The new Wesnoth logo and the Loyalist sigil

[pauxlo]

The problem is that decreasing any contrast in the reflection pattern, while making the text look more legible in small scale, will make it look less realistic and metallic in larger scale. I just have to strike the balance somewhere.

Couldn't there be made different versions for different scale? (I know about nothing about vector graphics and how feasible this would be, but I know that some fonts have a quite different look for smaller scales, to be more legible.)

[jacobolus]

The problem here is that there is not enough distinction between positive and negative space in the text, or between the letter shapes and the texture inside and around them. Both the letters themselves and the background are visually busy and made up of a mix of values from medium-light to medium-dark. The font also has letters which have a bit less negative space, and a bit more design uniformity than the previous font. The big problem though is with contrast in value and texture. If you squint a bit, the whole logotype turns into sort of a big blob. My advice would be to make the drop shadows narrower and darker, to make the letters lighter, to make the shading of the letters generally crisper/less complex, and to blur out the background a bit, and maybe tone down the shading on the mountains under “Battle” and move the trees a bit out of the way.† It might also help to make the font slightly thinner in weight, or ease up on the tracking, especially in the words “for” and “Wesnoth” which are pretty cramped. Also, to my eyes the letters are quite greenish; that’s not a problem per se, but is it intended?

It’s a bit unrelated, but I also think some taller stems on the t’s would help the words’ readability.

As usual, this sort of thing should be designed to work in grayscale first; use the “Lab” mode in Photoshop and put a hue/saturation adjustment layer that entirely desaturates the image for a reasonable preview. The old version could be better, but it’s still noticeably more readable than the new one.

† Ideally there’d be a region of the screen with a bit of a visual emphasis vacuum where the logotype could nicely nestle, not competing for attention. Right now the trees are much visually heavier than the logotype, which itself kind of blends into the background.

[Sgt. Groovy]

My advice would be [...], to make the letters lighter, to make the shading of the letters generally crisper/less complex,

I could do that, but the metallic impression of the text would suffer, especially in larger resolution. The letters are made of facets pointing in widely varying directions, so there should be strong contrast between them.

I've tried another way to render the text, with raised effect, but with a flattened, cast-looking top. The strong contrasts are now limited to the edges of the letters, making them less textured. It works pretty well on larger scale as well, but in general, the look is a bit boring and flat.

...and to blur out the background a bit, and maybe tone down the shading on the mountains under “Battle” and move the trees a bit out of the way [...] Ideally there’d be a region of the screen with a bit of a visual emphasis vacuum where the logotype could nicely nestle, not competing for attention. Right now the trees are much visually heavier than the logotype, which itself kind of blends into the background.

The relative placement of the background and the text is not fixed, therefore place-limited alterations to the background is out of the question. Ideally, I would have much more homogenous background for the text, like the menubox.

It might also help to make the font slightly thinner in weight,

Heavy weight and strong stress are hallmarks of blackletter fonts, as is limited negative space. The font has already been stripped of lot of interesting detail to make it more legible, it is about as simple and streamlined as it can to still pass as a blackletter. Reducing the weight would risk losing its character and the cultural connotations for which I chose this particular style. If I went that way, I might as well use a different kind of font altogether.

ease up on the tracking, especially in the words “for” and “Wesnoth” which are pretty cramped.

"Wesnoth" is pretty cramped, but this is to make it more in balance with "Battle" allowing "for" to be centered. I can ease up on the kerning, but the overall balance of the text will suffer.

Also, to my eyes the letters are quite greenish; that’s not a problem per se, but is it intended?

If it's not a problem, why do you need to ask?

Yes, it's intentional, the idea is to make some hue contrast between the reddish yellow metal of the shield and the yellow metal of the text. I didn't want to make it even more reddish, so it has to be greenish (actually it's just yellow, but dark, so it looks like green). As a benefit, this also creates a warm-cold contrast.

It’s a bit unrelated, but I also think some taller stems on the t’s would help the words’ readability.

I actually have two different versions of the "t", but again, using the pointy stem version, even though it might look better, or more familiar to modern audience, would make "Battle" narrower. Using the pointy stem on the wider version would look weird, though.

wesfont-blackl-2.png (1.25 KiB) Viewed 6281 times

As you see, while your suggestions would increase the legibility, it would always be at the expense of something else. An elaborate and detailed design like this is always riddled with complex trade-offs, and there is really no objective way to define the optimal point of balance in any of them. It all comes down to the point of view and priorities. Your POV is more of that of a graphic designer, while mine is more of that of an artist.

