Summary of Work For Off the Map/Report

11 week project.

In regards to FPS, the computer I worked on showed an average of 30 FPS.

I only realised how much lower it was on weaker computers right before hand-in, so if the level opens up at a much lower frame-rate than the one I talk about when optimising, it's because frame rate changes for different computers. Of course for the competition I will be working on getting the frame rate up to a much higher average so that it can have a good frame rate regardless of the computer it plays on.

This is a summary of the work I have produced for Off the Map. Most of it is final outcomes or simplified process, so for the full iterative process please refer to previous blog posts, and my Off the Map report where I clearly break down my technical and creative processes for this project.

Planning:

Our Gantt Chart for the whole project, the one I made, obviously not all of these targets were met and a lot of work had to be cut from the project, mostly due to people not meeting their weekly targets. On reflection I should have started lighting regardless of any assets in engine, however I was slightly intimidated by the task and wanted assets to build my lighting around, since our environment is a shadowy forest, and I wanted to know where needed illuminating and where could be left darker. I should have done this the other way around, thinking about it, where I place the assets into the illuminated areas, but it's too late to change anything now.

Also certain assets like the fairy doors and windows I only received in the last week, and these were really important in getting a lot of the lighting and mood right, mostly because they allowed me to create colour zones throughout the environment.

Furthermore placing assets was a nightmare since I only received the majority of assets on the Thursday before the final week. This meant I had to focus on getting the bigger assets down, and so missed out on placing my own smaller assets, the mushrooms and daisies as well as I could have.

The most frustrating thing this project was people's inability to plan or plan thoroughly. I was not meant to be team leader, however that is essentially what I became. I tried to give people weekly goals via the weekly timetable, and use meetings twice a week to check up on people, however people either didn't meet their goals, or they finished them early and rather than thinking I've finished early, let me do more work for the week, they finished early and then didn't really work for the rest of the week.

Anyway, here is an example of the weekly timetable:

Which I broke down into a day by day list of tasks too.

We also had an asset list that I pushed and pushed to get updated regularly, I also pushed for more assets to be decided on since our concepts didn't really actually show what needed to be made for the project, and initially we had very few assets assigned to people.

Here is part of the asset list as an example:

I was also heavily involved in deciding the look of our level, and in deciding on game play (especially since I did all the blueprints for the project).

Here are the mood-boards I put together:

Here is a concept of Oberon I did to guide the character artist in his look development:

I also produced a style guide and technical document for the project, which will be up as PDFs.

Of course despite the technical document I laid out for people work wasn't always up to scratch, so I wasted a lot of time chasing people up on fixing their models and textures.

I pitched this initial idea for the level:

I initially made a white-box of a Maze but that was deleted before moving onto the new white-box, which I created using a map one of the team members had given me.

I began working out the level scale and layout in engine using basic primitives.

Like so:

I then took the whitebox into Max and iterated on it:

After feedback however we cut down the level quite a bit:

Before I moved onto creating terrain to replace the white-box

On reflection I made a big mistake with the white box in that I didn't put in smaller scale references for players to compare assets to. I focused more on working out the distances the player would travel, and how navigating through the level felt. I will avoid this mistake in future.

Whilst I was still working out the shape of the landscape. Not amazing at all. (I made some super quick trees to mark out the forest), and we initially had this thin stream running from the second clearing out into the distance.

It evolved quite a lot over Easter, and I managed to get it looking and feeling a) more like a forest, and b) more contained for the player. We used high banks so that the player would feel more like they were in a forest, and not just a piece of land surrounded by a few trees.

Here are the changes the first clearing went through:

And the second clearing:

I used a lot of primitives in engine to block in where I thought certain assets should be, since I didn't have the assets themselves to build with, and I wanted to at least get an idea for the levels composition whilst sculpting terrain.

This image and the concepts provided by Hannah really helped when I was concepting this clearing:

I also did a lot of brainstorming for assets to be added to the level, and ways we could break up the high banks and make it feel more like a dense forest.

Whilst sculpting the path I tried to add in smaller areas of detail to make it more interesting for the player. So rather than clearing 1, path, 2, I made it clearing 1, start of path, interesting alcove, path, pool, clearing 2.

Trips to Bradgate and Abbey Park that I organised helped a lot with getting the terrain to look right, and work out how to start populating it.

(photos from trip). I also used them to gather my own photos for texturing.

