Pillars of Eternity II - Evolution and Paradigm Shifts

00:29
[Music]
00:36
all right so I gave this a very fancy
00:40
title breaking the mold RPG evolution
00:44
and paradigm shifts and pillars of
00:45
eternity two dead fire really what this
00:49
should be called is we tried to do some
00:51
new things and role-playing games some
00:52
of them worked and a bunch of them
00:54
didn't let's find out how all right what
00:58
in the heck is this talk about this is
01:00
ultimately it's a post mortem four
01:02
pillars of eternity two dead fire it's
01:05
about looking at the assumptions that we
01:07
made going into the sequel and what
01:09
proved right about those assumptions and
01:11
what proved ultimately to be very wrong
01:13
for those of you who are not familiar
01:15
dead fire is the sequel to pillars of
01:17
eternity these are both crowdfunding
01:19
games that obsidian entertainment made
01:21
there are four PC they're party based
01:23
isometric role-playing games in the
01:25
style of Dungeons & Dragons Arab I
01:27
Swindell Baldur's Gate and Planescape
01:29
torment they have a Dungeons & Dragons
01:32
style character development system so
01:34
you advanced over several levels with
01:36
different classes pick various abilities
01:37
as you go up in level we use 2d
01:40
backgrounds with 3d characters and
01:42
lighting systems and they were both made
01:44
in unity with our own proprietary tools
01:47
especially for game data editing and
01:49
dialogue what are we going to be talking
01:52
about so first we're going to talk about
01:55
the feedback that we received on pillars
01:56
of eternity because the feedback that we
01:58
receive proved to be the foundation for
02:00
our expectations for the sequel some of
02:03
that was good and some of it turned out
02:04
to be very bad changes that we made
02:07
based on that feedback and the reasons
02:08
for doing so and how we made those
02:10
changes I'm gonna talk a little bit
02:12
about the backer beta which was fairly
02:14
short the launch itself and what
02:16
happened after the launch I'm going to
02:19
go on to some controversial features
02:20
that we approached during development
02:22
some of these were controversial within
02:24
the team most of them are controversial
02:26
in the community but all of them are
02:29
worth talking about I think for for the
02:31
benefit of development in the future and
02:33
also a long-term post-launch support
02:36
last week sorry two weeks ago marked the
02:40
one-year anniversary of dead fires
02:42
launched we released our 5.0 patch
02:45
obsidian has become very dedicated to
02:47
doing long term support for its tight
02:49
especially pillars of eternity so we did
02:51
a lot of post launch patching and DLCs
02:54
and that made a big difference in the
02:56
long run little note here so I'm gonna
03:01
be talking about decisions that I made
03:03
that I'm completely responsible for
03:04
there are other decisions that were made
03:06
above my head it might sound like I'm
03:08
throwing people under the bus I am a
03:10
little bit but in order to actually do a
03:13
post-mortem on this and really talk
03:14
about the decisions that were made and
03:15
why I felt the need to sometimes refer
03:18
to decisions that were made above my
03:20
head because that will give you a
03:21
complete picture of what actually
03:22
happened during the development process
03:25
for those of you who do not know who I
03:27
am
03:27
I'm Josh Sawyer I'm the studio design
03:30
director at obsidian I started at Black
03:32
Isle Studios in 1999 I worked there
03:34
until 2003 I worked on primarily the
03:37
Icewind Dale series of games which is
03:39
one of the inspiration points for
03:40
pillars of eternity and dead fire
03:42
between 2003 and 2005 I was at Midway
03:45
San Diego working on a gauntlet game
03:47
that did not go super well then I went
03:50
to obsidian I've been there since 2005
03:53
either as a senior designer a lead
03:55
designer a game director and now for the
03:57
past few years the studio design
03:58
director currently I am between
04:02
directing projects I'm helping on
04:04
supporting internal teams as a studio
04:06
design director and I'm also working on
04:07
the pillars of eternity tabletop
04:09
role-playing game I did not expect I
04:11
would actually ever get a chance to work
04:12
on a tabletop role-playing game
04:14
professionally but now I am and it's
04:16
great so I'm going to talk about some
04:19
assumptions that we had going into the
04:22
sequel that's Sven bingo there by the
04:25
way we are in a space with divinity
04:28
original sin to kind of spend from
04:32
Larian it's a very great guy he's a very
04:34
forthcoming person with any data that he
04:36
has about the success of divinity
04:38
original sin and original sin - and one
04:41
of the things that came up when we were
04:43
talking with him is he had mentioned
04:45
that there was actually relatively low
04:47
overlap between the audience's for
04:49
original sin and pillars of eternity
04:51
which we thought was kind of interesting
04:53
but then again they are fundamentally
04:55
different games from a combat
04:56
perspective which is a big deal to a lot
04:58
of people divinity games are turn Bay
05:01
pillars of eternity is real time with
05:03
pause because that's what the Infinity
05:04
engine games were divinity original sin
05:08
to divinity original sin to the first
05:11
game reviewed well sold very well the
05:14
second game reviewed incredibly well
05:15
sold phenomenally well so we were like
05:18
well we reviewed and sold well so if we
05:21
have a similar approach to refining the
05:23
formula of the game we will review even
05:25
better and sell incredibly well Wow
05:30
wrong that did not happen but it worked
05:34
out pretty well anyway so now we're
05:35
going to talk about how we looked at
05:37
these criticisms and tried to move
05:38
forward
05:39
one of the first criticisms by the way
05:41
when I when I talk about these
05:42
criticisms
05:43
I'm only listing the ones that really
05:45
seem to come up a lot in user reviews
05:47
and in professional reviews there are
05:51
many many other complaints that are
05:54
totally valid that people could make
05:55
about these games but these are the ones
05:57
that kept coming up and we thought it
05:58
was very important to try to address
06:00
them one of the first is that combat was
06:02
hard to follow again
06:03
pillars of attorney was a real time with
06:06
pause game when Baldur's Gate came out
06:09
it was kind of a revolutionary that it
06:11
used real time with pause for RPG combat
06:14
for most people it was the first game
06:16
they had seen that used a system like
06:17
that
06:18
RTS a--'s were very popular in the late
06:21
90s early 2000s not very popular right
06:24
now
06:24
so when pillars of eternity came out
06:27
obviously all the veterans of those
06:29
games adapted fairly quickly but there
06:31
were a lot of people that had no
06:32
experience with this type of game the
06:34
pace of the combat was very fast and
06:36
there were a lot of problems that we
06:37
tried to address one of the things that
06:40
we approached was how we rendered a
06:42
visual effects as you can see from the
06:44
big mass of color in the middle of the
06:45
screen that's a couple of spells going
06:47
off what's going on under there who
06:50
knows um that was an experience that a
06:53
lot of people had playing pillars of
06:54
eternity where they would cast a spell
06:55
it would go off and then they would say
06:57
like I think someone's dying in there
06:59
but I don't know so we actually changed
07:02
our rendering pipeline to reduce that
07:04
visual noise and helped a great deal
07:05
we have unengaged a mechanic in the game
07:08
called engagement this is a crowd
07:10
control mechanic that tanks can use to
07:12
lock people down it proved to be
07:15
noisy in the original game you've had a
07:17
