"Dear Esther" design elements

[Holy_Diver]

It's just a matter of taste so it seems rather useless to say it's right or wrong. Obviously, if someone disagrees, they are free to use perfectly straight or parallel lines as much as they want. Pictures say as much as need be said as to the benefit of breaking up straight lines.

It's good to have that on the record. That said I don't see how the pictures would be different with straight lines. And you don't specify, are you saying curves? Odd lines? And what are the benefits? Are they technical? Etc. There is a lot to unpack. Here you just seem to be saying you personally prefer curves? Or am I wrong? Because you like women with curves? Either way curves always demand more polygons.

A modern city is almost all straight lines. The way this is presented is as if to say the image will appear more realistic or less artificial if somehow straight lines are used sparingly.

But I will mention that it's completely untrue that "ancient builders were far more precise than modern builders". Have you ever looked in a house from the 1500's or down an ancient street? Maybe you're thinking of the pyramids, Parthenon or such, but obviously, they are not the norm.

I am not sure the pyramids is a good example. But I am thinking of temple complexes, since that makes up the bulk of SOM's repertoire and is typical of a KF game. And its not hard to make a straight box for example. And it would be difficult to arrange boxes that are not straight. I just want to be sure amateur artists are not getting wrong ideas.

Even in modern times, things are not perfectly straight. Look at your roofline or a fence some time. The eye expects sagging, denting, bends. Perfect straightness looks artificial; that's why the whole technology of normal mapping was developed; that's why modern 3D games use high-poly models and have face textures showing pores and blemishes. If you think perfectly straight parallel lines look better, go for it. But I'd say you're in the minority.

I don't think its given that much thought. A sag can be very expensive to implement. You seem to just be suggesting that ramshackle buildings should not be straight. A better way to say this might be SOM could use more ramshackle tile sets.


PS: If you are not talking curves. As soon as the camera is off by even a bit most straight lines will no longer appear straight. You could force the camera to be off by not letting it ever be perfectly centered on the vertical axis. The odds of the player lining themselves up exactly on the horizontal plane is pretty slim unless they work at it. That said I was wondering if your problem was just that your prefer aliasing in the lines, since whenever things line up and the aliasing goes away, the line appears too perfect for a video game. Either way, enough antialiasing can pretty much make this a non issue.

EDITED: I've even thought about letting authors redefine the center of the look up/down cone. Since most people don't normally look straight forward unless their attention is on high alert. Or even changing the center slightly relative to the ground incline and where open space is, so you would naturally look up when climbing a hill automatically. But since you mention it this could be a good way to not let the player look straight forward so easily so that vertical lines would not be so vertical anymore. Maybe that would appear less artificial to you?

[HwitVlf]

This post is just a collection of my thoughts about what made Dear Esther especially pleasing visually and immersive so I don't know why anything needs to be "on record". Why are you talking about modern cities that have mostly straight lines? There's not a modern city in Dear Esther. And didn't you just say a couple posts up that ancient buildings had far more straight lines than modern ones? Modern cities have straight-ish lines, but if you look closely, you'll see dented trash cans, warped window frames, crooked fences, uneven cement blocks etc. Those imperfections are the difference between a "perfectly straight" artificial feeling environment and a realistically skewed immersive one.

Even on models of things like stacked crates, I wold make slight variations in shape. Obviously, as seem in the pictures posted earlier, adding irregularities to break straight lines can be done without being "very expensive to implement".

[Holy_Diver]

I was playing the game "Dear Esther" and I must say it's one of the best looking games of all time in terms of a realistic feeling, immersive environment. So I thought it would be interesting to look at some of the deign elements the game uses.

Call me crazy but you seem to be saying that this game is exemplar, and by extension SOM games would be better if they followed its example.

Regardless, I just felt that a critique is in order before anyone takes these recommendations to heart. A post like this could start a trend after all.


PS: a thread is a record, so now we have this "on record" as you say. And as for imperfections. These definitely add character. But I don't know if it is appropriate to describe imperfections in terms of geometry. It's simple enough to just say imperfections can add character. The problem with this however is, when you have a tile based organization. A tile with an imperfection must always have a duplicate tile lacking the imperfection. Otherwise you will see the same imperfection everywhere you look, and that will be more artificial than everything appearing squeaky clean.

[HwitVlf]

Dear Esther is just a nice example of one type of game design (realism), but there are many different approaches to game style each with their own appeal. What interested me mostly about Dear Esther is that it achieves a good "feeling" of realism while actually being mostly low-poly, low-resource.