I made the textures for the terrain. This involved a LOT of iteration and was helped by useful feedback. Especially when it came to nailing the transition from grass to dirt. (which could still be improved further by even more variation in textures and even better height maps to aid the blending between the two).

To make the terrain textures I used my own photos.

Initial grass texture used:

First Grass Albedo: (Although I did make about 3 versions of this before I was satisfied with how it tiled).

First Grass Normal:

Initial Grass Height I made (which didn't blend very well):

Final Grass height (for the grass I Initially used):

However I wasn't happy with this grass and so following feedback I completely remade it:

Photo I used:

Iterative process for final grass used:

Left is first attempt, right is final attempt:

I made another terrain texture for dirt, which had twigs and rocks on it, to add a bit of interest to the ground:

I tried using height for this but it made the transition worse so I stuck to only having a height map for the grass.

Finally I made the base dirt texture using substance designer. I say base since it's the main one, that I then paint the detailed one of top of to add a bit of variation. I realised AFTER hand in that I kept that texture at 4096 by accident. Needless to say I'm down-scaling that for the competition.

I would have liked to make roughness for the grass and main dirt, however the constant 1 seemed to work well enough, and as always, there's so much to do, that in places where it works well enough without any extra time needing to be spent on it, it's best to leave it until I do find the time to go in and do those minor details (and I say minor because the lighting is so dark in the forest that any specularity on the grass and dirt wouldn't really be seen anyway. The places I would have liked to do specularity for dirt are the banks, since they would have benefited from it.

Terrain Material:

I made other test textures for the terrain but didn't end up using them. For example this browned grass to ease the transition between dirt and grass, or this patchy grass I made for the same reason. I may still use them after some more experimenting with landscape blending, but for the time being they aren't being used in the level.

Asset Creation

The only assets I concepted before creating were the lantern and the purple flower. I probably didn’t need to concept the flower as it was being copied from life, but getting an idea of how I was going to push the colours for it was useful.

Other assets such as my lions head were concepted whilst sculpting.

I used plenty of reference for every asset I created, for example this is one of the boards I put together whilst texturing my donkey fountain:

I started with a rough silhouette in Max before sculpting my lions head, however the topology that resulted made unwrapping tricky.

Therefore I learnt from my mistake and spent time getting my donkey’s low poly right before taking it into z-brush to sculpt:

For the log I decided to re-topologise it in 3D coat because z re-mesh wasn’t giving me the results I wanted, and I didn’t want to lose important detail by going back to lower sub-divisions.

Before 3D coat:

After 3D coat:

For all my assets except the flowers and quill I used substance painter in conjunction with Photoshop to texture.

With the flowers I initially used photos but found the results noisy and unnecessarily detailed for such small assets. I found hand painting my albedo and height maps produced much cleaner results.

A comparison:

As always I would definitely have liked more time to polish my textures. A lot of the assets had to be made very quickly since engine work takes up so much time. Had I know that blueprints weren't going to be marked so much however I would have started making assets much earlier than I did, and left game-play towards the end of the project.

Here is an example of how Iterated on some assets to improve them when I had the chance:

Creating the waterfall and water were daunting at first, but after breaking down what needed to be done it was relatively straight forward. Here’s a screenshot of them in engine:

The waterfall could definitely be much better. I know it doesn't sit too well with the water, and is much more stylised in appearance. I would have liked more time on it, however the water and waterfall took me roughly a week and a half to create, and I couldn't afford to spend any more time on them.

I created lots and lots of test textures for the waterfall to work out what I needed to do to create it.

Here’s the final textures I created for them, along with the texture used for the water splash particle effect. I wanted to create more textures for the waterfall, but had to prioritise other work first.

Waterfall Alpha used to create streaky effect:

I would have liked to edit it further to remove seams even more, but unfortunately I didn't have much time for polish towards the end.

I blended a more subtle streaky/foamy effect with a white map to get a bit more variation in the waterfall. Since the player only sees it from far away I didn't create any other textures for the frothy effect, although I wanted to, but I had to prioritise other work.

Subtler texture blended in with plain white:

I also used a simple gradient to fade the edges of the waterfall out to avoid seams where it intersected the terrain.

Waterfall normal:

I used material instances for both the waterfall and water to quickly adjust values until I found a look I was satisfied with.

Waterfall Material:

To see the waterfall progress see my blog post Easter Blues.

