YT YouTube2Text

YouTube2Text → Transcripts → Konsoll 2018: Oskar Stålberg - Developing The Bad North Look

Konsoll 2018: Oskar Stålberg - Developing The Bad North Look — Transcript

by Konsoll · 10,311 words · 1,396 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00we are ready for the last talk of today
0:04and our next speaker has experience from
0:07companies like Ubisoft and us too he is
0:11the technical artist turned indie to
0:13create a very interesting game that
0:15North please welcome Oscar strolling
0:26yeah yeah as you can see I'm Oscar Saul
0:31Berg I'll just start by introducing
0:33myself I used to work for Ubisoft
0:35massive where I among as a technical
0:38artist where I did a bunch of shader
0:40work and among other things made the
0:42cool like holographic 3d map in the
0:45division I'm like one of my big
0:49interests is procedural generation and
0:51this talk is gonna be a lot about
0:53procedural general oh that's very
0:55interesting
1:01that is very there we go okay let's keep
1:04tabs on that yeah this talk is gonna be
1:08a lot about procedural generation
1:09because bad north is all about
1:11procedural generation basically all
1:12aspects of the game are procedurally
1:14generated in some kind of way here are
1:17two like earlier examples of procedural
1:19procedurally generated demos I've done
1:21one of them you get to build a little
1:23red brick house and one of them you get
1:25to terraform a sort of medieval era
1:28planet both of these you can find online
1:32and play in your browser if you just
1:33google my name and brick block for that
1:36one and just plan it for that one
1:38you'll find more just google my name
1:39you'll find them and that sort of so
1:43those I did on my spare time so that got
1:45me started round like thinking about
1:46procedural generation and more
1:47specifically thinking about kind of
1:50content driven
1:51tile based procedural generation in 3d
1:54worlds which sent the which later would
1:57turn into stuff that would become bad
2:00north so that's enough about me so what
2:04I'm gonna do in this talk is I'm going
2:06to assume gonna open up unity and take a
2:09poke around
2:10look at like what bad North looks like
2:12in the scene view in unity and like what
2:15happens when we tweak a bunch of values
2:16and stuff like that but we'll take with
2:19us some like core design principles or
2:22sort of aesthetic principles that I
2:25think permeates the design of bad north
2:28and these are mostly like these aren't
2:29things I came up with before it well
2:31some of them are but most of them aren't
2:32things I came up with before I started
2:33working on the game but things I kind of
2:35discovered and found work well as I was
2:38making the game so we'll just go through
2:40them quickly one of them that's pretty
2:42simple one
2:42borders not textures or like smooth
2:46gradients crisp lines so it's like
2:49instead of using like a tiling grass
2:51texture for example telling you over and
2:53over this is grass this is also grass
2:55this is also grass this is also grass
2:57you do most of your work defining areas
3:01like around the edges of the areas
3:03instead of like in the middle of the
3:04areas because that's that's really
3:06interesting is how things happen where
3:08different areas meet and that's where
3:10you want to put it like you can still
3:11have a texture but like tone it down and
3:13put some focus on the border since that
3:14the other one quite obvious symmetry
3:17like the idol game has an island in the
3:20middle that the camera rotates around so
3:22very natural we get the symmetric
3:23feeling for it and that's also a thing
3:25we try to play within like the
3:27iconography on the game and a lot of the
3:29like the art of the game and stuff like
3:31that it's kind of like a this kind of
3:34shining city on a hill feel to it
3:37the other one more sort of dramatic one
3:40order and chaos which sounds very
3:42dramatic and in a lot of ways we also
3:44have that very dramatically represented
3:46in bad north but if we sort of pull it
3:48down a notch we get something more like
3:50establishing patterns and breaking
3:52patterns or sort of make rules break
3:54rules which i think is a quite
3:56fundamental design principle and like
3:58all kinds of thing in game design in in
4:01art in music as well right like do
4:04things in a certain pattern and like
4:05twist it up a little bit and then when
4:07you get used to that you twist it up a
4:08little bit and there are a lot of that I
4:09think in the esthetical aspect of bad
4:12north to both on quite low levels and on
4:14slightly higher levels the next thing
4:17respect abstraction that sounds quite
4:19abstract it is will get some very very
4:21concrete examples of that though and
4:23it sort of ties into the idea of not
4:26seeing the forest for all the trees and
4:28striving to do the opposite of that and
4:30of course the next one is quite obvious
4:32minimal labour cost especially minimal
4:35repetitive labor because we are a very
4:38small team is me and one other guy
4:40full-time and then audio guy half time
4:42so like we don't have a lot of time to
4:44put into laborious workflow so the whole
4:48game is designed around everything being
4:50very fast to produce not necessarily
4:53like because sometimes obviously it can
4:55take longer time to build the system
4:57then build like do everything manually
4:59and so we don't always save time from
5:01that but it's also that very very
5:02repetitive tasks tend to get quite
5:04boring and they're also harder to tweak
5:05in the end I think that if you build
5:07very systems based things where you can
5:09like tweet individual variables and
