Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[PatrikLundell]

Thanks Meph!

[voliol]

Better? ^^

...I feel stupid for not considering that. Thanks!

[Starver]

I too, did not really consider this.

Partly because I was unsure what you'd want to do about it (if you wanted to, if you thought you even could/should). Which was a whole set of presumptions of my part.

(That other thread I linked indicated possibly less immediate interaction by Ironhand, who I really only knew as "one of the names" before the changeover, being more a passive spectator of graphics-packs than an avid adherent of them, so I really don't know how relatively responsive it'll be until I start seeing activity along those lines in threads like this that I'm already 'watching'. In other words, I don't really know enough to comment, but that's never stopped me before!)

[jecowa]

Can we make suggestions here again?

I've been looking through the ramp tiles. I'm not sure if these two look right:

[TILE_GRAPHICS:GRASS_RAMPS:1:11:GRASS_RAMP_WITH_WALL_NW_SE]
[TILE_GRAPHICS:GRASS_RAMPS:3:11:GRASS_RAMP_WITH_WALL_NE_SW]

You can see one of them in positions 1:1 and 2:2 in this mockup:


I think it's odd that they have 2 diagonal "lines" in them instead of only 1 diagonal "line" like in the adjacent tiles located in positions 2:1 and 1:2 in the mockup. I'm having a hard time visualizing how it would look in 3D.

Maybe it could look a little more like this?:

[Starver]

(Not sure it is a good place, actually, but replying anyway.)

Though I see where you're going, your 'correct' example makes tiles 0:2/2:0 and 1:3/3:1 into partially-flat upper-Zs, where they were wholly-flat in the original, and some other rather more advanced changes to tile profile (not the least the traditionally ∆-slope tiles of 1:2 and 2:1 gaining a flat upper-shelf diagonal to match its non-orthogonal neighbours) that relies on far more look-around knowledge to invoke than ...from what I've seen of this process... anything Tarn has actually managed to encode in the rendering system.

The cross-diagonals of the 1:1 and 2:2 pieces (lower-Z 'flat' groove \-wise and two upwards-shoulders splaying out /-wise, yes? ...plus the internal/inverted shoulders of the 1:2 and 2:1's /-wise axis) seem necessary to maintain correctness of corner-and-edge matching against the reality of each (especially non-ramped) neighbour, whilst also maintaining visibly consistent liquid-flow possibilities through the low-cut grooves by not forcing the tile-centre up too far from the base-Z level (in 1:2 and 2:1, the common corner sits at low-Z, in the original, each centre is at Z-and-a-half on the way up to the upper level, but has no necessary part to play in low-level (3/7ths or less) liquid flows).

Given the constraints and rare edge-conditions of trying to render low, high and ramp-connecting tiles (maintaining the essentially profile-agnostic ramp system of pre-Steam convention), it's as good a kludge as I think you might expect.


All this reply being contingent upon my correctly comprehending the photoclinometry of the shaded relief. I don't think we're looking at a knife-edge \-wise ridge between top-left and bottom-right, at this isolated glance, but I cannot yetvrule it completely out in my mind, without the necessary cues taken from larger expanses of gameplay/mockup imagery.

[Mike Mayday]

Oh jeez. My mastery of the English language is not good enough to comprehend all that, I'm afraid :/ With some visual aid and drawing of sketches at the same time, I could probably explain my reasoning. Except that even now I don't remember it fully. All I can say is that I DID consider your approach, Jecova, but ended up settling on the final one because it provided good compromise - I don't remember and can't tell right now if there was something obviously wrong with yours.

I was also looking for methods that would require as little coding from Toady as possible. Everytime we added something, two new opportunities to make things more accurate arose, that would require extra art and coding. All in all, the ramp system is a bit janky, but IMO good enough. With everything, there was a point where we had to call it a day and move to other stuff. Maybe later?

At this point, if I were to request improvements to the ramp shading system, I'd probably focus more on the occlusion shading of edges between z levels, e.g. when there are two ramps forming a single surface. This would allow for nice rounded slope corners. But I've seen someone make that mod already and it looks better than I thought it would.

[Starver]

To simplify my message (me? complicated? surely not!), my interpretation of how the system goes is that non-ramp tiles are consistently flat, with no regard for what ramps border them[1], whilst ramped tiles are selected to match the elevations of all four corners the 'best'[2], with a bias towards having a low-Z continuity in (e.g., as per example above) two opposite corners pinned-High and the other two pinned-Low. And it does this in (mostly?) in quadrants of flatness (but angled accordingly), formed from the X-shape of the two diagonals cutting across the perimeter.

(Is that more simple a description? Is it even accurate? It'll have to do for now.)

When we were still all pondering this, originally, I had played with the possibility of octants[3] and some 'bleed' effects into adjacent flat-ground edges (or even an octothorn/twisted octothorn subregioning) for more fine-grained level-matching, but I think the quadrant version works well enough. It just needs 45° diagonals (and the tile-edge horizontals and verticals), rather than messing with an expanding multitude of 22.5° slope-meets-slope variations, etc.

(Whoops, between that paragraph and the various feetnete, below, I've definitely gone complicated again. But I think it is a complicated subject, and no avoiding that. Definitely when discussing how it could be even less trivial than it currently is ... Not that there's any single good answer to it, but I think what we see is as good as we need.)


[1] We saw with streams, though, that the texture-style for flat ground upon the immediate bank was used to add a 'hang-over' of matching texture-style, rendered above the water's edge tile, e.g. grass stems hanging over, to break up the hard edge. I presume this still exists.

[2] Not sure how the example of ramps would be so touching two adjacent tiles, one low-Z and one high-Z. I suspect it'll be the ramps that would be there if the low-Z were also a ramp... If nothing else, to not have a cascade of earth-reshaping, beyond the single tile you might purposefully enramp/deramp. But I'd have to try this out with Steam version to confirm my prediction.

[3] Either divided with orthagonals and diagonals (N-S, NE-SW, etc) or the panels orientated that way with the dividers at the mid-angles (NNE-SSW, ENE-WSW, etc)

edited to resolve inadvertent bbcode invocation

[Claije]

What determines if an image uses the material color or the color straight from the .png in the steam version's tileset?

Metal Bars, Tables, and Trousers are all material colored, but cheese for example is not and has a base yellow sprite for all cheese.
If I recolor that cheese to the DEFAULT_TEMPLATE color, it doesn't recolor to material colors defined in the cheese.
I haven't located anywhere that implicates images to use palettes, but I've seen comments in the files that say different images are palette colored. Is this all hardcoded?

[Schmaven]

The official graphics are so nice.  They make it so easy to play the game without having to stop to inspect various tile contents.  In 1 of the old tile sets I had grown accustomed to, there was some sort of visual indication as to whether a tile of a farm plot had been planted or not.  My understanding of things is that graphical work is still ongoing, so I don't want to distract from that.  I'd guess more creature / plant sprites would be on the to-do list.  What else might you all be working on now that the premium release is out in the wild?  Keep up the good work!

[darkhog]

I hope you will add more gore to fully show what the ASCII version can show through the various text tiles. That, and also various environment tiles that look the same in the graphic version, but use different text tiles in the ASCII one.

[jecowa]

Open space (a hole) on the surface looks kind of similar to a patch of sand. Maybe add a drop shadow to the open space.

Spoiler: Current open space (click to show/hide)

Spoiler: Edited open space with shading (click to show/hide)