Official Specifications for Art Section of the Wiki

[artisticdude]

I'm currently in the process of updating the art sections on the wiki, and while reading through the How to create motion blurs page, I noticed that it says:

To draw that path of the weapon in wesnoth, you fill that arc which the weapon would swing in with the color of the swinging weapon (generally the brighter colors of it). We usually make it slightly transparent (around 30% transparent or 70% opaque), but in practice it tends not to make much difference if the swing is slightly transparent or not.

Which raises two questions concerning the current official specifications for motion blurs:

1. Motion blurs should use only the two brightest colors of the weapon palette?
2. Motion blurs can be semi-transparent?

My personal preference would be:

1. Motion blurs are always made with the 2 brightest colors of the weapon palette, although in some rare cases the third-brightest color may be used as well
2. Motion blurs should never be transparent at all; they should always be 100% opaque


The first is pretty self-evident, I think. My reasoning for the second is that, even though I'm by no means a 'pixel purist', IMO semi-transparent pixels in motion blurs cause more problems than they're worth for several reasons.

1. They mess up TC.
2. It can be relatively easy to mess up the consistent level of transparency in the motion blur when editing the frame.
3. Fully-opaque motion blurs just look cleaner.
4. Most, if not all, of the more recent animations have used fully opaque motion blurs.

I'd like it if I could get confirmation from one of the art directors on this, but I feel fairly certain that fully opaque motion blurs are the standard currently, so I have updated the wiki page I linked to above. That change can be reverted if necessary.

[thespaceinvader]

I would concur with your assessment. Jetrel has been using opaque blurs for quite a while IIRC.

[Jetrel]

Yeah, our nominal standard is transparent motion blurs, but it's consistency for the sake of consistency, rather than something inherently meritorious.

I'm all for having opaque motion blurs be the new standard; they look about the same in action (due to "persistence of vision" making stuff that only appears for a few hundred milliseconds look somewhat transparent), and they're a lot easier to deal with.


I have indeed been doing opaque motion blurs lately on a lot of stuff, because it's enough trouble just getting the work done in the first place.

[homunculus]

As said earlier, it would also be nice to have some recommendations (if not specifications) in the wiki for things like unit view angles.

Now I think I have managed quite a nice illustrative animation.
If the imaginary "floor" under a unit appears at 1/5 length (along the vertical direction on the screen), some characteristic mainline units seem to stand on that "floor" without visual problems.
However, 1/6 seems to be a close second, and I am somewhat undecided here.

some more explanation, if needed:

I did a red cross on the hex image, in the direction the unit is facing (assuming unit should be facing between south and south-east neighbor hexes), and scaled the hex with the red cross to 1/6, 1/5 and 1/4 height.
Then placed some mainline units over those "floors" to see if they look right.

Is 1/5 good as a recommendation that could be in the wiki, at least for starting spriters?

[francophone]

The mounted unit appears to correspond rather to 1/4. The orc rather to 1/6.
This is quite logical. The elements of an unit like the orc are best seen from the front. The converse to a mounted unit is best seen in profile. If the elements are lined and seen in front of one, others will be hidden behind the nearest; in profile, an arms hides some of chest that hides the other arm; in face, head and neck of the horse hide great party of rider who hides the back of the horse. The greater surface area and the most significant figure depend on the unit. Even unconsciously, we choose the angle of view based on that. (Perhaps wrongly, since reducing the perspective with this trend, which flattened the image of the unit.) So the angle is not always the same.
So if we want to find a standard, for a reference and not a coercive norm, we must make an average looking and not an angle that corresponds exactly to all units with a straight position.
So 1/4 seems to me (personally, the mage's feet and all the orc seem to me too facing). This is a subjective opinion, preference. So, it is difficult to establish a reference because the choice of angle is subjective.

But personally I prefer to measure the angle in a different way. The system of homunculus is logical but from a subjective choice of horizontal orientation. And is it is quite complicated. It does not mention any practical angle of talk about resizing report. What interests me is the angle resulting on the drawing. ie for "1/4", a line of 8 for 1 pixels for alignment of the shoulders, and 2 for 1 for gaze direction of the unit.

