit in terms of 3d models from what i can think of the most is probably final fantasy 9.. look it up. Next closest i would say chrono cross.
@@RaimonTarou Heavily disagree. That could be said of any PS1 game, but Crash Bandicoot is one of the few fifth generation games with facial animation, something that not even FF9 have (and the places are just pictures and not 3d enviroments except the world map). And Crash 2 and 3 looks much better, not even the polygons are noticeable. I didnt play Chrono Cross, but i would check it out one of this days.
@@sebastianaliandkulche ok crash was one of the few ps1 games that actually animated the faces ill give you that but ff9 models still looked better overall. maybe it was the way they painted the textures that made them look way better. and yea im only talking about character models, not backgrounds.
I appreciate the video. While I never played the old playstation games, I am a big fan of minimalist or limited art styles. These limits often end up leaving you more creative, which is always good in my book.
@@dontkickmychick6076 I prefer the Midground PS2 look, IMO I think this is the best for game graphics, not too unrealistic, not too (sometimes lol the PS2 had gorgeous games) realistic
"The first thing I did was create a 3D model" oh silly fucking me I thought that might be part of "learning to make PS1 graphics" "I learned how to paint masterpiece paintings in under 24 hours. First, I sketched a photorealistic woman with a slight sly smile"
@@sagehanson190 I didn't say it was hard. I said I expected this video about how to make PS1 graphics to be about how to make PS1 graphics, not about how if you already know how to make PS1 graphics, you can make them look more accurate.
@@MateriaGirl So you expected a tutorial going over the basics of blender and low poly? I mean if you're interested in that stuff why not just go look up an actual tutorial? PS1 graphics seem very simple but they're not where you start, they're stylized and have a certain aesthetic as you probably saw in the video. You need a small bit of knowledge before you can move in that direction.
THIS IS EXACTLY WHAT I WAS LOOKING FOR, I’m learning to make animations and render visuals for my clothing brand and I’ve been seeing a lot of these really cool low res video game style animations and really wanted to do my own but didn’t know how, thank you for this existing on RU-vid because I don’t know how I found it
Very straight forward felt like this was perfect video wasn’t too long kept me engaged the whole video, the other ps2 style videos were more based on game building less how to style the models and this broke it all down in a way I can understand. I’m for sure hopping in blender and trying this out I didn’t know the textures were just super tiny that actually makes sense now. subbed❤️
If you wanna nail the ps1 aesthetic, there are a few more things you can do. 1#. Simulate a lack of a z-buffer. The z-buffer tells the render engine what polygons to draw first and what ones to draw last depending on their distance from the camera. Lots of PS1 games had issues with this since the console didn't have a native way of doing this. If you've ever played a PS1 game, you might see that some polygons that are obviously behind other ones get drawn in front due to small errors. 2#. Get a shader that uses affine texture mapping. This will cause the textures to warp in strange ways depending on how the in-game camera looks at them. Affine texture mapping simply linearly maps the texture onto polygons without any care for perspective. The PS1 used affine texture mapping, so if you want to mimic the PS1 aesthetic, it'll be one of the things you'll need to use, but another tip in this tip, if you're gonna use this, you'll need to subdivide your meshes. Devs back on the PS1 did this to lessen the effect of affine texture warping, so if you wanna hit the aesthetic on the head, it'll be useful to take advantage of that too. Just remember not to make your meshes too high poly. 3#. Integer vertex snapping. The ps1 couldn't use decimal values for vertex positions, so it had to snap vertexes to screenspace Integer values. This cause the vertex snapping that we all remember. 4#. Colour Depth and Dithering. The PS1 could show 24 bits worth of colour or 16,777,216 different colours, but this mode was basically only for still image display. It was more common for games to use a 15-bit colour mode for actual gameplay. This means most games only had access to 32,768 colours during gameplay. If you want to simulate PS1 graphics, that's one thing you'll have to do. However, with such a small amount of colours, colour banding was basically a guarantee. To help with this, dithering is used to break up the banding and give the illusion of more colour. Personally, I'd only recommend using dithering on 3D graphics, but if you wanna use them on 2D art too, you do you. 5#. Vertex shading. The PS1 couldn't quite handle per-pixel shading, so it had to survive with vertex shading. This causes shadows to be calculated per vertex instead of per pixel. No doubt I've missed something or I've gotten something blatantly wrong, so if you have any input or a correction, please tell me!
This was the perfect video I needed to complete my current 3D project. I'm actually building a replica of some of the cars from the PS1 game LEGO Racers. I'm definitely putting a filter on them now to give them a classic look.
Generally i don't like this kind of non video tutorials on this topics but this video actually made me to learn something and it was actually well made. Nice work garbaj
Christ, that Press Start takes me right back to playing the home console version of Radical Bikers. Still have that PS1, and it still runs perfectly despite being nearly... oh wow it's nearly thirty. Christ.
Great video. I suggest for an even more PSX look: convert the textures to TIM files and back before making the atlas. It will introduce compression artifacts that, I think, are essential to the look. Oh, and vertex shading solid colors... Keeps texture maps from looking like a carpet sample book, reduces number of texture maps needed, reduces UV work, etc.