lot of UI feedback that was really
07:19
confusing to a lot of people so we made
07:21
changes to how engagement worked to just
07:23
reduce that overall clutter it also
07:24
helped to make the tank characters a
07:26
little more distinctive from the non
07:27
tank characters and then the most
07:30
controversial thing that we did is we
07:31
moved from a 6 character party to a 5
07:33
character party this was one of the more
07:36
criticized things in the game because I
07:39
Swindell in baldur's gate were 6
07:41
character party games and that felt very
07:43
traditional to a lot of people moving to
07:45
five characters just irritated a lot of
07:47
people but ultimately it was an
07:49
important change for us to make to help
07:50
make the combat easier to follow and
07:53
finally we had an AI customization
07:55
system the customization system really
07:58
added a lot both for players it allowed
07:59
them to build their own custom AI sets
08:01
but it also allowed us to do a lot on
08:05
the NPC side the monster side so for
08:07
players it made combat easier to manage
08:09
because they could automate a lot of
08:10
things and for us it allowed us to make
08:12
better AI behavior and that helped a
08:14
great deal this is just a contrast right
08:17
here this is a typical battle actually
08:19
that's not a typical battle that's a big
08:20
battle from pillars and here is a battle
08:23
from dead fire the visual effects
08:24
rendering changes made a big difference
08:26
there's a lot of visual clutter that's
08:27
gone from engagement and things like
08:29
that it did help a lot another criticism
08:33
is that the stronghold felt very
08:35
disconnected from the main story i think
08:36
that was a totally fair criticism
08:38
we included strongholds because people
08:39
liked strongholds and it turned out we
08:41
didn't have the resources to really
08:42
integrate them as well as we wanted to
08:44
so we decided to make the stronghold a
08:46
ship this was something that we argued
08:48
about for quite a while at first we
08:50
thought about doing multiple little
08:51
strongholds all around the archipelago
08:53
ultimately it seemed like it made more
08:55
sense to just say look you're gonna have
08:57
to travel by ship around the game let's
08:59
just make that your stronghold it's
09:01
necessary for traveling so you're gonna
09:03
keep going back there it's a convenient
09:05
place for all of your companions and
09:07
other little NPC buddies together so
09:09
that way if you want to swap party
09:10
members out you're always going to wind
09:12
up going back to the ship at various
09:13
points and it's very appropriate for a
09:16
pseudo age of sail game pillars of
09:18
eternity is not it's really not a
09:20
medieval game it's more supposed to be a
09:22
early modern game 16th century plus and
09:26
so going into the age of sail felt like
09:27
a natural
09:28
from the original game which was early
09:30
16th century another big criticisms that
09:33
the stronghold mechanics are boring also
09:36
can't argue with that
09:37
they weren't that interesting so we
09:40
added a ship crew and navigation
09:41
mechanics that was relatively
09:43
straightforward the crew was an easy
09:45
place to put NPCs that you found for
09:47
example if you had a quest that ended an
09:49
NPC might join your crew as a reward
09:51
that seemed like a really cool neat
09:53
thing to do and then we got into the
09:55
ship to ship combat system which is
09:57
again one of the more controversial
09:59
things that we added and we're gonna
10:01
talk about that later voiceover was very
10:06
inconsistent pillars of eternity is a
10:09
big role-playing game with tens of
10:11
thousands of lines of dialogue and it's
10:14
very expensive and it's very
10:15
time-consuming to voice all that
10:16
dialogue so we had partial vo for some
10:20
characters almost none of the characters
10:22
in the game actually had full voice over
10:25
this was not liked by a lot of people
10:28
only the chapter breaks were even
10:30
narrated so any of the prose text that
10:33
we wrote was just plain text that you
10:35
had to read streamers which were
10:37
starting to become very prominent in
10:38
about 2015 when this game came out they
10:41
really really hated seeing a lot of
10:43
unvoiced text and it turned a lot of
10:46
people off from streaming the game just
10:48
kind of frustrated people and right in
10:51
the middle of dead fires development
10:53
divinity original sin 2 came out and it
10:56
had Fulvio for everything so that set a
10:59
new standard I hate them for it they did
11:03
that though and we had to respond to it
11:04
we couldn't ignore it and we're gonna
11:07
get back to that too there were too many
11:10
trash fights partially my fault actually
11:14
largely my fault there were does
11:18
everyone know what a trash fight is it's
11:20
a fighting role-playing game where
11:21
basically it just feels like it doesn't
11:24
offer anything tactically or
11:25
strategically that's different from the
11:27
fights around it it feels like a waste
11:28
of the players time just to drain their
11:30
resource you go through a dungeon and
11:32
you'll have like you know every square
11:33
inch-- is populated by essentially a
11:35
carbon copy of a fight that you've
11:37
already seen there were a lot of these
11:39
unfortunately in pillars
11:41
so in debt fire we moved so that block
11:44
outs that people did for their levels
11:46
had to including counter maps so we
11:47
could see where the encounters would be
11:49
documents also had to define those
11:51
encounters so for example if someone
11:53
said well I only have three encounters
11:54
in this level and then we found out that
11:57
they were all essentially the same
11:58
encounter right so I cut one of those
12:00
encounters or change it or change both
12:02
of those encounters into something else
12:04
but a lot more work went into the
12:06
upfront design of saying please don't
12:08
flood every level with trash fights
12:10
because the players they just don't
12:11
enjoy it and then encounters received a
12:15
system design pass so we have area
12:17
designers area designers know the
12:20
systems of the game to a certain extent
12:21
but the fine tuning of combat was more
12:24
on the system design side so the idea
12:27
was when an area designer would put an
12:29
encounter in a level than we would later
12:31
have a system designer go through and
12:32
they would do a pass on tuning that
12:34
unfortunately there was so much system
12:37
design work to do that only a small
12:39
number of encounters received that
12:41
tuning passed prior to launch and that
12:43
turned out to bite us in the ass because
12:45
the game wasn't tuned very well the
12:49
environments felt too static again this
12:51
game was a 2-d game with 3d characters
12:53
in pillars of eternity
12:55
none of the vegetation was dynamic there
12:58
wasn't dynamic cloth really except for
12:59
on the characters the lighting felt a
13:01
little stilted and often the shadows and
13:03
the characters didn't integrate into the
13:05
background it got what I would call the
13:07
scooby-doo effect where you can see the
13:09
the moving characters that feel very
13:11
separate from everything else in the
13:12
scene we approach this in a couple of
13:15
ways
13:16
some of it was graphical and some of it
13:17
was behavioral so we added a new NPC
13:19
scheduling and behavior system
13:21
this allowed characters to sort of have
13:23
their own lives and routines it made the
13:25
environments feel more alive because
13:26
instead of characters just sort of
13:27
standing in one place all day long they
13:30
had their own routines that they would
13:31
go through this was mostly used in
13:33
cities and things like that but it was
13:35
also used for monsters and patrols and
13:37
then we did a lot of rendering work
13:40
together
13:41
it's funny because like once the game