I wouldn't say that I'm just endorsing adding "flaws" to models; it's also about choosing to design things so that they don't end up with lots of straight lines. For instance, in the bridge screenshot earlier, I don't think it would look as nice if they had chosen to design a perfectly flat "box" bridge. Because polygons are artificially flat by nature, anything you can do to can do to mask that, without over-using resources, makes things look better.

If your talking about SoM map Pieces specifically, I'd say you should always make 2 to 4 slightly different versions of each Piece in order to "break" artificial, repetitive straightness. It also helps to make irregular shaped Pieces slightly oversized so the different versions can overlap in a variety of ways at the seams.   If you're making a single outdoor building model, you can design it any way you like then chop it into ~2m chunks and make each chunk a unique map Piece. Collision shapes stay simple/flat of course.

I thought Ben did an excellent job at breaking up Map Piece repetition in his last Rathmor demo. 


If were talking specifically about SoM, 

[Holy_Diver]

In the past I have more than once had to develop software from scratch in order to inject straight lines into models developed with very high end 3D modeling software that is geared toward non-engineering applications. I don't know how things are nowadays but game and CGI artists have really never valued straight lines (enough to demand tools for making them) even when it comes to seams straight down the middle of a model.

If you don't just mean wabi-sabi or whatever by "flaws" then you have to come up with a technical justification. I wouldn't worry so much about artificial straightness, it just comes with the territory. You won't be able to make a map that doesn't look like a grid with SOM. And that's a good thing if you ask me. I bet Esther (not having played it) is modeled entirely from scratch without any repetitive elements outside of plants and things like that. To me that is not a scalable approach. It lends itself to completely disposable graphics. In other words you can't share and recycle them (it will be possible to make maps with disjoint elements one of these days, but its important to wholly embrace tiling in my book)

The Rathmor Crags demo is actually constructed out of a single columnar tile boundaries wise. You can't exactly do that either. It already takes a whole host of tiles to slap together a working tile set. I can't envision variation tiles or objects being worth the overhead. Although it would be worthwhile to consider developing a layer over SOM's file formats that could efficiently represent slight variations at somewhere down the road. Like no sooner than 2 years from today if I am doing it.

I don't want to wander too far off topic. I've said what I wanted to say. And also I've been replying in the hopes of winning some replies in kind vis a vis my own thread. So if don't get some soon I'll probably wander off

[HwitVlf]

If you're looking for some kind of specific help to get your game going, I just made a new "Help Wanted" section.

It's actually quite easy to break up repetitious tiling effects even in SoM. With 2 versions of a single floor Piece, each Piece can be rotated 4 directions that's 16 possible "looks" for each floor tile from the player's view. That means there are 256 different possible combinations for any two Pieces. And, if the Pieces are designed to overlap into the adjoining Piece, each Piece's apparent shape is affected by all 8 Pieces around it for almost limitless possibilities.  That's just from 2 different floor tile Pieces. Granted, it's harder to break visible wall repetitions, but still not that hard. The first screenshot shows the effects of only 2 different floor Pieces. The second also shows just 2 different wall Pieces. Tiling is not especially apparent in either and would be even less so after Objects are added.

[Holy_Diver]

If you're looking for some kind of specific help to get your game going, I just made a new "Help Wanted" section.

No, was just looking for you or anyone else to follow up on the replies that have already been made. I still have at least two more years of banging on SOM before I can even consider diverting time to any of those new KF games, but it wouldn't hurt to get an early jump if people want to. So just feedback in other words.

It's actually quite easy to break up repetitious tiling effects even in SoM. With 2 versions of a single floor Piece, each Piece can be rotated 4 directions that's 16 possible "looks" for each floor tile from the player's view. That means there are 256 different possible combinations for any two Pieces...

Yeah I don't doubt that. Right now I am very keen on getting a Paint system going in Armored Core tradition so texture/colouring changes can be trivial. And as far as I know there has to be a solution for rotating (really transforming) UV maps independent of the tiles, because otherwise you will always end up with the texture being mirrored if identical tiles are placed side by side floor and ceilings wise. With some textures that is acceptable, but for most it isn't. Maybe you can come up with a more creative way to work around this limitation by combining complementary tiles as you say...

Still I think this is a problem, because like if you have a hardwood floor that runs into the wall, the grain can't go both ways. It would nice to be able to handle those kinds of scenarios with a single tile. I tend to think more in organizational terms, because I don't see SOM as being worth a whole lot unless it can be turned into an online social thing. And for that you need databases, and whether they are centralized or distributed, they will need to be efficient (but for the time being making two walls, running either direction, is not so bad either as long as we can make floors lineup better than SOM's somehow)