You could also simplify the lighting too because most games of that era had either no lighting, baked lighting, or vertex shading where the vertex would be assigned a color depending on the position of the lightsource and the normals
Also used shadowed light source. Shadows had to be baked or faked. Also skeletal rigged the model when rigging wasn’t possible on the ps1 until very late in its lifespan (2000-2002) Not sure they had the right culling limit either.
This is something I can imagine a lot of people wanting to do and you did it with free software that anyone can download and use on their computer, so brilliant job!
Note for the Chads who want Saturn style visuals based on this video, instead of just PS1 or N64- Saturn was more color saturated than PS1 and also had no texture filtering, in addition, there would sometimes be flickering polys. 3D games on Saturn also had a more reflective look to their polys since the first party titles used what would later become Sonic Adventure's Lantern Engine lighting, palette based. Really good examples to reference are Burning Rangers and Panzer Dragoon Saga if you want to make something that looks nicer, or Tomb Raider and Sonic R if you're trying to go full Saturn faithful. Resolution for like 90% of games on the console was 320x240, but it also had a high res mode that a few games used at an absolute maximum of 704 X 480
I was about to cry when you said retro graphics. It seems that people have started considering PS1 era games as retro which means I’m now officially old.
joints were not connected in old 3D games because lack of proper skinning that would work in real time so it was easier to separate body into parts with pivots in joint spots and animate it like that...
For anyone wanting to do stuff that could run on the real deal: Most textures in PS1 games were about 64x64 and often did not have more than 16 colours each, the models also had to be carefully designed to avoid depth sorting issues as it didn't have a Z-buffer, triangles close together would flip back and forth which one was in front/behind. You could have about 3000 triangles on screen at one time. Per pixel lighting (like the circular lights you get in unity or unreal) did pretty much not exist and was far too expensive. Instead you tended to see vertex lighting either baked on a model or dynamic.
@1:51 The PS1 also supported hi-res modes. Wipeout 3: 720x480 (480i), and various other hi res modes like Ergheiz: 640x480, Tobal 1+2: 512x512 (which scaled to 640x480), Crash, Kain, Soulreaver, Quake2: 512x240, Tekken 3: 384x480 (also scaled to 640x480) The problem of course, these screen modes leave little texture space. Depending on the art direction, vertex colors alone would do the job...
2 pretty important things if you wanna make it look as convincing as possible: 1. make a shader that rounds vertex positions in view space. This gives models that strange wobbliness as the ps1 didn't have floating point values for those positions (pretty sure, don't quote me on that) 2. Use affine texture mapping and subdivide your models on large flat surfaces. This gives textures that strange warped look as the camera gets closer.
Cool, thanks for this. Honestly after growing up in the 80's and witnessing the 8bit wars and then the 16bit and into the 3d gaming world, not much has changed. For a while games were just being re created over and over with just better graphics. Today, the playing field is level and you have to make games with creativity, just showcasing what the GPU can do isn't enough anymore.
PS1 maximum resolution is actually 640x480 interlaced (in NTSC) but this was rarely used because the entire screen takes up part of the same video memory used to store textures, and there isn't much VRAM.
I'm not complaining. It is a tried and tested look. It allows one developer working on a project to focus more on actually creating the game and programming it rather than fiddling around for thousands of hours making photorealistic assets that can still look uncanny in the end. That way you can also spend more time on the gameplay features and art style instead and adding little things. Oh, and you can also run it on anything. So that is a bonus. And let's be real here, do most aspiring game developers who are just starting out with no programming experience really need those 8th gen AAA graphics to fulfill their vision in every scenario? Horror games especially come to mind here.
Dope! How long do you think it would take to recreate the last cutscene from the first kingdom hearts game by yourself? Those graphics blow my mind. And you should recreate the scene a ps1 version hahaha. Maybe you could make a series where you take certain cutscenes from games and recreate them as ps1 versions. I think it’d be dope
For PS1 look you'd need a texture filtering/distortion shader as well. I have it somewhere in my project since I also made a PS1 aesthetic prototype some time ago.
its suprisingly difficult once you try and do it yourself and try and make game ready assets and characters when there's no good tutorials explaining everything and you spend hours dive bombing down a youtube tutorial rabbit hole jus to learn the little nuggets of information you need that are being KEPT away from you
You might be able to achieve the PS1 sorting artifacts by marking all your geometry as transparent even though it isn't really, so it goes through a front to back sort with depth buffer disabled in your engine of choice.
it's only easy if it's not done right. it's more than just a low poly mesh with pixelated textures. most people doing 'ps1 style games' hat absolutely no idea what's important. vertex lighting instead of pixel lighting, NO real time shadows, correctly compressed textures (both in size and colors), no texture filtering, affine texture mapping, reduced float precision of vertices, no depth buffer, correct screen resolution and aspect ratio, no mipmapping (unless it's faked by LOD), fake CD loading times, draw distances, framerate. done right it's a really difficult task
textures are still filtered. fog should be dithered (i know it looks smooth on crt), add jittering due to lack of floating point unit and affine textures. can you emulate the software depth sorting in a simple shader??? or maybe fake it by rounding the depth value?
Some old games from PS2 era had special 16:9 mode that makes everything squashed. Then if you happen to have a 16:9 CRT or LCD the game looks nice and proportional. It works like anamorphic lens from 4:3 video camera and the final picture looks good on 16:9 TV Monitor or projector. The same way was working with funky resolutions that old consoles had but ratio standard was 4:3.