13:43
got to this point its it kind of just
13:45
looks like a 3d game now that just has
13:47
an isometric camera but that background
13:49
is actually all 2d but it has 3d
13:51
lighting integrated into the 3d
13:53
characters and everything like that so
13:56
we move to physically based rendering we
13:58
moved from unity four to unity five it
14:00
seems like a no-brainer to go to
14:01
physically based rendering and start
14:02
using various tools to just make
14:05
everything look a lot better on the
14:07
character side we introduced vegetation
14:09
in cloth physics all the trees there in
14:12
the background those are all dynamic and
14:14
they blow with the wind the wind was the
14:15
dynamic element that would move cloth
14:16
and vegetation it added a ton a huge
14:21
amount of work went into unifying the
14:23
character shadows and lighting into the
14:25
world so that when shadow when
14:27
characters moved through darkness they
14:28
really felt like they were integrated
14:30
into the environment instead of being
14:31
separate from them we had a parallax Lee
14:34
excuse me we had a parallax layers
14:36
parallax layers obviously this is very
14:39
old technology going back to like 8-bit
14:41
games and even earlier where you have a
14:42
scrolling layer in the background or
14:44
multiple scrolling layers that move at
14:46
different rates of speed this is most
14:48
useful in areas that have great height
14:50
where there's a big distance to
14:53
something down below it was actually a
14:55
surprising amount of work to get
14:56
parallax layers to work given our weird
14:58
rendering stuff but it did add a lot in
15:00
the areas that we used it and then we
15:02
had a new water technology the water in
15:04
pillars one was basic and functional we
15:07
licensed a water technology called Seto
15:09
and then we altered it pretty
15:11
significantly it made the oceans look
15:12
really beautiful so that when you're on
15:14
the coast or on a ship which is a lot of
15:16
the places in the game the water really
15:18
looked great yeah so this is a criticism
15:24
of both games we did our best to try to
15:26
reduce them but it was a very difficult
15:28
challenge load times were very bad this
15:31
is especially difficult in areas where
15:33
you were going in and out of buildings
15:35
over and over we switched to a hub based
15:39
loading system so for example in a city
15:42
or any area that had a large interior
15:43
with several interiors we would take all
15:47
the child areas and cache some of the
15:50
data about those areas that when you
15:51
transitioned in and out that data was
15:54
already sort of pre-loaded that made
15:56
those transitions in big areas a lot
15:57
easier there was a ton of memory that
16:01
was being taken up in a ton of loading
16:02
that was happening with unity data all
16:04
of our gameplay data all of our spells
16:06
all of our creature
16:07
were in unity asset bundles and we would
16:11
load those dynamically every time we
16:12
would load into an area that was very
16:15
time-consuming expensive so we moved all
16:17
of that out of the unity editor into our
16:20
own standalone editor that was using XML
16:23
or JSON so it's text all that stuff
16:25
would load at the start of the game and
16:27
that that helped eliminate a lot of the
16:29
load time as well and then finally the
16:32
save load system was rewritten this
16:34
might not seem like that big of a deal
16:35
but when we rewrote the save load system
16:39
we also reauthorization trying to reset
16:51
all these variables and do all this
16:52
crazy stuff companion reactivity was bad
16:57
I think people liked our companions
16:59
overall but there was a lot of plate
17:01
there were a lot of places where
17:02
companions either wouldn't react to each
17:04
other NPCs wouldn't react to them they
17:07
wouldn't leave the party it was kind of
17:09
like it felt like they were always along
17:11
for the ride but they weren't really
17:12
fully integrated into the game as well
17:14
as they could have been because obsidian
17:16
is known for writing this is something
17:18
that we thought was very important to
17:20
address so right away we had more
17:22
in-depth budgeting for writing what I
17:24
mean by this is that all the writers sat
17:26
down and we said okay how big our
17:28
companion is going to be how are we
17:30
going to break up the writing
17:31
responsibilities in terms of their hubs
17:34
which is their main question node their
17:37
interjections in conversations their
17:39
banters with each other the
17:41
relationships you can have with them
17:42
their personal quests so we broke those
17:44
down into ranges of nodes nodes are the
17:47
unit of measurement that we sort of
17:49
talked about when we talk about how much
17:50
writing there is to do for a character
17:51
and we budgeted that very early we also
17:55
staged the writing so on most of the
17:58
games I've worked on all the companion
18:00
writing was done very late in
18:01
development usually because we reasoned
18:04
oh well when it's done very late then we
18:07
can react to everything that we've
18:08
written earlier in the game the problem
18:10
was companions are huge and that's an
18:12
enormous amount of work so after we did
18:15
the budgeting we also broke up the
18:16
writing for companions into three phases
18:18
that made the most sense for development
18:21
this removed an enormous amount of
18:22
burden from us because we could focus in
18:24
each stage development on a different
18:26
aspect of companion writing and it made
18:29
the the tail end of a companion writing
18:31
a lot easier for everyone we also
18:35
include a companion leave conditions
18:36
companions were no longer going to
18:38
follow you to the ends of the earth if
18:40
you pissed them off they would get into
18:42
an argument with you they might just get
18:43
up and leave immediately but they were
18:45
unique to each character and it added a
18:47
lot of sort of depth to them that was
18:50
lacking in the first game and then the
18:52
final thing is that we added a new topic
18:53
system with reactions and relationship
18:56
break points this did not work very well
18:59
I think it was a cool idea and I'm gonna
19:02
talk more about it later a lot of people
19:07
criticized that it had a very slow start
19:08
pillars of attorney had a very slow
19:10
start and your motivation was very
19:12
ambiguous the mean villain of the game
19:15
was kind of hidden for most of pillars
19:16
of eternity he was a very cool villain
19:18
Theo 6r Catan but exactly why he was
19:22
doing what he was doing and why you were
19:23
chasing him was not super clear so from
19:27
the beginning of data fire pacing was a
19:28
priority the very beginning of the game
19:30
is your castles destroyed and this God
19:32
breaks his way out and you're almost
19:34
killed and then the god of death holds
19:36
your soul ransom and says please go find
19:38
athis or I'll just snap your soul into
19:40
nothing so very clear very clear
19:43
motivation for why you're doing what
19:44
you're doing
19:46
áthis this big big dude with the glowing
19:48
forehead he is the main antagonist he's
19:50
obviously the main antagonist right away
19:52
and we did iterate on the pacing a lot
19:55
so especially at the beginning of the
19:56
game there were a lot of times where we
19:58
would go and we would cut out small
20:00
loops we would cut out notes of dialogue
20:02
just to make sure that the introduction
20:04
to the game was a lot faster and
20:06
smoother than it was four pillars one
20:09
the factions were underdeveloped in
20:12
pillars one for those of you who played
20:15
pillars one you meet the faction
20:17
starting at act two and defiance bay and
20:19
when you leave defiance bay you will
20:21
never see those factions again and if
20:24
you play something like Fallout New
20:25
Vegas the factions are present almost
20:27
