Monthly Archives: October 2019

Jetboard Joust Devlog #105 – Trailer Trash pt 2

Please wishlist Jetboard Joust on Steam here.

Well, the trailer’s finally done – thank God! As with the last instalment I ran into a few issues along the way and got distracted by (necessary) bugfixes. Here’s a brief summary of how things panned out…

1. Oh, Bugger!
The thing about doing a trailer is that it makes you capture and scrutinize a shedload of gameplay footage AND deviate from your ‘learned’ methods of playing your own game. The consequence of both these things is that you uncover new bugs.

I ran into a few new bugs when working on this section of the trailer, most of them related to trying out different weapon types on the bosses. The most significant of these was that there were major deficiencies in the raycasting algorithm I was using to work out damage caused by explosive weapons. Turned out I was only calculating collisions based on axis-aligned bounding boxes (fine for smaller enemies, not fine for the more complex outlines of some bosses). So I had to redo the algorithm to work properly with convex polygons.

There were also issues in the way damage was distributed via ‘proxies’ (e.g. different parts of the same boss) which had to be looked at and problems with the ‘particle storm‘ weapon which were very fiddly to debug.

2. Enemies and Weapons
The next twenty-ish second chunk of the trailer aims to give an overview of the variety of weapons and enemies in the game. I split the into eight 2.5ish edits, taking the timings from the ‘Enter The Gungeon’ gameplay trailer which seemed to pace the edits about as fast as you can get whilst still making sense of what’s going on. Managed to squeeze a brief glimpse of the weapon upgrade process in here too – I wanted to convey visually that there’s upgrade paths available for the weapons.

The edits were made by grabbing a shedload of footage and then whittling it down to the ‘best’ short segment. This was a time-consuming process. I set up levels artificially so I was only dealing with one enemy at a time – ‘real’ gameplay footage proved too confusing as there was just too much going on most of the time.

3. Bosses
The next twenty seconds or so provide a five second ‘reveal’ of the first four bosses. That’s only enough time to sneak a look at the first attack phase of each but I think that’s enough – I’m bit worried about giving too much away as it is. As with the previous stage, I had to grab a ton of footage and then whittle it down to the ‘best’ edits. I also had to bear in mind the transitions between the edits whilst grabbing the footage, i.e. make sure the player was positioned onscreen at a point where a smooth transition would be possible so the viewer’s eye never has to make a disconcerting jump. I found this was particularly important when dealing with very short cuts. I had to re-shoot a couple of edits in both this section and the last in order to get these transitions smooth.

4. Out With A Bang
Initially this section was simply a glimpse at the final boss fight but I felt that just cutting to titles after this seemed rather lame. Consequently I worked on a more ‘scripted’ take where the player flies over the final boss and then takes it out with an R.P.G. It’s not possible to do this in the actual game (the fight has about five stages) but it acts as a significantly more punchy ending to the trailer.

I had to do multiple takes of this to get it right as I needed to fly over the boss, shoot it accurately, and then position the player accurately in the middle of the screen all without doing anything visually unappealing like crash into the edges of the screen or flipping left/right too much. there were plenty of bloopers – including some where I fired the R.P.G. in completely the wrong direction.

5. Outro Titles
I ended up doing these in code and positioning them over the last section in After Effects using chromakeying. Initially I thought I’d animate them in After Effects but I just found this far too fiddly. Fortunately I was able to use the in-game logo ‘entrance’ without having to change anything much so it was mainly about animating the subheadings with the release date etc. I like the way the smoke is still dissipating in the background as the titles appear!

6. Back To The Beginning
Lastly, I went back to the start of the trailer. I was every-so-slightly over 1:30 seconds so I edited the initial titles down a bit. As some people had commented that the player appeared to come ‘out of nowhere’ a bit (I kind of agreed with this) I tried panning across to the player before going back to the alien babies/abductions. This seemed to work. I also made the player blink in this cut to bring him to life a bit (done in After Effects) and added ‘abduction’ and ‘mutation’ text with pans and zooms with the rest of the action.

Now I think we’re finally there. Next task – Steam page and assets!

Dev Time: 8 days
Total Dev Time: approx 300 days

previous | next

The Final Trailer In All Its Glory

mockup_3x
Raycasting Now Works Correctly With Convex Collision Polygons

mockup_3x
Fixing ‘Proxy’ Collision Detection Bugs

Jetboard Joust Devlog #104 – Trailer Trash pt 1

Decided to split this devlog in two as making a videogame trailer is a fairly long process, particularly when you don’t know what the hell you’re doing! Plus there’s been a couple of other things that needed attending to before I felt I could really progress as I’d like.