The water splash particle I created to hide the seam where the waterfall intersects the water, and where the next level starts, uses this texture:

Water splash particle material:

The normal I created for the water was made after about 5 attempts to get anything decent:

Iterations:

Final Water Normal:

This is the height map I used in the end:

The water material:

The dancing lights were simple, the material was created in unreal, and then I experimented until I achieved a look I liked.

I went for blue and yellow ones, since they complimented the colour scheme of the level. For example blue complimented the orangey lights, and yellow complimented the purple shadows.

Dancing lights material:

Screenshot of them in engine:

Looking at them now I think I should have made them glow a bit more to match the rest of the scene:

Asset Evidence

And now a compilation of the assets I made in unreal, along with texture sheets, sculpts, renders, wire-frames and unwraps. I would have liked to have spent more time on the roughness for my materials, but that’s down to planning my time even better next time.

There are little fixes I wish I could have made such as bringing certain lilies above water so they all glow, just small things I didn’t get time for.

Or not placing all my assets for hand-in as I prioritised bigger stuff like plants over my unique mushrooms.

Here are all my assets in the final level:

Assets In neutral scene:

Wire-frames:

Textures: There's a lot I would like to have polished with these, such as hiding seams better, and making my roughness maps more accurate, however at the risk of sounding like a broken record, I did the best I could in the time I had. This is why I want to get better at creating high quality work quickly, when I spend the time I know I can get great textures, it's just a matter of making great textures in less time (and of course improving on the quality even further).

Donkey

Lion

Flowers; Daffodil, Lily and Daisy.

Ink bottle:

Lantern

I made a separate texture sheet for the lantern head on its own, so that I could add in a string alpha for hanging it from branches.

Pretty basic:

Glass:

Large rock textures (all rocks were made in a day since they weren’t originally my assets, but they turned out better than the persons rocks so I kept them in as I didn’t have a chance to polish)-So unwrap and textures would be much better given the time.

Large rock

Medium Rock

Small rock

Matinee Prop

Log (Used bitmap 2 material with substance and Photoshop for this)

Purple Flower

Quill

Unique Mushrooms:

With the unique mushrooms I actually made a blueprint that let me choose which variation I'd placed and what colour it would be:

The four different mushrooms I made I shared between two texture maps. So these mushrooms share these textures:

And these:

The tall thin mushroom uses fresnel to glow from within:

I also used an emissive mask for the small flat mushroom so that only the top of it glows.

Engine renders:

Renders:

I also made the moon

Which was just a simple sphere with a diffuse texture on it, and a plane with a gradient on it to create the atmosphere around it.

Overall there was no tri budget, it was more about using just enough triangles to make the asset look good, and giving more triangles to the larger more visible assets, such as the fountain, including more isolated assets which wouldn't cause so much of an issue in relation to tri density.

I took into account what the player would see, and as a result made assets less or more detailed depending. This isn't something I got straight away however, which Is why I ended up spending a bit too much time on some of the smaller assets, when I should have spent it making more assets to populate the level.

Some assets were iterated on constantly until I achieved the results I wanted. For example I unwrapped and baked my lion 5 times until I was happy that there were no obvious seams. Others sadly didn't get as much love, which is something I'm disappointed about it.

Lighting

So this didn't start out as well as it should have. I think because I'd set up a day-night cycle for the level, I didn't think to start lighting the scene at night time (because it'd switch to evening the moment I clicked play. Though of course this is really silly on reflection, using a day-night cycle makes getting lighting right very tricky, which is why in the end it was scrapped, and I focused instead on lighting a scene at midnight.

This was the original lighting during day time (not my work-the work I had to fix):

I took lighting over from here and did my best to fix it for the time being:

When I switched to a completely night-time scene, my life became a lot easier.

I made the environment intensity a little lighter, so that it wasn't so hard to see everything, and I ticked off my post processing volume so that I could get the lighting right, and then play around with exaggerating everything. I found that having strong effects in my post processing also meant that the colours in the scene were a little washed out and unvaried, for example at one point pretty much everything looked purple and pink

which is magical but also quite boring and means you lose a lot of the colour variation of the flowers, mushrooms and foliage.