from the beginning of the game all the
20:28
way to the end so that was a misstep
20:30
dead fires factions are there pretty
20:33
much throughout the whole game
20:34
as soon as you get to ma che island do
20:37
you meet the wana you eat the prenup or
20:39
not the Prince PU meet the villian
20:41
trading company shortly thereafter you
20:43
meet the principie and then the royal
20:45
dead fire company so very early on you
20:47
meet all four of the factions and
20:49
they're there with you almost literally
20:51
up to the very end of the game so we did
20:54
a much better job of including the
20:55
factions they're also central the
20:57
political conflict of the dead fire but
21:00
that did confuse the plot this is my
21:02
fault I have to take full responsibility
21:04
for this so we had this very strong
21:06
start with athis going blank I'm in the
21:09
knees you know like going across the
21:11
ocean and the gods like you got to
21:13
follow that dude and then as you're
21:14
following him there are all these
21:16
factions that are fighting over the dead
21:17
fire and players logically we're like so
21:21
I'm following that guy it seems like
21:24
there's a lot of fighting going on here
21:25
am I supposed to be helping figure out
21:26
who controls the dead fire or how does
21:29
this tie together and I didn't do a very
21:31
good job of actually tying those things
21:32
together there were two cool plot points
21:34
that didn't mesh very well but overall
21:39
we did a lot more planning and plotting
21:40
about what the factions should do I
21:42
think that helped a great deal and every
21:45
faction had a companion champion this
21:46
was extremely important so all the
21:48
characters at the bottom are associated
21:50
with a faction and they have very strong
21:52
feelings about that factions role in the
21:55
world depending on the alliances you
21:57
make some of those characters might
21:58
leave you for example pala Jeana is
22:00
extremely sort of hard-ass about being
22:02
dedicated to the villian trading company
22:04
but there was actually something that a
22:06
long time ago Chris Avallone had made a
22:08
very strong point where it said like you
22:10
should always have a companion that is a
22:12
mouthpiece for one of the factions in
22:13
your game so we did a better job of that
22:15
in this game we did have a backer beta
22:18
we had a backer beta on pillars one as
22:21
well there are some tricky things about
22:23
backer betas it's very limited content
22:26
and it's a very limited audience I think
22:28
anyone who's done early access also
22:30
knows that there are dangerous to this
22:31
for those of you who want to do this
22:33
type of thing understand that your
22:36
audience is self selected to the most
22:39
hardcore enthusiastic people they give
22:42
very good feedback but it is also
22:43
feedback for psychotic gamers like they
22:47
have the most
22:48
strongest opinions they're like they're
22:50
gonna just grind your game into the dust
22:52
it is very valuable but it can kind of
22:55
skew your understanding of how your game
22:57
plays for a wider audience um also the
23:00
longer a beta goes on it becomes a well
23:02
trod path um that content that you put
23:05
out unless you're gonna put out a bunch
23:06
of new content every time you update the
23:08
backer beta people are gonna just run
23:10
that over over and over and over again
23:12
and it can lead to a very strange
23:14
perception of difficulty for example or
23:16
quality of writing because no writing
23:19
really seems that great if you've seen
23:21
it 20 times so there's a little bit of a
23:23
problem sometimes with content that gets
23:25
overused and over iterated on for us we
23:29
didn't have we didn't have any companion
23:32
or stories elements in the game we
23:34
wanted to keep that under wraps so we
23:35
didn't get feedback on that also our
23:38
this is our earliest content and that's
23:40
often the the worst content and
23:42
sometimes it's the hardest to iterate on
23:44
I know a lot of times I've worked on a
23:47
game and the very first stuff you do not
23:49
only is it sort of bad in execution but
23:52
when you get under the hood it's such a
23:53
mess that if you try to like pull out
23:56
and redo it it's a problem so sometimes
23:59
backers can get the sense that you
24:01
aren't really doing very much with the
24:03
content when in fact it's just hard to
24:05
do anything with it
24:06
it also personally warped my sense of
24:09
difficulty because I actually had this
24:10
opposite effect where because everyone
24:13
who's playing the backer beta was really
24:15
good and aggressive and they played it
24:18
so many times and they said all this is
24:19
so easy I was like you're just saying
24:22
that because you're a super crazy gamer
24:23
and you've already played this 20 times
24:25
I'm sure the difficulty is fine wrong it
24:28
was not fine it was actually really easy
24:30
which we found out when the game came
24:32
out so what we heard in the back row
24:35
beta a lot of the new mechanics were
24:37
confusing or punishing we did get great
24:38
feedback on that we made a lot of good
24:40
changes based on player feedback one of
24:43
the big things is that everyone just
24:44
moved way too fast this was also a
24:46
criticism and pillars one that we didn't
24:48
I didn't really heed as well as I should
24:49
have we did slow characters down a lot
24:52
just physical movement speed and the
24:55
shifter ship combat sucked we're gonna
24:57
get back to that but people were like
24:58
this sucks and we're like how about this
25:01
iteration that
25:02
sucks - how about at the game launch it
25:05
still sucks and there are a lot of
25:07
technical issues so it was worth doing
25:09
they did really find a lot of stuff both
25:12
from a gameplay perspective and also a
25:13
technical perspective that helped the
25:15
final game in the long run so the game
25:20
came out yay
25:22
well I'm not going very fast time I'm oh
25:25
well so what are we here the game is way
25:27
too easy there were lots of performance
25:30
problems the Fulvio was awesome everyone
25:34
loved it the main plot was too short
25:36
which is sort of true if you just crew
25:39
straight through it can be a very short
25:40
experience the game itself is very big
25:42
there are ambiguous stakes in the main
25:45
plot again there is this confusion
25:46
between the main plot and the sort of
25:49
the factional plots the companion
25:51
relationship mechanics were crazy and
25:53
didn't make any sense at all and a lot
25:56
of people we had these minor characters
25:58
called sidekicks they weren't full
25:59
companions a lot of people they always
26:02
want more companions so we had added
26:04
these mini companions we said we can't
26:05
write full like we can't write 10 full
26:08
companions so here's a couple of minor
26:11
companions called sidekicks and people
26:12
are like we want them be full companions
26:14
why did you include them at all
26:15
which you know is questionable I still
26:18
think that the game is better for having
26:19
them but people did want them to be
26:21
fully fleshed out Hey
26:23
another the ship ship combat is bad the
26:25
reviewer said it was bad the players
26:27
said it was bad so in the short term
26:30
after launching there were a couple
26:32
things that we couldn't really do
26:33
anything about anything that had to do
26:35
with the main plot and the villain you
26:37
know that's that's my fault
26:39
that's how I wrote the main plot of the
26:41
story there's not a whole lot you can do
26:43
in the short term to fix you know it's
26:45
just sort of fundamental structural
26:46
issues with the game and the story
26:48
ship-to-ship combat we did make changes
26:51
to the UI we made balance changes that
26:53
helped again though fundamentally people
26:56
who didn't like the system just we're
26:57
never gonna like it the companion