1. Audio FX
Firstly I had to do some more work on the in-game audio. There were a number of actions I felt required fx that didn’t have any (some pretty important such as unlocking weapons and worlds) and a couple I wasn’t happy with. I spent a couple of days on this. There are now around 270 individual effects files in the game, and that’s not including background music and loops! I think the audio is a pretty distinctive part of the game due to the fact it’s all produced from scratch on analogue gear – no stock samples here, folks!

2. Performance Optimisations
Secondly, as I was recording a bunch of action footage I began to notice frame-rate drop at a few points where there was really intense action on screen. This was exclusively at places where enemies that release a bunch of ‘offspring’ are destroyed by a weapon that kills the enemy and all it’s offspring in one fell swoop (e.g. the jetboard attack). The resulting explosion-and-bonus-fest was just a bit too much.

So, I worked on some optimisations for the above. This included adding object pooling for every object that’s generated when an enemy is destroyed, combining multiple smaller explosions and/or smoke clouds that are very close together (and instantiated in the same frame) into one larger one, and pre-caching of the terrain elements that pickups might hit as they fall (rather than calculating this every frame). I’ve also let some offspring ‘escape’ in the above scenario as I didn’t like it when absolutely everything got destroyed. I no longer see any frame rate drop now, even when there’s a shitload of fireworks going on, and I’m running a Mac from 2008!

3. Start The Trailer
Before starting the trailer I read through a number of very helpful blogs on the subject by Kert Gartner and M. Joshua. Here is a particularly good one.

3.1 Choose The Tools
One of the most useful practical tips I picked up from these was a pointer to an app called Screenflow that will grab 1080 game footage at 60fps even on my ancient Mac. I’ve been through a bunch of these screen capture applications (Snapz Pro, Capto, Screenium) but Screenflow is the only one that will do this. Capto is cheap and neat (this is what I’ve used for most of my animated GIFs and devlog posts) but will sometimes compress really heavily for no apparent reason, Screenium is almost as good as Screenflow (and cheaper), will let you record a set area AND remembers this rea (really useful) but it still compresses a bit even at the highest settings (plus I found it’s editing tools a tad clumsy and prone to crashing). For the purposes of recording gameplay footage for trailers Screenflow definitely comes out tops.

3.2 Intro
Probably the hardest part of the trailer to get right is the first 15 seconds. You don’t want to lose the user’s interest and you have to attempt to communicate the core mechanics of your game in as short a time as possible (without being overly didactic). I’m still not sure I’m 100% there but I think what I’ve come up with does a reasonable job of communicating abductions, mutations, the jetboard attack and the general carnage of the game in that timeframe. To get this footage I used a combination of scripted events (i.e. faking things through coding) and playing through a set sequence over and over again until I managed to one-shot all the enemies in a way that looked effortless and ‘readable’ enough.

3.3 Shock & Awe
After this initial intro section comes just under another 15 seconds of what I’m referring to as ‘shock and awe’. This is a high-octane segment that focusses mainly on the destruction wrought by the jetboard attack but also features a couple of other weapons. This is all ‘real’ gameplay footage, I just recorded a load of stuff to get a variety of enemies and palettes. I deliberately try and move the action from one side of the screen to the other here.

3.4 Breathing Space
Lastly comes just over 10 seconds of ‘breathing space’. A longer cut in which we see the destruction of a boss, the opening of a treasure chamber and the discovery of a new weapon. This section has a certain amount of scripting (reducing the bosses health and getting rid of some onscreen distractions) but is largely the result of playthrough after playthrough to get things looking slick and pulled off in as short a time as possible.

I’ve tried to keep the player’s position onscreen consistent between cuts so that the viewer’s eyes can easily track what’s going on. I’m also zooming/panning across the action where appropriate in order to avoid ‘dead’ areas of screenspace and create variety. I thought this might be overly distracting but it doesn’t seem to be.

So now we have just under 45 seconds of trailer done which is approximately half of it. Next step, around 20 seconds on enemies and weapons, 20 seconds on bosses, and a five second closer!

Dev Time: 10 days (inc 2 days audio and 2.5 days performance optimisations)
Total Dev Time: approx 292 days

previous | next