Who would like the CSG to be 2d or non-rotatable?

[Charles-cgr]

I think it depends entirely on the ability of the artists and the level of fidelity. Some people have better workflow with 2D others with 3D, what the team's artists are best with is what they should go for.

Once a model is created you can animate it any which way whereas a 2D asset needs to be redrawn.

This depends on approach. There's 2D animations that are done with rigging instead of sprite sheets.

Also you can get away with very little animation if you make the transition more dynamic by using effects and the camera, see Darkest Dungeon for example.

One of those things I knew without knowing. I just watched a video on Spine and it's interesting. Probably too late for current projects as you need to design the assets accordingly but I'll be looking further into rigging in the future.

 

[naossano]

Rotating camera can be nice if done well. Sadly, the last game in which i enjoyed rotating camera was released 18 years ago... (and it was more an RTS than an RPG)

 

[HoboForEternity]

2D really require more efforts imo. not only you have to draw everything ,every frames, every single animation from attacks to movement.

with 3D you only need a single model for a character, then you can you them anyway you can. i prefer 2D with quality sprites, but 3D is the safer choice these days

 

[Excidium II]

2D really require more efforts imo. not only you have to draw everything ,every frames, every single animation from attacks to movement.

It depends a lot of what kind of game you are making. Also again, a lot of 2D games are just pre-rendered 3D including favourites like Baldur's Gate and Fallout.

 

[Immortal]

2D really require more efforts imo. not only you have to draw everything ,every frames, every single animation from attacks to movement.

with 3D you only need a single model for a character, then you can you them anyway you can. i prefer 2D with quality sprites, but 3D is the safer choice these days

"You only need a single model for a character"..

Walking loops are no cake walk in 3D either though.
I would rather (pay someone to ) create 4 - 6 frames of a walk loop on a sprite sheet then animate / rig the skeleton for a model. Then properly animate them walking trying to make each joint appear natural.

You make it sound so easy.. in practice it's a little more effort.

 

[Mustawd]

You make it sound so easy.. in practice it's a little more effort.

I didn't intend to give that vibe off at all. Animating a 3D model is not easy. Neither is animating it. It's the same issue you face with 3D. In terms of realistic movement and expression. You can't learn those overnight. Regardless, of 2D or 3D.

However, 2D does become more difficult because as an artist you're forced to animate and create as you go. At least with the 3D model it already exists in space. Perspective is already a thing. Proportions are already a thing. With every frame of 2D this must be recreated again and again. Are there shortcuts? For sure. With photoshop you can cut down the time it takes to create each frame. So there's no need to start from 100% scratch if you already have the first frame.

However, what you can't shortcut is the knowledge of perspective or anatomy that goes with doing that. At least a 3D artist can animate the model without worrying about proportions, perspective, or lighting. Again, not saying animating a 3D model does not take skill. But it's a whole other ballgame that takes a ton of ingrained knowledge for 2D. I mean anatomy takes fucking years to be proficient at it.

 

[Excidium II]

You have a strange notion about 2D. You can have dynamic lighting with 2D too. After all the information of how things are affected by light is a 2D map.

Also you people seem to be talking at a level of expertise that is mostly irrelevant. All that talk about anatomy mastery. wut. Most people are just concerned that it doesn't look horribly wrong.

 

[Immortal]

You make it sound so easy.. in practice it's a little more effort.

I didn't intend to give that vibe off at all. Animating a 3D model is not easy. Neither is animating it. It's the same issue you face with 3D. In terms of realistic movement and expression. You can't learn those overnight. Regardless, of 2D or 3D.

However, 2D does become more difficult because as an artist you're forced to animate and create as you go. At least with the 3D model it already exists in space. Perspective is already a thing. Proportions are already a thing. With every frame of 2D this must be recreated again and again. Are there shortcuts? For sure. With photoshop you can cut down the time it takes to create each frame. So there's no need to start from 100% scratch if you already have the first frame.

However, what you can't shortcut is the knowledge of perspective or anatomy that goes with doing that. At least a 3D artist can animate the model without worrying about proportions, perspective, or lighting. Again, not saying animating a 3D model does not take skill. But it's a whole other ballgame that takes a ton of ingrained knowledge for 2D. I mean anatomy takes fucking years to be proficient at it.

There is a best of both words.. Many games from the late 90's early 2000's would use 3D Models to generate their sprite maps.

One example example is Nox by Westwood Stuidos.
The entire game is made up of 2D objects with "cylinder hit boxes" but originally was created with 3D objects that you could use to generate your sprite sheets.

It means you can get the fluid / lower requirements of 2D rendering without needed to redraw from scratch. It works very well for Isometric.

Super Mario RPG also did this -I think-

Spoiler

You have a strange notion about 2D. You can have dynamic lighting with 2D too. After all the information of how things are affected by light is a 2D map.

Also you people seem to be talking at a level of expertise that is mostly irrelevant. All that talk about anatomy mastery. wut. Most people are just concerned that it doesn't look horribly wrong.

Making animations that "look off" is extremely distracting for your game.