27:01
relationships
27:01
seemed all over the place the tuning was
27:03
really crazy so we did we did some
27:05
tuning we did some new tagging of
27:07
relationships I'll go into that the most
27:10
important thing I think we did is that
27:12
we displayed a record of changes between
27:14
your companions relationships
27:16
because a lot of times people would get
27:17
to a point where they'd say like why do
27:18
these characters hate each other I have
27:19
no idea why
27:20
so the companion relationship page would
27:22
actually show all of the steps that had
27:24
led to that point there wasn't any
27:27
silver bullet for performance
27:28
optimizations we did what we could but
27:30
we found that in a lot of cases our AI
27:32
especially in big areas like NACA taça
27:35
the main city was just really atrocious
27:38
there wasn't an easy way to split those
27:39
levels up we didn't want to pull a lot
27:41
of NPCs out and potentially screw things
27:43
up so that was difficult to deal with
27:45
and then difficulty there were two
27:47
things that we did if we found what I
27:48
call the tall blades of grass we would
27:50
cut those so people hate this people you
27:54
everyone knows that people hate this so
27:56
do it as early as possible if there's
27:57
one character build that is just
27:59
ludicrously powerful and like nine out
28:02
of ten people use it because it's just
28:03
such a joke while at the same time
28:05
saying I hate how easy this game is nerf
28:08
that build like very quickly just change
28:11
it cut it out as quickly as you can
28:14
the other stuff that we did is we
28:15
started doing all the encounter tuning
28:17
that we should have done earlier so now
28:21
we're going to talk about controversial
28:22
features the five character party again
28:25
we had feedback that combat was hard to
28:27
follow larger parties usually mean that
28:30
there are more enemies in the battle
28:32
makes sense
28:34
pillars characters are more active than
28:36
characters in Baldur's Gate and I
28:38
Swindell on average characters like
28:40
fighters and rogues and Rangers in the
28:42
old infinity engine games didn't have a
28:43
whole lot to do in pillars potentially
28:45
they have a lot more to do that means
28:47
that you're spending a lot more time
28:48
managing every character we found that
28:51
our UIs were cramped by the in the HUD
28:55
area and in the inventory they're really
28:57
cramped by having six characters also we
29:02
followed the same sort of approach that
29:03
Baldur's Gate and torment did where you
29:07
start with the main character and you
29:08
add more characters over time one of the
29:11
struggles with a system like that is
29:12
that you have to work very hard to keep
29:14
the players attention with regard to
29:15
pacing you're starting with one
29:17
character doing a very small number of
29:19
things then you have to then you have
29:22
three than you have for then you have
29:23
five as you're getting more characters
29:25
they're advancing in levels and they're
29:27
getting more things to do per character
29:28
so does this really
29:29
crazy explosion of potential maintenance
29:31
that you have to do so this is a big
29:33
challenge in pillars of eternity you
29:36
would usually get a dare as your first
29:38
companion followed shortly by a LOF or
29:40
vice versa
29:41
the last companion most people got was
29:43
sigani and because she was a ranger she
29:45
actually added two characters so your
29:47
party was 7 which was just a lot to
29:49
manage and a lot of times we heard
29:52
people give feedback they said I always
29:54
forgot about one character there was
29:55
always one character in a battle that I
29:57
just like I just forgot what they were
29:58
doing they were just Auto attacking over
30:00
in a corner you know like shooting their
30:02
wand or whatever and that didn't seem
30:04
great so implementing this was actually
30:07
really easy we were already planning on
30:09
rescanning the HUD and UI anyway
30:11
lowering the party limit was not hard
30:13
the battle sizes came down we had fewer
30:16
enemies per fight because there were
30:18
fewer characters to fight against and
30:20
with all the other changes that we had
30:22
made based on previous feedback the
30:24
combat had become a lot easier to follow
30:26
the reception to this a lot of people
30:30
grumbled about it grumbling though
30:33
here's the thing it's ok like if
30:36
someone's like oh that's dumb I don't
30:39
like it but they're not like really
30:40
angry if it actually is making a good
30:43
change in the game just just do that
30:45
just keep going there were some people
30:47
that really really really were mad like
30:49
I'm not playing this game it has fewer
30:52
than 6 characters which is you know a
30:54
weird thing in this world to be that mad
30:56
about but there were some people that
30:59
really didn't like it and that's just
31:00
how they how they felt um
31:02
the logic was like well baldur's gate
31:03
nice when Dale had 6 character parties
31:05
Ergo and pillars won at a 6 character
31:08
party ergo kind of ignoring the fact
31:10
that we had received all his feedback
31:11
like wow this is a lot to manage overall
31:16
people said that the combat was easier
31:18
to follow so what I do it again yeah
31:21
probably I mean a lot of it depends on
31:23
the way your combat is structured if we
31:26
didn't have real time with pause combat
31:28
maybe six characters are fine maybe
31:30
maybe 10 characters are fine go play
31:32
some battle brothers get 12 characters
31:34
in there you know like who cares
31:35
you could have huge battles if you
31:37
really want to but a lot of it depends
31:38
on the context but for this game I would
31:40
totally make the choice again
31:42
the companion relationship mechanics I'm
31:45
gonna dive a little bit into the weeds
31:46
here just because there's a murky system
31:49
here that was very confusing and
31:50
frustrating for people what are you
31:52
talking about the companion relationship
31:54
mechanics so the way that companions
31:59
would change their relationship to you
32:01
and to each other was largely through
32:02
the system called topics so you see
32:05
these little icons around palutena here
32:07
they represent things like being
32:08
prevailing Republic or pro-ana man see
32:11
or anti-religious these are things that
32:13
she likes she likes people that are
32:16
anti-religious she likes people that say
32:18
Anna man she is cool every time a
32:20
character says something that's tagged
32:21
with that topic she goes yeah and
32:24
they're sort of like in a procedural and
32:26
not on a moat but basically a reactive
32:28
line or pal Geno's like yeah and and
32:32
then her reputation goes up with that
32:33
character and then over time that builds
32:37
toward a break point either positive or
32:39
negative and then once you hit that
32:41
break point you know there's an
32:42
emotional sort of explosion the
32:44
character is like I think you're awesome
32:46
thanks for being so great and hating
32:47
religion or they say like you know like
32:50
I love you you're the best and then you
32:53
as the player can either interact with
32:54
that or just say mm-hmm
32:56
sorry I don't want to deal with it which
32:59
actually a lot of people did there were
33:01
certain conflicts where players are just
33:02
kind of like back away and let things
33:04
take their course
33:05
so what problem was I trying to solve I
33:08
believe that there is a space between
33:11
doing purely hand scripted reactivity
33:14
for characters and things that feel more
33:16
topical and I don't want to say
33:18
procedural but a little more
33:19
mechanically driven if you look at
33:21
something like fallout 4
33:22
you have companions that will react to
33:24
things like picking locks or stealing
33:25
which are sort of devoid of context and
33:28
that's mechanically driven on the other
