The Dev Log of J.H. Grace

Here I blog my successes and failures.

[mihaivdev]

[Tutorial] [Intermediate] Recreate Need For Speed Carbon’s Autosculpt in Unreal Engine 4 and Blender using Shape Keys and Morph Targets

Before following this tutorial, please note that this is not a step-by-step tutorial and it assumes you have some familiarity with Unreal Engine and Blender or a similar 3D package. If you’re a complete beginner Tesla Dev has a good video tutorial on Morph Targets ( https://www.youtube.com/watch?v=r0Jlu3mI6v0 )

Intro: When I first played Need for Speed Carbon I was amazed by the Autosculpt system and wanted to know how they’ve done it. While I didn’t find out at the time, while playing with Morph Targets in Unreal Engine recently, I figured they could be used to recreate a system similar to Autosculpt.

For those who haven’t played the game, Autosculpt was a feature that allowed you to do fine visual tweaks on different car parts such as wheels, bumpers, spoilers, side skirts, etc by adjusting a few sliders. These adjustments allowed you to vary several properties such as size, position, rotation, pulling on various groups of vertices to achieve numerous unique variations of a car part. 

Here’s what we’re going to achieve:

Part 1 - Blender First of all we have to prepare the mesh. I’ll be using Blender for this but the following is achievable in most 3D packages such as 3DS Max and Maya. You can also make this in Unity instead of Unreal Engine.

The key feature in Blender which will help us achieve our goal is called Shape Keys. Shape Keys allow you to define several variations of a mesh and smoothly blend between them.

We start with a fairly simple car rim and tyre as one mesh with the modifier stack collapsed. This is needed due to Shape Keys not working with modifiers. So if you built your rim using the Array modifier you’ll unfortunately have to apply it to continue.

It’s important to keep your topology as simple as you can, avoiding any unnecessary topology such as no loops on the spokes like below. You’ll see why in a moment.

Important! Before you go any further make sure your mesh’s scale is applied as it will cause issues if not. To do this select your mesh, press Ctrl + A and select Scale.

Now go to the Data tab in the Properties panel and under Shape Keys click the “plus” icon to add a new shape key. This will add your Basis Key which defines the base shape of your mesh. Then click once more to add a new key, double-click it’s name to rename it and call it Split, followed by setting it’s Value slider to 1. 

Now press Tab to go into Edit Mode and tweak your mesh to look similar to below. After you’re finished exit Edit Mode.

Note: This is where not having an edge loop on the spoke helps as you can bend it out of place by just moving the top vertices retaining its overall shape. This is a bit tedious and while it might be tempting to use the Array modifier to model just one spoke of the rim and duplicate everything it is unfortunately not possible due to the incompatibility of Shape Keys with the Array modifier.

After modifying your mesh like above you can drag the Split slider and see how your mesh blends nicely between the 2 variations.

Now you can move the slider to 0 and click on “plus” to add another Shape Key. 

The process from here is the same:

1. Add Shape Key 2. Set the value to maximum 3. Enter Edit Mode > Modify your mesh to represent the maximum deformation for that Shape Key 4. Exit Edit Mode and play with the new Shape Key to check if you’re happy

I’ll quickly demonstrate all my Shape Keys and let you to either try and do the same or come up with your own.

Width - modifies the top width of the spokes

Poke - pokes the centre of the rim in

Offset - moves the entire spoke assembly in and out (deep dish style)

Twist - rotates the centre of the rim only

Radius - changes the radius of the rim (hint: use Proportional Editing to make this)

Tire - changes the tire width (if you want fat Lambo tires)

All that’s left to do in Blender now is to export this as a .fbx file and import into UE4. So go to File > Export > FBX and use the below settings.

This is for the Blender part now let’s jump to Unreal. Part 2 - Unreal

In Unreal the feature that allows us to use the Shape Keys we created is called Morph Target. These are a property of Skeletal Meshes so we’ll have to import our wheel like one.

In a project of your choice import the .fbx you exported earlier and match the settings underlined in red. You can leave everything else as default.

After you’ve done the settings press Import and double click your rim in the Content Browser. Go to the Mesh tab and your Shape Keys/Morph Targets are on the right. You can now adjust these as you did in Blender and combine them to achieve rather unique variations of your mesh. 

As a bonus UE4 allows you to move the slider negatively down to -1 so you can get the opposite result to the original intent of your shape key. This might end in some interesting or just outright funny results!

To adjust these sliders within a game, use the Set Morph Target node in Blueprint which you can drive using a UMG slider or whatever else you prefer.

Challenge: Try and use the above techniques to tweak another car part! Did you know?: You can use these techniques to also make customisable doors, windows, stairs and many other things. 

Hope this was helpful and if you have any questions send me a message and I’ll try to answer.