This is basically Uncanny-Valley at work. It's a real issue game devs have to consider. Which is why most indie games go for cartoonish aesthetics or avoid close ups of humans. (Or use cheap unity bought assets if they need to)

 

[Excidium II]

Super Mario RPG also did this -I think-

So did almost every 2D game when computers were fast enough to render decent 3D models.

'sup

Spoiler

 

[Immortal]

Super Mario RPG also did this -I think-

So did almost every 2D game when computers were fast enough to render decent 3D models.

Most pixel artwork doesn't start as 3D models for ~aesthetic~ games. I think it depends on the game / view port / art style.

It would probably be the way I go though. Since I can't do art for shit.

 

[Excidium II]

Pixel art is its own thing, working with a set of self-imposed limitations. 'state-of-the-art' 2D is another story, it's just 3D with rendering time being the only limitation.

Interesting post: http://polycount.com/discussion/166800/pillars-of-eternity-the-white-march-art-dump

The workflow that tended to work best was often to throw something in zBrush for 30 minutes or so, decimate it to a reasonable polycount, and then toss a tiling texture on it. An example is the skeleton in the first image. I think the skull is somewhere around 100k triangles and then I just blended an ice texture and an oxidized bronze (although that piece had a fair bit more time in zBrush than just half an hour). There wasn't a lot of time to put things into a 'proper' workflow. I spent a lot of time making materials that would hold up well at our camera angle and distance. Things like a refractive ice shader or a material that projects snow on the top faces of surfaces came in really handy.

One of the nice things about this being pre-rendered was that I could often just throw polygons at it until I got a result that had the requisite level of detail and then just let material definition take care of the rest. Going back to that first image, the walls of the abbey are all brick meshes. They show off the snow shader well and it's easy to take a few out to make room for additional geometry (windows, doors, whatever). Since we only ever see one camera angle, we can just bash together pieces that are hidden from view.
We do use some more modular approaches in places, notably interiors. This only gets us so far though because the designers tend to do things... unconventionally and with little regard to standard sizes in their layouts. I end up making some kind of pillar or edge piece for each tileset that I make just so I'm covered in case of weird designer shenanigans and hide any unfortunate seams.

Each scene tends to be made using wildly different techniques so if you have any specific questions I'd be happy to answer them.

So, we didn't have any real hard poly limits, but we did have to work within the limitations of our own machines. It wasn't until relatively recently that we had a distributed rendering system up and running, so render times were a concern. There were times where it became a bottleneck in the process because we just couldn't get our renders out fast enough. Poly count didn't have a super huge impact in comparison to things like lighting and reflection, but it was definitely something to keep in mind. We also had to work with lots of layers because with large scenes we would get significant viewport lag.

The big dwarven fortress scene in the first expansion clocked in at around 200 million triangles before its revision pass (which added a fair bit of geometry in addition to what was there). We have to keep things in layers and hide and unhide as we go otherwise it's just impossible to work in the scenes. That aside, yes, we weren't overly concerned with poly count or texture resolution and given our timeframe, even if we had wanted to make things 'efficient' we wouldn't have been able to fit it in the schedule.

As for interior areas, in the base game our lead environment artist, Hector Espinoza, came up with a method of lighting where we would just use non-rendering planes with emissive materials. These were actually much faster to render than real Maya lights. They don't cast shadows though, so we can't really use them for key lighting, at least not by themselves. We would use them as ambient lights in an area or to achieve lighting effects that wouldn't actually happen in real life. An example of that is that we would make a horseshoe shaped light plane around doorways to give them a highlight and make them stand out from the room behind them. We used these in conjunction with point lights for interiors.

We render out a number of passes with our master beauty shot including a world position pass. We use these passes to properly occlude characters and even allow us to have moving dynamic lights affecting the pre-rendered background.

So, everything dynamic is an actual 3d mesh in Unity, the characters, effects, doors, etc. The background plane with the rendered image is just sitting behind them and their materials are set up to respect the world position map that we rendered out. This way the engine knows where every pixel is in that background image relative to everything in the scene and sorts them accordingly. All of it is just programming witchcraft combined with our render passes.

We render shadows in the static passes, but dynamic objects cast their own shadows as well. We have directional lights in the scene mimicking the angle of the sunlight or (in interiors) set to something more neutral that lights the characters well. The shadows obey the same world position data that all the dynamic objects do.

 

[Mustawd]

You have a strange notion about 2D. You can have dynamic lighting with 2D too. After all the information of how things are affected by light is a 2D map.

Sure. Does the layman know how light works? I doubt it.

Also you people seem to be talking at a level of expertise that is mostly irrelevant. All that talk about anatomy mastery. wut. Most people are just concerned that it doesn't look horribly wrong.

Wut? This coming from a person who was graphic whoring for three pages about how Tyranny environments looked crappy?

I mean that's a perfect example. You need good art direction and skills to be able to say, "That tree looks like it's just sitting in a brown marshmallow".