33:30
hand if you have a thing where
33:32
characters only respond to very specific
33:34
lines that every character says that's a
33:37
lot of work and it feels very special
33:38
Casey like you have to take certain
33:40
routes through the game to hit a certain
33:42
breakpoint with a character so what this
33:45
was trying to do is say hey there are
33:46
all of these topics that people talk
33:48
about throughout the world let's use the
33:50
fact that the player makes hundreds of
33:52
dialogue choices that sort of intersect
33:54
these topics
33:55
and companions make tons of offhand
33:57
comments and use that to do mechanical
34:00
work behind the scenes so the player
34:02
will feel that it's a more organic way
34:04
of building a relationship so that when
34:06
two characters decide that they're
34:08
buddies or they hate each other's guts
34:09
the players like make sense because
34:11
every time this character keeps saying
34:13
something that's anti-religious this
34:14
other character keeps going mmm until
34:18
they just explode and go crazy and the
34:20
players like yeah I saw that happening
34:21
makes sense so the implementation was
34:24
actually really easy the mechanics of
34:25
this were very straightforward um but
34:29
the designers had a difficult time
34:31
really understanding it part of that is
34:33
because I designed a system that was
34:34
probably one layer of complexity beyond
34:37
what it needed to be you had a lot of a
34:40
lot of adjustability and fine-tuning
34:42
that ultimately proved to be very
34:44
difficult to wrap your head around it
34:46
tuning was very difficult for a number
34:49
of reasons so again the goal of this is
34:51
to produce something that feels natural
34:52
and organic so that's a feeling and
34:55
trying to get that result with a
34:57
mechanical system can be very tricky but
34:59
how nodes are tagged as arbitrary for
35:03
example anti-religious what counts as
35:06
anti-religious that's a subjective thing
35:09
racist what counts as racist that's also
35:12
a subjective thing we also had strengths
35:15
to that this is a minor racist comment
35:17
this is a major racist comment to
35:19
different designers can interpret those
35:21
very differently it's not resulted in a
35:22
lot of ambiguity between those things
35:25
the values on individual topics were
35:27
largely arbitrary so you know you'd say
35:30
like ah well this topic comes up about
35:33
15 times in the game so it's 8 points
35:35
per thing this topic comes up five times
35:37
so it's 20 points per instance that was
35:40
done for flexibility but ultimately it
35:42
was just very difficult for people to
35:44
wrap their heads around an organic
35:46
system is hard to test that's part of
35:48
the point is that if you take a
35:49
different route to the game you can
35:50
arrive at that relationship through a
35:52
different way and what feels good is
35:55
ultimately subjective so someone might
35:57
say yeah this feels totally great and
35:59
another person might say this feels
36:00
completely rushed or fast or why is it
36:03
taking so long for this character to
36:05
react to the choices that I've made so
36:07
the reception
36:08
not good this did not go well there were
36:12
lots of behaviors that were just way out
36:14
of whack um
36:15
Alif was one of them al Hoff came across
36:17
as this supercritical like he would just
36:20
he was constantly frowning and like
36:22
being mad about almost everything you
36:24
did and it was because his topics were
36:26
marked very correctly but the instances
36:29
of that were just so frequent that you
36:30
were just constantly seeing a lot like
36:32
threat about everything you did which
36:34
was not really the intention and then
36:36
the other one that came up is that show
36:38
D and palutena had these like terrible
36:40
conflicts um show D is a priestess pala
36:43
Geena hates the gods and all religion
36:45
and there were a couple of places where
36:47
you could like have showed in your party
36:49
get pala Geena and then go into her hub
36:51
and be like tell me about yourself
36:53
palette you know and she's like well I
36:54
hate the gods also I think all priests
36:56
are stupid morons they're complete idiot
36:58
it's unlike Bing Bing Bing Bing Bing and
37:00
so over and over again show Dee is just
37:02
like going crazy and then after five
37:05
minutes of going through palo jeana's
37:06
hub you could exit and showed he would
37:08
immediately explode on this person that
37:10
she met five minutes ago that was not
37:12
intended but that's how it worked so the
37:16
goal is to be more organic and natural
37:17
but these failures felt especially
37:19
artificial they felt like really really
37:22
bad so it was the complete opposite of
37:23
what I wanted would I do it again only
37:27
if I simplified the system I do think
37:29
that there is a cool thing that was
37:30
going on there but it was just too
37:32
complicated for people to use it's
37:34
entirely my fault I did want to try a
37:36
new system I think we learned important
37:38
things from it and it could be valuable
37:40
in the future if it were tuned much
37:42
better ultimately I feel like sometimes
37:46
you have to go too far to know where the
37:47
limit is I found it now we're going to
37:51
talk about ship-to-ship combats this is
37:53
very controversial this is the UI for
37:56
ship-to-ship combat so instead of having
37:58
like a thing on the open seas where
38:01
ships would fight when two ships
38:03
encountered each other you would do this
38:05
turn-based minigame that was abstracted
38:07
you had your ship position relative to
38:09
the other ship at the bottom
38:10
you had turns that went in sequence this
38:12
was very abstract this is very bored
38:14
gamey also very disliked by a lot of
38:19
people I thought it was fun but I could
38:21
tell that a lot of people are not
38:22
like it as soon as the ship became the
38:26
stronghold I remember very clearly I
38:29
said hey guys let's not turn this into
38:32
the pirate game and that is exactly what
38:34
it became so it's a it's a ship minigame
38:37
where you maneuver around and blast each
38:39
other it seems very logical I have a
38:42
ship there are other ships sailing
38:44
around I should be able to shoot cannons
38:45
at them okay I guess that makes sense
38:48
it's a cool idea maybe for some people
38:51
if you like this style of presentation a
38:53
lot of people really did not so how do
38:57
we implement this it really is a full
39:00
game within a game like you could have
39:01
we could have had like a whole indie dev
39:04
studio just make this game it seems very
39:06
simple but there is like so much stuff
39:08
going on in it it's a ton of stuff it
39:11
really could just be a game on its own
39:12
not a very fun game for a lot of people
39:15
but a game it took us a huge amount of
39:17
resources to get to alpha feedback was
39:20
very negative on the team they were like
39:21
we don't like this there are a couple
39:23
people that did I actually thought it
39:24
was had promised even more resources
39:29
went into the first iteration on it and
39:30
the feedback was still extremely
39:32
negative so at that point I said goodbye
39:37
this feature has to go away it's it's
39:40
quicksand it's if you've been in
39:42
development long enough and you see like
39:44
people just keep marching into this pit
39:46
and disappearing and then you just keep
39:48
sending more people into the pit and
39:49
nothing's getting better that's
39:51
quicksand
39:52
it's a quicksand feature and so I cut it
39:54
and my boss brought it back as a
39:57
crowdfunding goal this is where throwing
40:01
people under the bus happens
40:03
I said hey man I think this is going to
40:05
be really expensive and really difficult
40:07
and I don't think a lot of people are
40:09
gonna like it
40:10
but he brought it back as a crowdfunding
40:11
goal and so we were gonna do it it was