As usual, this sort of thing should be designed to work in grayscale first;

When working with an image that is supposed to be used also as B&W version, this surely holds true. Then the image really should be defined by value and the colour is only and added decoration.

But with an image that will be used only in colour, this kind of formulaic thinking can be restrictive. Here's a classical example: Monet's Impression, Sunrise. When an image has a light source, like sun, one would expect it to have higher value than it's surroundings. The sun and it's reflection surely look bright. Let's see how it looks in black and white. WTF, where did the sunrise disappear? It turns out that the brightness of the sun is entirely created with contrasts in hue and saturation, not in value at all. The whole morning lighting of the painting is really made of colour, in black and white it looks like it's night.

use the “Lab” mode in Photoshop and put a hue/saturation adjustment layer that entirely desaturates the image for a reasonable preview.

I really hope people would stop telling me how to do things in Photoshop when the whole point of this project is to produce the logo in vector format and the work is being done in Inkscape.

[enchilado]

The texture on the letters makes the whole thing look a bit... cheap. Partially because the texture isn't present on the sides of the letters, but also because applying textures like that is never going to look particularly amazing.

[Sapient]

The texture on the letters makes the whole thing look a bit... cheap. Partially because the texture isn't present on the sides of the letters, but also because applying textures like that is never going to look particularly amazing.

There are some harsh critics in here! I don't know, to me it looks great. I would go so far as saying that it may be hard to improve upon this latest attempt. Feel free to prove me wrong, though.

[Captain_Wrathbow]

[peanut_gallery_comment]
IMHO, this latest version is much too flat, although I can't attribute that to anything but my own personal preference; that is, there's really nothing wrong with it artistically, I just don't like it myself.
[/peanut_gallery_comment]

[JAP]

I realise that this is much a matter of taste and the following is just my opinion.

I much preferred the previous version, it looked excellent.
The new version probably has a better readability but the older version was plenty good enough in that aspect.
I mean the human brain is specialized on pattern recognition, its what we do best of all.
So if a better looking version takes 1/100 of a second longer to read... i still prefer the better looking version.

[Jetrel]

I also prefer the previous version.

I realise that this is much a matter of taste and the following is just my opinion.

I much preferred the previous version, it looked excellent.
The new version probably has a better readability but the older version was plenty good enough in that aspect.
I mean the human brain is specialized on pattern recognition, its what we do best of all.
So if a better looking version takes 1/100 of a second longer to read... i still prefer the better looking version.

[jacobolus]

Heavy weight and strong stress are hallmarks of blackletter fonts, as is limited negative space. The font has already been stripped of lot of interesting detail to make it more legible, it is about as simple and streamlined as it can to still pass as a blackletter. Reducing the weight would risk losing its character and the cultural connotations for which I chose this particular style. If I went that way, I might as well use a different kind of font altogether.

That’s fine. The problem is when you have very busy letter shapes above very busy background shapes, both with busy textures. In order to make the text legible, there needs to be *some kind* of sufficient contrast. In this design, there wasn’t enough value contrast, or texture contrast, or shape contrast, to make the letters stand out enough so that it can be read without excessive effort.

As usual, this sort of thing should be designed to work in grayscale first;

When working with an image that is supposed to be used also as B&W version, this surely holds true. Then the image really should be defined by value and the colour is only and added decoration.

But with an image that will be used only in colour, this kind of formulaic thinking can be restrictive. Here's a classical example: Monet's Impression, Sunrise. When an image has a light source, like sun, one would expect it to have higher value than it's surroundings. The sun and it's reflection surely look bright. Let's see how it looks in black and white. WTF, where did the sunrise disappear? It turns out that the brightness of the sun is entirely created with contrasts in hue and saturation, not in value at all. The whole morning lighting of the painting is really made of colour, in black and white it looks like it's night.

Monet’s example works because it uses extreme color contrasts instead of value contrast. Worth noting: Monet’s shapes are simple, and rendering text in Monet’s color scheme would make it entirely illegible and give readers a headache.

The reading of text relies on discrimination of edges and fine details, which is done by the human visual system mostly in a grayscale similar to the L* of CIELAB. Since chromatic content has so much less impact on text legibility than lightness content, it doesn’t really matter whether the ultimate design is to be viewed in color or black-and-white: a preview of just the lightness content is very nearly exactly as readable as the color version, but a artist looking at it doesn’t get distracted by the extraneous color in trying to make that determination.

use the “Lab” mode in Photoshop and put a hue/saturation adjustment layer that entirely desaturates the image for a reasonable preview.