[thespaceinvader]

There really isn't a standard for that. It's been 'what looks best' for most of the existence of the game - but that's not to say that standardisation is bad, just that it's never really happened.

[homunculus]

@thespaceinvader:
Well, I didn't really mean firm standardization, just some suggestions about what issues to pay attention to.
And it is probably just a coincidence that the conclusion that different spriters came to about the view angle is similar enough.
Recently I posted a C&C which had a list of some rather formal issues with an image (to make it fit where it was supposed to), each of those would probably have taken only a few minutes to correct, but the trouble was that the list was about 10 points long (I don't remember how long exactly, could have been more).
Figuring out such formalities, each spriter on his own, is in my opinion, not the optimal way to spend the effort of starting spriters, while it could be in wiki in a nutshell form.

@francophone:
If you want the ratios to be like 'a line of 8 for 1 pixels for alignment of the shoulders, and 2 for 1 for gaze direction of the unit.' it might be a motivation to see things in a way Wesnoth units really are not.
It is not such a great effort to draw a line that alternates 2 and 3 pixels, it is much worse if you get the view angle wrong in the final sprite.
And, you only need to do it once, if you save the forward and sideways lines as an image you can drag it to your editor as a layer whatever sprite you draw.
The mage, for example, is too much side view for 1/4.
The hooves of the horse fit best in 1/5 if both forward and sideways directions are accounted for.
But the hooves fit in 1/6 just as well or even better, if only the forward direction is accounted for.
Just the body of the horse seems to be drawn from more top-down view than other units, and would fit better with 1/4.
And also, looking at the geometry guides it becomes more apparent that the "shoulders" and "hips" of the horse might not be perfectly aligned, but a little bit shifted as if the horse was belly-dancing (thumbs up for the glamour).

And finally, to clarify why such recommendations would be nice for starting spriters:

the somewhat anecdotal story of the young yeti (yeti toddler):

Young yeti was supposed to be a somewhat funny looking lev 2 unit, about as large as a full grown orc, but with awkward fighting moves like characteristic to toddlers, so I analyzed quite a large number of toddler fighting videos (=spent some time and effort).
Next I did a stick figure animation for attack and defense (and, more or less, movement) in the south-east direction, the way my best guess was at the time, and posted it in the forum.
Unfortunately, it was not pointed out that the unit view angle and the attack animation direction were wrong.
So, as this was my best guess, I went ahead and did blocking for the full animation (=spent some more time and effort, now idea how much).
And here is the blocking of the base frame (wanted to do it in grayscale rather than with different colors, so that smooth shading is not what was supposed to be in the final sprite):

young-yeti-blocks.png (4.63 KiB) Viewed 4576 times

As you can see, the view angle is wrong, and probably also the direction the unit is facing is wrong.
Also, the movement direction was wrong in the animations.

The image "says" it was created 'Wednesday, December 08, 2010, 10:23:25'.
Quite a lot of time and effort went to waste, and we didn't get an animated young yeti at that time, and I doubt I am going to resurrect the project in any foreseeable future.
Could have just read a list of recommendations in the wiki, and the effort would have been spent on drawing the sprite instead.

[francophone]

"A line that alternates 2 and 3 pixels" is not a problem for me. This is 5 pixels for 2.
What I mean is that it seems more convenient to indicate in a tutorial a report of 5 pixels for 2, rather than report of the enflatment of a cross.

[homunculus]

"A line that alternates 2 and 3 pixels" is not a problem for me. This is 5 pixels for 2.
What I mean is that it seems more convenient to indicate in a tutorial a report of 5 pixels for 2, rather than report of the enflatment of a cross.

oh, i know i am terrible at explaining things to people who just want to draw.

of course it does not need to be written "my way", which just attempts to be more transparent about where those lines come from.

using 1/4 may result in simpler ratios, but i suspect there will be more trouble that way, because foreshortening of the vertical direction becomes more important, and that is probably more trouble than the 1/5 or 1/6 lines.