[jhgrace-blog]

This is really cool!

We use a similar system in Automation, though much simpler (only offset and scale for the rim are in the morph target mesh, all the tyre settings are procedural meshes in UE4, and there aren’t any of those awesome modifiers you have there).

You would have noticed though that the normals of the morphed objects in UE4 get more and more mushy as you bring the morph target sliders away from 0 (Your last image shows this off pretty well). This is because actual .FBX files don’t support updating the normals in the file format itself.  What this means is:  the morphed meshes will have the same normals/tangents as the base mesh.  This will make everything look Wrong™. To get around this, I select all the hard edges and ‘split’ them so that smoothing groups are in separate mesh elements, and then I just set all edges to smooth.

This allows us to keep our nice hard edges while still getting the awesome rim offset/scale options! it does mean you’ll have to rebuild all your morph targets though as the base mesh will have changed its topology enough that everything will break if you don’t:

[jackw-gamedesign]

Undoing Unhealthy Game Design

As Game Designers, we have a responsibility: To create an enduring, fun, and captivating experience for our players. Within that experience we can pursue various mechanics or stories to elicit emotional or satisfactory feedback from our players, but ultimately we have something to keep in mind as well: our livelihood.

Game Design is a career, it is an industry, and as a result we have to encourage players to play our games so that we can earn an income. The question I ask is… can we create ethical play when we are trained to keep a player engaged ( / addicted) to our games? 

(Image source: https://www.youtube.com/watch?v=z6ZrYQNjLs4)

How long do we expect players to sit and stare at a screen to play our titles? Is 30 minutes enough to make a profit? 1 hour? 4 hours? 12 hours? How long should our loops be? How far until the next save point? Is a 2 hour cutscene too long or a 3 hour boss battle asking too much?

Ultimately, do we exploit our players’ health for our own profits? (A secondary argument: It is the player’s choice how long they play a game for)

Some games are coming through that are mindful of this (ie Pokemon Go), and some companies run health awareness messages before their games begin, but on a whole it is not part of design and it is important to keep these things in mind. Are we encouraging unhealthy and potentially damaging behaviour? Can we develop systems within our games that reward our players for taking a break?

I believe ethical design should become common practice in game development and could be harnessed to create engaging experiences that encourage our players to upkeep healthy habits.

[jhgrace-blog]

Extra Credits did a great video on humane game design that I highly recommend you check out: https://youtu.be/GArkyxP8-n0 The bit I took away as the most important to me from that video is the idea of 'Exit Points', where you add pauses and moments where the player has a choice to either keep going, or stop. I like exit points because not only does it give the player an out if they so wish to, but it can be useful as a tool to maintain pacing, as instead of having to ramp up the "awesome' for each successive encounter you can instead design around these exit points, to create a pacing curve that is more manageable.

[jhgrace-blog]

working on some transparency post process effects in UE4

[jhgrace-blog]

Project: Starlifter

So I’ve decided on a name and logo for a project I’ve been intermittently working on.

I figured if the idea is out there somewhere then I’d work on it more often.

It’s a hybrid 6DOF-RTS game set in the far far future, where a group of people have spent several thousand years living in a huge artificial space station without any memory of how they got there, and the game has a different level set in a different extra-terrestrial mega-structure (starlifting statites, Dyson swarms, black hole starships, Shkadov thrusters/nicoll-dyson beams, Ringworlds,   Krasnikov Tubes, Arcologies and Ecumenopolises, Matrioshka brains, Shell worlds/Disc worlds/Loop worlds etc...

I haven’t got much in the way of a functional prototype yet (I’ve got 6DOF movement, swapping between ships, selecting/focusing on groups), still need to work on constructing ships + ai movement + formations, but I do have a logo:

It’s meant to look like a Shkadov Thruster or a Nicoll-Dyson Beam, but there’s still a lot i dont like about it (it looks very 90′s).

Anywho, that was all.  Move along.

[jhgrace-blog]

you could almost say @AutomationGame was isometric

[jhgrace-blog]

that feel when the camera breaks and accidentally does something cool

[jhgrace-blog]

Baked some lightmaps in UE4 earlier.  Forgot to fix the mesh UVs.  Oops.

[the-huskarls]

Devlog - Finding the Fun

The end of alpha is just around the corner, we are at a crucial time of trying to find the fun factor of our gameplay. Radical tests and experiments are what we need now.

Keep reading

[jhgrace-blog]

I did the thing Sources: 1, 2, 3, 4, 5, 6, 7, 8

[jhgrace-blog]

factory: hacked

[jhgrace-blog]

A teaser of things to come

[jhgrace-blog]

So I made a gradient mapping function for UE4.

Been wanting to do this for some time, but never could figure out how to do it.