I really hope people would stop telling me how to do things in Photoshop when the whole point of this project is to produce the logo in vector format and the work is being done in Inkscape.

I’m not trying to tell you what to do. I was offering one suggestion for a way to preview what the design would look like in a grayscale roughly akin to that used by the human visual system. As far as I know, Inkscape and GIMP have no way to work with images in CIELAB or any other perceptually relevant color space, whereas Photoshop does; even if you don’t make the image in Photoshop it can be used to render a useful gray preview. Feel free to figure out some other way to make one, or not bother with it at all, if you don’t want. In general, designing logotypes with some reference to how the human visual system works is helpful if you want them to be legible. But if you’re happy with the result otherwise, then go for it.

[Sgt. Groovy]

Monet’s example works because it uses extreme color contrasts instead of value contrast. Worth noting: Monet’s shapes are simple, and rendering text in Monet’s color scheme would make it entirely illegible and give readers a headache.

Speculation aside, let'ssee how it would actually look with the Monet colours.

While I definitely wouldn't render a longer a text this way, this sure isn't "entirely illegible." The lack of value contrast makes the edges look blurry, but the surfaces stand clearly apart, contributing to the eerie, otherwordly luminosity of the text that would be much more complex to achieve with value. While it makes the text less legible, it might make it look more interesting, motivating the viewer to spend the extra effort it takes to read it, in turn making it more memorable.

In general, designing logotypes with some reference to how the human visual system works is helpful if you want them to be legible.

That's true, and I really don't disagree with your suggestions regarding legibility. But legibility isn't everything and your suggestions aren't unambiguously improvements when other considerations are taken into account. If I was only worried about legibility, I would just use solid colour filled letters with a good contrasting border. If this was a commercial project, where the logo would have to compete with attention with other logos on the game store shelves, that's what the marketing department would probably tell me to do, artistic values be damned.

But this isn't a commercial project, and the new players are mostly guided to the game by word-of-mouth on the internet. The main purpose of the logo isn't to advertise the game but to be displayed inside the game to people who already have decided to play it and already know what its name is. It's supposed to influence the general atmosphere of the gaming experience, and I think this puts more weight on the artistic aspects of the design. When used in other contexts, the logo should be used with a much more plain background.

[jacobolus]

You’re right, it’s possible to read text in a color scheme with *some* value contrast (from the scheme you drew from Monet’s painting the CIELAB lightnesses of the background and foreground are respectively 48 and 57). (see image)

If you get rid of the value contrast altogether though, it becomes much more difficult (and headache-inducing), as we can see if for example we take the same colors and move both their lightnesses to L* = 52. It’s still vaguely possible to read, because the color contrast is so extreme, and because we’ve eliminated all texture. (see other image)

See what happens if we further reduce the color contrast to half its current amount. (see other other image)

Anyway, I’ll refrain from dragging this discussion any further off topic. Hopefully the input about human vision wasn’t too big a distraction.

[Sgt. Groovy]

Not at all, it's always a learning experience.

Anyway, back to the topic, here's the sceptre embossment completely redone. It doesn't make much difference in the small scale, but looks better and more consistent with the background in large scale.

I'll probably have to yet redo the leather and the bump patterns, not because of how they look, but because I've used a gimmick that I've since learned to be undefined in the SVG standard (meaning that it might not render the same way in other renderers and maybe not even in future versions of Inkscape).

[LordBob]

Since we're back to the shield, I'm going to rise again a concern which has been discussed before - sorry about that.

I think we had already mentionned how contrast in metallic reflexions very much depends on the environnement, and how metallic parts should reflect something in order for the material to be recognized. However, while the edge of the shield is now believeable, I'm having trouble with the center plate which looks a lot like plastic, especially the upper half and the textured background surrounding the scepter. I'm not sure if it's a lack of contrast between highlights & darks, a case of hue/saturation, a lack of simulated environnement, or said environnement not matching that of the edge reflexions. But even wih the full text in front of the shield, it still shows.

Sorry I can't be more precise, it's a matter that I would myself approach through trial and error. Still, I hope this still helps improve.

[Sgt. Groovy]

Would this help?

[Sgt. Groovy]

Here's the updated shield with a small reorganisation of the text. I found it natural to scale down "for" since "The" is also scaled down. It should be a consistent feature across translations to scale down the articles and prepositions (or any other similar constructs).