5:10everything changes yeah so those are
5:14sort of the basic things that hopefully
5:15will keep coming back to us we take a
5:17look inside the game so yeah this is bad
5:20north how many here have played bad
5:22north so a few people how many people
5:25have seen bad north more people okay
5:28well everyone who roasts that had the
5:30second type but not the first byte yeah
5:34so this like the campaign view it's
5:35changed a little bit from the version
5:37that's out there right now and from the
5:40campaign we can enter one of the many
5:42islands and of course this is
5:44procedurally generated so we'll start
5:46taking talking a little bit about what
5:48goes into making these islands as I said
5:51they're procedurally generated they use
5:53an algorithm called wavefunction
5:57collapse which I didn't name it it's
5:59quite a weird name but it comes from
6:01sort of an idea of like building a
6:03possibility space and then sort of
6:04collapsing it until you have one final
6:06solution so the way but the way I
6:11interface with it which is actually
6:12quite easy to understand is I just give
6:14it a bunch of like tiles that I build in
6:16Maya 3d tiles that looks like this and
6:18then the algorithm figures out how to
6:20assemble them into complete levels that
6:23also ties into the very I guess I said
6:26minimal work for minimal workflow thing
6:28where I can just throw in new tile
6:31assets generate new levels and I'll get
6:33to see what those look like
6:34so obviously right now I have like quite
6:37a few tile
6:37going on here I think there's 400
6:40something like that which obviously like
6:41took a long time to build but I did that
6:44over two years time but when I was just
6:47trying out the algorithm in the
6:49beginning and like trying to nail down
6:50the basic feel of the shape of the
6:53island like how big are my bevels what
6:56are like how frequent are my detail
6:58stuff like that basically everything I
7:01worked with was like these tiles over
7:03here something like that so I needed
7:05like 10 tiles and then the algorithm
7:06figures out how to build that into
7:08separate levels so a benefit you have
7:11from that is you don't have to like
7:13sometimes with procedural content you
7:15have to make a lot of different
7:16permutations of things before it starts
7:18working but with this algorithm you
7:20don't which makes it much quicker to
7:22work with and you don't have to make all
7:23the weird like diagonal fitting pieces
7:26to work with every single use case
7:28because the kind of the tiles themselves
7:30restricts what kind of levels it can
7:32build out of them again of course if you
7:36build a game out of tiles like this or
7:38any game out of tiles that again ties
7:41into the sort of order thing where like
7:43you establish very very clear pattern so
7:45you have a tile based game grid based
7:47game and that's something we've
7:48obviously largely leaned into it is a
7:51very gridded island it looks like a like
7:53it's a more blocky than an island would
7:55look like in real life
7:56and obviously the basic gameplay of the
7:59game as well like the grid squares you
8:02move your units to are on the same grid
8:05as the game is being generated on but
8:07then of course make patterns break
8:09patterns we do a bunch of little small
8:11things just to make it like to break up
8:14the super obvious blockiness of it one
8:17of the things we do is to implement it
8:20and this is like one of also one of my
8:21contributions I think to to this because
8:23I didn't invent this algorithm but one
8:25thing that I brought to the algorithm is
8:27adding these sort of tiles that span
8:29across multiple grid squares so here's a
8:32good example of that this little tile
8:35here like actually consists of our
8:39module we call them this little module
8:40actually spans across across the two
8:42tiles which makes that we can put like a
8:44little bit more of a crack into it that
8:46like makes it a little bit larger sort
8:48of rounds off the shape a little bit
8:49more
8:50removes that super obvious sort of grid
8:53size from the rest of the game so that's
8:57a quite subtle way of doing at that that
8:59also means that we can sneak in stuff
9:02like this like this little sort of
9:03tunnel here or yeah we have some like
9:10bridge just going about some caves and
9:13bridges going on and of course having
9:14things stretch across multiple tiles
9:16also means that we can do stairs which
9:19you like you need so the game doesn't
9:20all play on take place on one level also
9:25just like from making this tool as I
9:27said workflow is very very important and
9:28when you make procedural content
9:31especially if you're not the same person
9:33writing the algorithm and doing the
9:36producing the art for it you might end
9:38up with tools that require you to put in
9:40a lot of custom data like you make some
9:42tiles export them to the game and then
9:44you have to say import them into the
9:45engine and then like mark things up to
9:47tell them what's going on in different
9:49tiles but that's all been automated here
9:50so as you see if I highlight an edge
9:54here it will highlight in yellow all the
9:56other edges that look the same so that's
9:59just like the import algorithm like
10:01looking through the mesh looping through
10:03the edges and see what fits where yeah
10:07and if I like I like this side you'll
10:08see everything with a straight edge
10:09everything with a straight horizontal
10:12edge stuff like that
10:13another little fun thing that I'm it's
10:16quite tiny but I'm quite proud of it
10:17that big breaks up the repeted
10:19repetition a little bit is like if you
