FMP - week 18-19 - polish, bugfixing, and art integration

This sprint:

• improved landscaping based on feedback

• polished village

• integrated animations with forest swoop logic

• bugs bugs bugs!

• New intermediate build for testing

Archer model

In terms of night loop artwork this is the biggest task on my todolist still: to make a 3D model for an archer. This archer will be used in the castle attack minigame where they will be shooting at you from the castle.

I really hoped I could get away with a fully armoured archer so I wouldn't have to model a face, but after researching what medieval archers looked like I found that, unfortunately, most wore primarily cloth with a helmet. Considering our target audience, I felt it was important to try and stay believable with these kinds of details to help with a sense of immersion. I know this is my imaginary game world and technically I could do whatever I want, but I think it would hurt the player experience. Oh well, so be it! It just means it will take me more time. But since some of my other tasks got done a lot quicker than anticipated I think I should be fine in terms of scope.

This model will require modelling of clothes: a skill I do not yet properly possess. I have tried short cuts before, for instance with my 740 Melinoe model, but I noticed that the result is not great, and I end up losing a lot of time anyway fixing details. So, I decided to research different methods, I found the following that would work in Blender:

• Using the sewing option with the cloth modifier (van den Hemel 2022)

• Using a shrink wrap modifier combined with snapping turned on (Abbitt 2018; CGI Zone 2019)

• Sculpting from scratch and doing retopology (Schmidt 2018)

I started with the shrink wrap modifier flow with snapping, but no matter what I tried, it did not work as smoothly as Abbitt and CGI Zone showed in the video. Since my blender version was newer I couldn’t select the exact snapping settings they had so maybe that’s why. The resulting topology was such a mess. Only trying to get started with the basic shape took me over 2 hours. It was also extremely frustrating that I couldn’t see my vertices anymore with the combo of the snapping and the shrinkwrap. The latter also led to a lot of accidental movement of vertices.

I decided to start over as it was clearly not leading me anywhere. I think it might have been that I tried to move multiple vertices at once and that that doesn’t work properly with snapping. Sculpting from scratch would take me too much time, especially after wasting 2 hours trying to get the previous method to work. I looked into sewing but this meant that I had to find a proper sewing pattern for the clothes I needed, which proved harder than anticipated.

With not a lot of options left I decided to give the shrinkwrap modifier another chance, but this time without the snapping. This was surprisingly effective and also fairly quick. Because I couldn’t use snapping I had to make sure that the clothing mesh followed the body closely, to avoid unexpected behaviour of vertices. In the end I managed to model the full archer in about 8 hours, including texturing. I was quite happy with that result. It was still a time investment of course but it was in line with what I expected and I was satisfied with the result. The resulting archer model can be viewed below.

As you can see I decided to not model the string of the bow and keep the bow itself very simple. The reason for this is that the archer model will never be seen up close in the game, and it will be night time when we use the archer models. A user would thus never notice the string of the bow or any details and I could save time by leaving those aspects out of the modelling process.

Polish and bug fixing

The rest of my week was mainly polish and bug fixing! Not very excited but very essential. I fixed a ton of things that came up in user tests, like:

• Floating wooden logs

• Floating fences

• Wrong orientation of player

• Polished up the village and fenced it off

• Fenced areas off and changed landscape to direct player better through the map

• And a lot more!

Many players noted that they got lost in the day loop map. I applied some of the learnings from GDC talk and made sure that the light sources where coming from the direction the player needed to go, I fenced off areas and creating natural barriers to give a corridor effect while the world still seemed large. There is a lot that can be done here with environmental storytelling still, but I will leave that to a later polish sprint again.

In the screenshots below you can see the difference between the first version of the village and the new more polished version.

Wrapping up forest swoop

Rob had gotten very far with the forest swoop minigame, which meant that it was time for me to integrate the art into the game logic. A lot was already done by Rob, the main thing that I needed to do was syncing and triggering of animations.

For this scene this wasn’t completely straightforward as I wanted the dragon to tilt as it turned and this wasn’t an essential part of the game logic. Therefore, I needed to adjust Rob’s code to determine the amount of tilt to use on the dragon.

To achieve tilt I created a blendtree (Unity Technologies 2021a) for the dragon that would shift the dragon from completely left tilted to completely right tilted based on a tilt amount. In the code I tracked for how long the user was holding the A or D keys to increment the tilt amount the longer they were turning their dragons.

I also still wanted the dragon to occasionally flap its wings, for which I added the following code:

The code is easy but effective. I think the animations could use some more spice, but the code is done for now and it's just a matter of polishing animations in a polish sprint.

After putting it into Unity and testing it out I did realise that the dragon didn’t flap its wings in the tilted state properly, It would switch back to center to flap. I wanted the dragon to also flap while being tilted. To achieve this I put the tilt animation logic in a separate layer of the dragon animation controller(Unity Technologies 2021b). This way the dragon would switch properly from soaring, to flapping to swooping, while still having the tilt act on top of the animations.

There is still a lot to be done with the environment of the level, but the dragon controller now works properly with animations and all.

List of references

ABBITT, Grant. 2018. ‘Sculpting Armour Workflow’. YouTube [online]. Available at: https://www.youtube.com/watch?v=Gbc0DY950Ho&t=153s [accessed 2 Feb 2023].

CGI ZONE. 2019. ‘Cloth Modeling Using Shrinkwrap Modifier in Blender 2.81’. YouTube [online]. Available at: https://www.youtube.com/watch?v=QF3aPHDHuus&t=83s [accessed 2 Feb 2023].

VAN DEN HEMEL, Jan. 2022. ‘Blender Secrets - Cloth Simulation Sewing Basics’. YouTube [online]. Available at: https://www.youtube.com/watch?v=5t2WQJMoQDM [accessed 2 Feb 2023].

SCHMIDT, Jonty. 2018. ‘Blender Clothing Process’. YouTube [online]. Available at: https://www.youtube.com/watch?v=9zNAATDadJc [accessed 2 Feb 2023].

UNITY TECHNOLOGIES. 2021a. ‘Blend Trees’. Unity Manual [online]. Available at: https://docs.unity3d.com/Manual/class-BlendTree.html [accessed 26 Feb 2023].

UNITY TECHNOLOGIES. 2021b. ‘Animation Layers’. Unity Manual [online]. Available at: https://docs.unity3d.com/Manual/AnimationLayers.html [accessed 26 Feb 2023].