Developer Diary 45 - The Art of Choice (Patreon)

Last time we talked about the art for the full-body drawing, and that’s the topic of today’s devlog as well. Based on your love for the tilted/angled view, I felt I had no other choice than to pursue this and see where it could lead me.

I will try to break down each view, and where I ended up with it both artistically and mentally.

Doll-like: Initially I really liked this. The more aggressive stylization seemed to offer a good excuse to push things a little bit further. To my surprise, your response to it was neutral to negative, mostly citing that it didn’t feel ‘Albatross-like’ or ‘Bubblegum Sunday-like’. This feedback initially didn’t work for me as I felt I could make it work, but over time it too began to align myself with this feeling. It didn’t take long for me to completely fall out of love with the idea. I also realized I would need to stylize all characters in this way, and I did not see how I could make rough male characters in this style. The doll-like view had one major ace up its sleeve though: it showed everything beautifully. A decent neutral pose (although stiff), plenty of room for expanding hips and finally: it showed a great view of the shoes, which the original did not. In the end I did not go any further with the doll idea, but took what I learned posture-wise and used that for future attempts.

(Clothes made it a little better, but I had now decided I wanted to keep the realistic approach)

Angled view: This one was the crowd favorite, which came as a big surprise to me since I had little love for it. In theory it seems perfect, since it shows so many things in just one pose. In practice however, things weren’t the same. I didn’t like how the character didn’t feel centered on screen, how she wasn’t really making eye contact, and after many failed attempts, I came to realize I could not work out the arms on this pose at all. (I’d like to quickly mention that I refer to my model as she/her, but you will free pronoun choice). One thing broke the camels back on this idea though: Hip expansion looks atrocious on it. Because of how the hips are angled, you see very little, and thus it almost feels insignificant to have wider hips at all. This really bothered me, and that made me not want to work on it any further.

(Lack of being able to connect with this view emotionally and the weak hip changes visually made me scrap this design)

I had a new idea. Let’s use the doll pose we have, which we know ticks every box we need, and use a more realistic body shape to bring it all together!

I had renewed energy, and this seemed like the ideal solution. I also approached this one from the opposite direction. I knew I disliked the hips on the angled view, so I wanted to make sure I was really happy with them for this one. So instead of starting her skinny and ending with the thick version, I began with the thick version and worked my way back. This turned out to be a really great way to work, since it helped me find much better proportions (many of you pointed out I was drawing my characters too tall). Once I got this on the page, I felt really happy and accomplished. But once again, side by side with the original, something gnawed at me. The original had such a natural pose, whatever I had here was really stiff feeling. On a side note: the original BGS also has a fully symmetrical body and face, with only the details like clothing and hair, etc, to break it up. I’d say nobody would look at the old BGS and say “Wow this feels symmetrical”.

With this in mind, I figured I’d try a little trick and sneak in a little ‘contrapposto’ to give her some pazzazz.  Giving her shoulders and hips a 3 degree tilt in opposite directions did quite a it to her posture, making it feel more casual and a little more relaxed. However, compared to the original, it still seemed stiff and rather undeveloped. I was getting pretty frustrated at this point. No matter what I did, I could not seem to overcome my initial attempt for the art. I decided to come at it logically: I knew I had scrapped the doll-like approach and the angled approach, since both had too many downsides to justify them. I had this newer version which seemed close, but no cigar, compared to the original attempt. Well, I figured: Let’s combine what works with these two and blend them together. I know it seems like a lot of work for a problem like this, but actually, this is a really important thing to figure out. This pose is so crucial, I would compare it to getting the jump in the original Mario right (which took over a year to fine-tune). This is my absolute base for the art. If this isn’t perfect, everything else will suffer. This is why I am so incredibly stubborn and demanding of this piece of art. 

(The result of taking the doll-like approach and making it more realistic. Close, but no cigar)

One more thing I’d like to mention about the art side of things is the size of the artwork. It’s hard to explain, but I think a picture might help make it a little more clear. Each head is shown at 100% zoom value, this means that the pixel density of the artwork is the same as my screen. This clearly shows how large the head is of the original BGS art. Bigger head means that more detail can be put into the art, but it also takes more time. The middle one was my original attempt, with the left one being my last attempt for body art. As you can see, they get smaller each time, and not without reason. New BGS will have a complete visible body, whereas BGS only showed a torso. If I would draw the art for this game at the same resolution, there is no way I could ever get this project anywhere near done. I want to simplify the art style, so I can add many more clothing options to the game. This means downsizing the art to compensate. The middle one might appear smaller, but compared to the original, it would still take more time to make art as the total image is still bigger. This is why I wanted to try working even smaller. The problem is that you cannot just downsize by any amount. You always have to go twice smaller (50% original size) or twice larger (200% original size). The small one would mean less detail (and less reason to zoom in) but many more clothing choices, whereas the one in the middle would offer the opposite. It’s a tough problem, and I’m still tackling it. The future will make it more clear how long exactly it would take to make a piece of clothing.

(My final rough attempt at merging everything I've learned together. This needs more work, but it feels pretty good)

Finally, Vivien (our programmer) has made progress with the camera functionality. Most people seemed happy with controls but pointed out it felt a little claustrophobic. This is because as you zoom, you lose the ability to go left and right (and losing out on background to look at). We did this initially because it seemed logical to keep the focus on the character and make sure she was always somewhat in view (zoom in on her hand and you’ll see how it appears centered on it). We realize now this gives a confined feeling, and we’ve undone this change. You will now be able to explore the full background, regardless how far you are zoomed in. We’re also looking at making changes to the double-tap functionality, which currently always zoomed out completely. In the future it will zoom out in steps, which will allow you to go from the furthest zoom to the portrait zoom, and only then showing the full view.

(Notice how in the After the No Parking sign is fully visible)

As always, I truly appreciate all the support and feedback you give us, thank you.

Alba & Vivien