I mean if it was so easy people would just do it and make games. 2D in a very linear style, meaning animish or flat, is a lot easier to do because it's a simplification. So you do see some of those that are just created by one or two kind of artistic people. But what about real environments? Realistic looking animations? Those do take expertise. and it's not like you can gain that expertise in a year or two. It can take 3-4 years of work to get to being average or ok.

 

[Immortal]

Pixel art is its own thing, working with a set of self-imposed limitations. 'state-of-the-art' 2D is another story, it's just 3D with rendering time being the only limitation.

That's fair. When I say pixel art.. I don't just mean 8-Bit.. Though

Most if not all 2D side scrollers are just animated sprites.. Even recently released ones.

Usually you only get your moneys worth from Modelling out sprites when your using a isometric view.. Especially if your walk animation along the Z axis is different from the animation along the X axis..

IE: Your back / front turns to the screen.. as opposed to doing some kind of goofy side walk along the Z Axis.

EDIT: I guess it would be the y axis since it's a 2D game.. but you're giving the illusion of depth. So I used Z


EDIT EDIT: I would also argue that games like Darkest Dungeon are ~state-of-the-art~ 2D.. if you consider the aesthetic they were going for.. I can almost guarantee those were not 3D models first.

INB4 Celerity.

 

[Mustawd]

Pixel art is its own thing, working with a set of self-imposed limitations. 'state-of-the-art' 2D is another story, it's just 3D with rendering time being the only limitation.

If your example is so simple, then I guess we'll be seeing all kinds of cool 2D rpgs then?

 

[Excidium II]

Immortal
Fair enough, of course there were lots of full 2D titles deeper into the 90s, specially arcade ones. Had mostly PC games in my mind I guess.

EDIT EDIT: I would also argue that games like Darkest Dungeon are ~state-of-the-art~ 2D.. if you consider the aesthetic they were going for.. I can almost guarantee those were not 3D models first.

Wouldn't call it state of the art since it's mostly static assets. They make good use of effects and camera to make it seem more dynamic. Still, I think that is a good approach.

Pixel art is its own thing, working with a set of self-imposed limitations. 'state-of-the-art' 2D is another story, it's just 3D with rendering time being the only limitation.

If your example is so simple, then I guess we'll be seeing all kinds of cool 2D rpgs then?

Right as we start seeing all kinds of cool rpgs.

 

[Mustawd]

COme to think about it, I might be overstating it from a 2D sprite perspective. Yes, having a real artist would make it highly polished, but probably the trick above could work. But you still need portraits and environments. Which I assume are the more difficult portions.

I probably got on my soapbox about anatomy and animation as I was studying it a lot myself.

 

[Excidium II]

Static portraits aren't a problem, enviroment will depend on the level of fidelity. I think full 2D enviroments are a terrible idea unless it's tiled.

 

[Lurker King]

Hey, Vault Dweller, this is another example of the advantages of ignoring rotatable camera.

 

[naossano]

I like rotating camera when it is well done. Unfortunately, for 95% of the games in played that had rotative camera i consider that i was badly done, getting in the way of the gameplay. So fixed camera is the safe approach for me. It can't disapoint.

 

[Shadenuat]

I would only ask for fixed camera pls.

 

[Lurker King]

Sidescrollers can look better than most cRPG nowadays.



Just saying.

 

[Zer0wing]

I like to know what's going on in combat and align the necessary camera angle myself.

Usability > Graphics.

 

[cherry blossom]

Beyond the question of quality and polish between 2D and 3D, an interaction model that is designed to be as hassle free as possible, in a way that doesn't require the player to wrestle with camera angles often, is paramount to gameplay, wouldn't you agree Vault Dweller? Which of your favourite isometric cRPGs had you mingling the camera all the time? Fallout 1&2, PST, Arcanum, ToEE; apart from being 2D, all provides you with a simple and effective mode of interaction: Everything you need to see in the game world, in order to effectively interact with it, is in your view at all times. You only pan the camera as you move around.

Age of Decadence suffers from the environments designed around the necessity to rotate the camera.

 

[Zer0wing]

Beyond the question of quality and polish between 2D and 3D, an interaction model that is designed to be as hassle free as possible, in a way that doesn't require the player to wrestle with camera angles often, is paramount to gameplay, wouldn't you agree Vault Dweller? Which of your favourite isometric cRPGs had you mingling the camera all the time? Fallout 1&2, PST, Arcanum, ToEE; apart from being 2D, all provides you with a simple and effective mode of interaction: Everything you need to see in the game world, in order to effectively interact with it, is in your view at all times. You only pan the camera as you move around.

That's a stockholm syndrome right here.
Nope, not always. Objects in these games, at least in Fallout, Arcanum and PS:T tend to overlap each other, creating various problem of collecting loot and finding and actually ending fleeing trash mobs, most of which would be fixed with rotatable camera.

 

[Lurker King]

That's a stockholm syndrome right here.
Nope, not always. Objects in these games, at least in Fallout, Arcanum and PS:T tend to overlap each other, creating various problem of collecting loot and finding and actually ending fleeing trash mobs, most of which would be fixed with rotatable camera.

Which could be easily fixable without making the game 3D.