40:15
the most expensive feature in pillars of
40:18
eternity two dead fire in terms of dev
40:21
time and money spent it was a drain on
40:23
every department especially testing but
40:25
every other department had to contribute
40:27
to it it never really seemed to satisfy
40:30
players I think this was the thing is
40:32
that there were players that really
40:33
liked it I liked the system I thought it
40:36
was pretty fun but there was a big
40:37
portion of our players who were like I
40:39
hate the whole system they hate the
40:41
presentation they just they just hated
40:43
the whole thing quicksand
40:46
yeah it just it always seemed to need
40:48
more stuff so the reception to this was
40:52
technically mixed because again there
40:54
were people that really did like it
40:56
most people hated it lots of reviewers
41:00
it was the most criticized feature in
41:01
the game it was a very abstract system
41:04
with limited mechanics would I do it
41:06
again no way like again it's not that I
41:10
don't think that I don't it's not that I
41:12
personally don't think it's fun is just
41:13
that for something that most players
41:16
will have to engage with knowing that
41:18
many of the people playing the game were
41:19
never going to like it I would not do it
41:22
again because ultimately this is the
41:25
fantasy we needed to deliver on which
41:28
was deck to deck combat we already had
41:30
that remember this this is what we
41:33
showed players you'd be doing you would
41:34
be shooting people in the face at
41:36
point-blank range like I did really want
41:39
to get tentacles attacking your ship I
41:40
never got that so sorry about that but
41:42
ultimately fighting on the deck of the
41:44
ship this that's what we essentially
41:47
showed the player that we were going to
41:48
do and that's what we needed to do the
41:50
ship to ship stuff I think just wound up
41:52
being a lot of time full voiceover
41:55
alright I'm rushing through these guys
41:57
so as I said before divinity original
41:59
sin to set a new standard we had a lot
42:02
of internal debate about Fulvio I
42:03
thought it was too late because it was
42:07
halfway through development and we did
42:08
not start off thinking that we were
42:10
going to be able to that we were going
42:12
to do Fulvio we were going to do a lot
42:14
more vo
42:15
but not fooled you I was the only person
42:18
really on that team that had previous
42:19
experience working on a game of Fulvio
42:21
which was Fallout New Vegas and that was
42:23
65,000 lines of dialogue and it was a
42:26
ton of work so what I said is let's do
42:29
full view in the early game and not do
42:31
side characters in the later game let's
42:33
like you know sort of hedge our bets and
42:35
just and manage that I was overruled
42:37
that the owner levels owners believe
42:39
that we really really really needed
42:40
Fulvio so we did it
42:43
Fulvio implementation critical role came
42:46
out to do the major character voices
42:47
which was very very cool when we had to
42:51
do prep that was with to production
42:53
milestones left so people are going hey
42:56
why aren't the writers writing anymore
42:58
I'm like because they're recording vo
43:01
and that takes all of their time for a
43:04
game that has again tens of thousands of
43:06
lines of dialogue we were recording
43:08
across three time zones including two am
43:11
recording sessions which were very very
43:13
bad if people missed sessions either due
43:16
to being very sleepy or overlapping we
43:20
would get casting and recording errors
43:21
and then we would need pickups or we'd
43:22
have to just live with bad vo initially
43:26
the owners would not move the schedule I
43:28
will go on record here as saying this
43:30
was the most intense personal pressure
43:32
I've ever had on a project and that
43:33
includes shipping Fallout New Vegas in
43:35
18 months so this was extremely
43:38
extremely fatiguing and draining took me
43:43
to the breaking point this was very very
43:44
difficult to get through
43:45
eventually the schedule was pushed by
43:47
two and a half months we needed that
43:49
there was just no way we were gonna get
43:50
the game done without that extra time so
43:52
we did get that extra time and we got
43:54
through it how did it go over
43:58
I went over very well people loved it
43:59
streamers reviewers players loved it
44:02
critical role added a huge amount is
44:06
this a hard limit am I gonna die if this
44:09
timer runs out I guess I'm gonna find
44:11
out would I do it again absolutely not
44:14
under these same circumstances even
44:16
knowing how much Fulvio added to the
44:18
game knowing that we were not going to
44:20
get the extra time initially I would
44:22
never do that
44:23
it was extremely demoralizing for the
44:25
narrative team and it put everyone under
44:28
a huge amount of
44:28
stress I would I would still advocate
44:30
the same partial VL solution with proper
44:32
time yes because now it's an expectation
44:35
which kind of sucks but that's what it
44:37
is
44:37
oh my god many times up
44:41
post-launch the dlcs seeker Slayer
44:44
survivor beast of winter and forgotten
44:46
sanctum I think the DLC teams did a
44:48
really fantastic job these games are
44:50
much more focused than the original game
44:52
awesome teamwork they did awesome job I
44:55
just want to call out all the work that
44:57
they did is being very very very good
44:58
there was internal debate about doing a
45:00
big expansion versus three DLCs owners
45:03
wanted to do three DLC so that's what we
45:05
did or rather what the expansion team
45:07
did it was better than doing the split
45:10
x-mansion
45:11
but it was still very stressful doing
45:14
three DLCs on a tight schedule is very
45:15
hard lots of overhead for every single
45:18
launch because if you do three DLCs in a
45:20
row you're essentially launching three
45:22
products and then you're supporting each
45:23
one of them as you go on alright in the
45:26
long term difficulty we made lots of
45:30
balance changes over months and months
45:31
and months of time we also tuned almost
45:33
every crit path in major side area and
45:35
the game encounters that's work that
45:36
should have been done early on we had a
45:39
god challenges these are optional
45:40
gameplay modes that made the game more
45:42
difficult for example ma grens challenge
45:43
made it's that you had limited time to
45:46
pause the game and in turn based mode
45:48
which we added later your turns would
45:50
actually run out and just skip Macbeth
45:52
God challenge any character knocked out
45:54
for more than 10 seconds would die so
45:56
these are pretty significant gameplay
45:58
challenges we added 11 overall and then
46:00
an ultimate that wrapped it up that was
46:01
very cool we continued iterating on the
46:04
ship UI we did actually include
46:07
narrative changes to explain some of the
46:08
narrative shortcomings these were
46:11
included in the 5.0 patch I think people
46:13
have received them very well again it's
46:15
my fault that it wasn't clear to begin
46:16
with but we did try to address it in the
46:18
long run and then finally those
46:20
sidekicks those miner companions they
46:22
did each get expanded into sort of a
46:24
mini companion with the DLCs again I
46:26
have to thank the DLC team for doing
46:28
great work here
46:29
Edwin Edwin Constantine and Ficino each
46:32
got expanded into a mini companion with
46:34
their expansions or with their DLCs the
46:39
biggest thing that was added post-launch
46:41
that made the biggest difference over
46:42
was turn-based combat this was the thing
46:44
that we didn't do from the beginning
46:45
because the Infinity engine games are
46:48
real-time with paws but about a billion
46:50
people bought divinity original sin too
46:52
so let's give it a whirl um this was not
46:55
my idea this was the idea of Nick Carver
46:57