I won’t go in to the details of the function itself because I’m trying to get it on the UE4 marketplace, but it works like this:

[jhgrace-blog]

Global-Illumination pre-pass renders look like point-clouds

[jhgrace-blog]

Scene I put together for an art test for Camshaft Software I had a lot of fun making this.

(Camshaft gave me permission to post this)

Here you can see how the fresnel effect is used to dull-out the displays.  the oscilloscopes fade out, the table LCD has a low field of view, and the CRT displays have a subtle colour fade effect:

DOF was used to blur objects that were either too far away or too close to the screen.  Grain was also added to give the objects a 'worn' look:

I took screenshots of my desktop to use for the monitors here.  The oscilloscope graphs were generated in Adobe Illustrator.  A mask texture was used vertically and horizontally for the graphs to give them their refresh rate.  The CRT displays use a 29x multiplier for the panning mask to give them that flickering look.  (why 29?  half of their actual refresh rate, because 60 was too much, minus 1 to give it an uneven look):

The table was given a detail noise texture to make the reflections look a bit more realistic for this type of material:

Most of the graphs displayed on these monitors are from Wolfram|Alpha.   I'd really like to go back and improve some of these interfaces, they seem a bit blocky to me:

There's something beautiful about just looking at the lighting in UE4 too.  I wasn’t a fan of lightmaps and baked lighting before (I was a CryEngine fanboy for a while), but UE4 really changed that for me:

You'd never thought that flat colours and low-res geometry could look so good in the end, yet here we are.  Totaled about 24,000 triangles and 10MB of meshes/textures/lighting:

Took about 3.5 weeks. The first two weeks were a bit sketchy, but after I got my internet sorted out (and therefore didn't ragequit when google images wouldn't load when trying to look for reference material, because holy smokes does that get old quickly) things started to go smoother.  Clocked in at about 40 actual work hours total, though a lot of that I think was spent on baking lightmaps.  Jeeze, the issues I ran in to with that.. Never have I had to go in and snap each UV for a modular floor to stop lightmass from making weirdly stupid errors with the bounce lighting.  Still eventually had to just replace the whole walls and floor with a single mesh.

Final post-processed image here:

Had a lot of fun using static colouring instead of textures for a lot of this, I was using Alien: Isolation as a lot of reference for the props.  I loved the way they hit that aesthetic right on, but it did bug me that everything looked too clean: There wasn't enough 'noise' in the materials for me in the textures..

The image Camshaft gave me was a good challenge:

 The low resolution meant I had to think a fair deal about how light bounces around in the smaller areas, what some of the smaller details might be..  I didn't even see the slight curve in the wall until a few days in, and it wasn't until just last week I think that I saw the wall behind the far rack system lead out into another area.  The lighting was a good brain teaser too, I wasn't sure how much of that was light bleed into the lens, how much was actual albedo, or how much was from the lights.  I think I got a good balance between post-processing, albedo, and lighting colouration in the final product.

Things I'd like to continue on would be improving the UVs for everything, since I basically said 'screw it' for everything that didn't need a texture, adding some grunge by way of decals to areas like the front of the tables to demarkate where peoples' arms would polish the surfaces, water marks and light-wash under some of the things that lean against the walls, and really increasing the resolution of that chair because I didn't remember I wanted to do that until after I took all the pretty screenshots and I really shouldn't just keep iterating on this one thing, everything has to be finished eventually.  I'd really like to have another go at a lot of the interfaces, too.  Especially that central one on the large oscilloscope machine.  I think I got a nice minimalist finish, but nothing beats a good high-resolution detailed interface.

If I can find out why when I package the project the camera is stuck at 0,0,0 and I can’t move or do anything then I’ll post up the download link for people to have a walk around.

[jhgrace-blog]

And now for some programmer art

So I’m trying to teach myself some UE4 Blueprinting, and I’ve come up with this thing.

I followed the UE4 Blueprint Quickshot series, and I got to the end, and I was loking through the comments, and I saw someone had posted something interesting.  They wanted the blueprint to spawn objects randomly, but with their z-axis clamped down on whatever mesh was below, so they could do what I’ve just done now which is make objects random but latch on to whatever was below them. Using what I learned in one of the earlyer videos, I knew I could use a line trace to get the location of whatever mesh was below the object and make that location the place where the object would be spawned.

I’m still not very good at describing this because I’ve only been blueprinting for like 2 days, but messing around with things is fun and you can have a look at the pictures to see how I did it :D

[jhgrace-blog]

Rover

Rendered in UE4 Modeled in Maya Post-processing and materials all done dynamically in UE4

Started this yesterday, finished today.  Still a lot more I could do, but I’m trying to work quickly to improve on my modeling skills.

Reference:  [1] [2]

And actually while I’m at it, this was an extension from a project waaay back, that you can see here: [link]

[jhgrace-blog]

Super Secret Progress