Without post processing:

I struggled with lighting and I know it can still be a lot better. Mostly it was finding a balance between having it look good, but also keeping a relatively high frame rate. I started out by getting the directional light right first, then moved on to placing the smaller lanterns, before finally going in with the fairy windows. It was also hard to make it feel like a dark shadowy forest, but also keep everything fairly visible. This still needs work of course, but feedback helped a lot with this, because initially my lighting was like this:

But I was told to up the attenuation radius and lower the light intensity so that it made everything more visible, without adding in loads more lights, and having it unrealistically bright.

So my lighting changed to this:

I also changed the intensity of the god rays since I was told they were too bright, and made the green lights yellow to avoid having clashing colours

I used the fairy windows to create variations in colour with the lighting, for example I wanted to surround Titania in cooler colours, so that she stood out in a circle of warm light. I did this giving the trees surrounding her blue and purple lights from the windows.

Fairy windows creating colour zones:

To keep frame rate high I only set the trees to cast dynamic shadows since they are the main contributors to shadows in a forest, and I have them moving.

Lighting change:

I also put global illumination up in post processing to brighten the indirect lighting in the scene.

Without GI:

With GI:

And used reflection spheres to capture more highlights.

With reflection spheres:

Without:

Having the skylight on movable also made a huge difference to lighting

Stationary on the left, movable on the right.

I also edited the sky sphere from this:

To this:

With the god rays I tried to get them to point in the same direction as the directional light. Adding light shafts to the directional light also added to the magical look of the scene.

I couldn't get the spotlights to illuminate the floor where the god rays are, this is an issue I want to fix, but didn't manage too before hand in. Ideally I want to create a dappled look on the floor for them.

I used warm light from the lanterns to create an inviting contrast to the cool midnight light, and edited the shadow colours to be slightly purple to contrast with the greenish blue light of the moon.

Here is a screenshot of the level demonstrating this lighting set up:

I also changed the exposure from auto in project settings, then in post processing I added my own exposure- in auto exposure. Upped exposure bias to a setting that worked. This stopped it turning really dark after the player looks at a light and away from it.

I would have liked to do some colour grading too, but I'm going to have to do that after hand in.

Game-Play/Blueprints

In regards to creating blueprints I used tutorials and advice to achieve the results I wanted. I created many blueprints, a few of which were cut from the final level. The more I created them the more I began to understand how to go about solving certain problems, such as using variables to get events to trigger only once a condition had been marked true.

Used in the final level:

Level Start:

Blueprint in the widget for the level start main menu, it causes the image to fade to white and begin play. Originally it fades in to white then back out again to reveal the first clearing, I'm not sure why it isn't doing that any more, but I will fix it!

Level End:

Creates end screen, (a white image), this is an invisible image that fills the screen and fades in to white to end the level.

Play end credits and sound

Fades level out to white once you enter the trigger volume:

Pick up Script.

Checks to see if actor has collided with sphere collision around character:

Creates material instance on object (creates highlight material):

Detects if you are looking at the object

Fades out highlight the further away you are:

Pick up the object and create widget for it:

Detects if you have picked up object and if so triggers matinee event:

And the material for any object that will use this blueprint:

(they also need to have their collision changed to custom, and other settings adjusted).

Scrapped blueprints:

Day Night Cycle, plus changing light colour with time of day change:

Map change:

Flying in a certain zone:

Turn on Light/Turn Off Light: scrapped due to frame rate issues that couldn’t be resolved in time for hand-in.

Variable walk speed:

I scrapped this so that the player had to move at a leisurely stroll through the level, it made more sense with a walking simulator.

Basic Audio trigger:

First clearing audio triggers, plus creating tutorial UI:

Destroy first clearing tutorial UI (Press E):

Finally I was responsible for nearly all the placement of assets in the level, and creating all the collision to block the player’s path to certain areas, and fixing any collision that didn't generate correctly such as the collision for the trees. Here’s how I made the main collision for the players path:

I also fixed the issue with scale in the level by sizing up the character, and manually scaling assets that weren't quite right. (I Imported box references from max that were the right scale from real world proportions).

Finally I was in charge of optimising the level as much as possible. Such as balancing lights with frame rate, making sure light maps are the correct resolution, setting up cull distances, and so on. I also had to take peoples assets and do mock ups for them, like showing someone how to make grass using 200 tri's, not 2000.

This was a challenge since people put their work into the level so late that fixing any issues with peoples work and lights was very tight, and frustrating.

Ultimately I made a lot of mistakes this project, but that also taught me a lot. Of course there is still a lot that could be improved upon, which I will be doing for the competition hand in, so any constructive critique is always welcomed.