working with a programmer named Brian
46:59
McIntosh I initially resisted it because
47:01
it did not accurately preserve the
47:04
action economy of the real time with
47:06
pause combat system the real time with
47:07
pause combat system was based on seconds
47:10
and tenths of seconds and the turn-based
47:12
system that they could implement
47:13
reasonably couldn't sustain that it had
47:16
to kind of break the action economy but
47:18
I realized that it was an argument of
47:20
perfect versus good yes we could not
47:23
faithfully adapt the original system but
47:25
tons of people who were turned off by
47:27
the real-time of pause combat would be
47:29
drawn into the game even with an
47:30
imperfect implementation so we did it by
47:33
far the most significant implementation
47:35
are a thing that we added to the game
47:36
this was don't get me wrong this was a
47:40
lot of work but compared to things like
47:41
the ship combat system this was like
47:43
nothing like this was a very sort of
47:45
straightforward implementation there
47:46
were lots of bugs to work out but the
47:48
reception - it was almost universally
47:50
positive it was a great thing for us to
47:51
add and we added it you know nine months
47:54
or so after the game launched and it
47:57
people responded very well to it all
47:59
right I actually made it through so some
48:02
lessons learned here we made some
48:04
assumptions about real time with pause
48:07
and turn-based combat that maybe don't
48:09
apply anymore
48:09
it's not the late 90s or early early
48:12
2000s I know somebody's gonna read this
48:16
and say oh my god obsidian is never
48:17
gonna make a real time with pause game
48:18
again I'm not seeing that but what I'm
48:20
saying is that I've always personally
48:22
really liked turn-based games I think
48:24
we're now finding that there are
48:25
audiences that also really appreciate
48:26
them I think the fact that our
48:28
turn-based motorists issue so well means
48:30
that we don't we don't have to just
48:32
constrain the piller series to doing
48:35
real-time with Paul's combat in the
48:37
future don't make any assumptions about
48:40
that but it's just to say that
48:41
turn-based
48:42
is a totally viable thing that we can do
48:44
now easier difficulties are very
48:47
important but hardcore RPG players they
48:49
really do need a challenge when the game
48:50
launches keep in mind the first people
48:53
to buy the game are the ones who are
48:54
most excited about it
48:55
often they're the most passionate
48:56
they've played your back or beta they
48:58
played early access all that stuff so if
49:00
the game is not challenging its gonna be
49:01
really really disappointing
49:03
I'm not saying to make the game
49:04
incredibly difficult for everybody but
49:06
if you neglect the difficulty too much
49:08
as I did it can bite you in the ass and
49:11
that's not good when you see a quicksand
49:14
feature you got you'll know when you see
49:17
them you just you just see resource
49:18
after resource going to them try to cut
49:20
them as soon as possible fight like hell
49:22
to get rid of them because they're
49:23
quicksand they will just consume
49:25
resources until the game launches and
49:27
ultimately if they don't really work
49:28
well if your design it's not gonna be a
49:30
good fit and unfortunately I would say
49:34
now the expectation of Fulvio is now
49:36
very real in these games it's not just
49:38
because of divinity it's also because of
49:40
us we contributed to it I actually saw
49:42
reviews of owl cats
49:45
Pathfinder kingmaker where all these
49:47
people were going like why doesn't set
49:48
up full Bo this is such BS dead fire and
49:50
divinity original sin to have Fulvio all
49:53
cat was working on a much smaller budget
49:54
and it's it's it's kind of a ridiculous
49:57
expectation but now it has become this
49:58
expectation where players just think
50:00
that everything will have full view it
50:01
sucks
50:02
umm I don't really know what to do about
50:04
other than to plan plan plan
50:06
Fulvio and a big game is extremely
50:08
expensive and it takes so much time you
50:12
have to lock your dialogue so much
50:14
farther in advance than you're expecting
50:16
to and I mean again I think it was
50:20
62,000 lines of dialogue for dead fire
50:23
like it's just it's a logistical
50:25
nightmare so if you if you go down that
50:27
road just understand that you're gonna
50:28
have to plan a lot for it and ultimately
50:31
again the plot should have been about
50:32
one struggle not two
50:34
we had two very cool neat ideas that
50:36
didn't really quite converge entirely my
50:38
fault
50:39
but lesson learned also I should
50:41
probably direct a different type of game
50:43
next I mean I've worked on I Swindell
50:47
Icewind Dale two Neverwinter Nights 2
50:49
and then pillars of eternity one into
50:51
these are all party based fantasy real
50:54
time with pause games I'm a little
50:57
burned out on them dead fire was again
50:59
the most stressful sort of directing
51:00
experience I've had so far so hopefully
51:02
the next thing I direct will be a little
51:04
bit different and then maybe I I still
51:06
like these games a lot but I need some
51:07
time
51:08
perspective to come back to them you can
51:10
burn out you can burn out on directing
51:11
the same type of game over and over
51:12
again and then you're not really doing
51:14
good work because you're just burned out
51:16
thank you very much thank you very much
51:26
is there a question time bombs one or
51:29
two if it's a right okay hello fact
51:36
thanks for the great lecture thank you
51:39
I was wondering because a year ago I was
51:42
also working on RPG game and was big
51:46
game and I was wondering how many times
51:49
have you actually played the game from
51:52
beginning to the end like in one one
51:55
size file because I actually never did
51:58
it so I think it's important personally
52:01
because I'm director and often the lead
52:02
designer I always try to play through
52:05
the game at least once from beginning to
52:08
end every game that I've worked on in a
52:10
significant capacity I've always worked
52:11
on I've always played through the whole
52:13
game with one character the whole way
52:14
through I think it's very important to
52:16
get that experience on dead fire I had
52:19
one full playthrough and three half
52:23
playthroughs I think total so one of the
52:28
problems is that dead fire is an
52:29
enormous game there's also some debate I
52:31
think which is a separate topic should
52:34
directors design content or should they
52:37
purely just play the game and get
52:38
feedback I I want to work on the game
52:42
like I also want to make things in the
52:45
game so like in pillars one I did almost
52:47
all the system design on pillars do I
52:49
did some system design that was the co
52:51
narrative lead with Carrie Patel but I
52:53
also then played the game and gave
52:55
feedback it's a difficult balancing act
52:57
but I think if a person has
53:00
responsibilities on the game it's very
53:02
difficult unless the team leadership
53:05
makes time to play the game so one thing
53:08
that we did do a little bit better on
53:11
pillars then on debt fire was we have
53:13
team play days and they're enormous ly
53:15
valuable for getting people's eyes on
53:17
the game because sometimes you're so
53:18
deep in implementation you just have no
53:20
idea what's really going on
53:22
so but the thing is people camp I mean
53:24
they can make time on their own but
53:26
that's a lot to ask of people usually
53:28
what we try to do is schedule time for
53:31
people to play through the game so they
53:32
get more exposure to it it altum Utley
53:33
makes the game better in the long run
53:37
okay a thing we have to and here so in
53:41
order to prepare for the next speaker
53:42
right all right thank you thank you very
53:44
much
53:45
[Applause]
53:46