10:22know is this very basic corner here it's
10:25not entirely symmetrical like this is
10:28not a 45-degree edge it's like slightly
10:29slanted which means that if you put
10:31together like a sort of pillar it will
10:38end up having slightly different like it
10:40can be a bit pointy like this or it can
10:42be a little bit duller or like it can be
10:44a little bit a symmetrical stuff like
10:45that this is a very very basic thing to
10:47break up that they're the most common
10:49parts of the level without adding
10:51additional clutter to it because that's
10:53another important thing and in addition
10:56to that I have another very tiny thing
10:58which is like these little right because
11:01that you want to intimate some
11:02additional detail too so this is a
11:03little sort of
11:04dr. piece that has a little bit of a
11:06crack in it and it flips the asymmetry
11:11so you can have like a plane going one
11:13way and then this little crack shows up
11:15and then the play goes another way so
11:16it's a way to introduce like a little
11:18bit of extra detail almost kind of a
11:20little bit of a texture to the cliff and
11:23again the tech the sort of texture crack
11:25here appears on the edge not on the side
11:28of the cliff because the site isn't an
11:30interesting place what's interesting is
11:31what happens when the cliff bends but it
11:34also produces some meaning in that it
11:35flips the the side so it sort of creates
11:37a bigger pattern than than just the
11:39crack itself this is a very small thing
11:42another thing we do to break up the
11:47tiles that's actually not in the
11:49released version of the game right now
11:51but we'll see in an update fairly soon
11:53because I've just made this the other
11:54day and I'm quite happy to show you it's
11:56quite subtle again like a lot of things
11:58in this game
11:59is this little thing here oh yeah yeah
12:08so it's very very subtle but it's a way
12:11to just like smooth out the island a
12:13little bit so this is what the release
12:14version looks like now so all the edges
12:16are straight and of course like if you
12:18work with these kind of tile based
12:20things you're like that can easily
12:22happen because of course this piece here
12:24doesn't know that the that there's going
12:27to be a like an end to it here right so
12:31it has to be purely straight but then I
12:34add this little effect afterwards and
12:37this also like it's a way to smooth out
12:39the entire thing but it does very much
12:40tie-in to the fact that the game is like
12:43voxel-based and gridded from the
12:44beginning because what happens is if we
12:47just look at the island generation and
12:50then we'll find the right Island here's
12:52the island there's like a voxel space
12:57yeah we create a voxel space from the
13:01island some of it's quite obvious it
13:03happens from the pieces but also as you
13:05can see here the trees add a little bit
13:07of sort of voxel coverage to them this
13:09is used for lighting as well we'll get
13:11more into that later but if you have a
13:13voxel space space that means that you
13:15can also drip drive normals from that
13:16and those normals you can use
13:18for the lighting but you can also use
13:20them to push the edges a little bit of
13:23the entire thing to sort of smooth out
13:24the entire thing this is quite a nice
13:27little thing very small bit quite nice
13:33yeah and to this sort of abstractions
13:36part there's a like I've tried to be
13:38very very limited in what kind of things
13:40I put into the island right there's very
13:42very few colors going on in the base
13:44island there's basically three elements
13:45here there's walkable areas non walkable
13:48areas and cliffs and that's it
13:50right so that's because that's what you
13:52care about when you play the game you're
13:53like where can my troops go where's the
13:56like and where are the where are the
13:58different layers and where are the paths
13:59between the different layers so I don't
14:01need to have a rock texture telling you
14:03over and over again there's a rock here
14:04Rock here Rock here Rock here what
14:06you're interested in it's like breaking
14:08it down what's walkable what's not
14:09walkable so that's like all I'm playing
14:11towards with yeah but the colors here
14:18and also the the forests of course play
14:25into this as well and we here we also
14:26have a very clear case of this sort of I
14:28mean in a very literal sense trying to
14:30avoid not seeing the forest for all the
14:32trees because it's a common area you can
14:33do especially maybe I said quite junior
14:35artist is you would make a very
14:36beautiful detailed tree but then you put
14:39it into an environment and then that
14:40tree is like way too complex and if you
14:43put a thousand of those trees around it
14:44becomes a very very cluttered whereas as
14:46a player you're probably just interested
14:47in like there's a forest here so my
14:49trees are like deliberately like they
14:51have a little bit of a texture to them
14:53and if you look very very closely you
14:54can see you know what kind of trees they
14:56are and you see that from the silhouette
14:57as well but mostly they just blend
14:59together in this larger blob which helps
15:02you read the entire environment faster
15:04and that also ties into the lighting
15:06where like you can see if we just get
15:11back the sunset yeah when the side comes
15:16into the side that from the side the
15:18trade again they use these kind of
15:19normals from the voxel space so when the
15:21light comes from the side they light up
15:24the trees as a blob instead of lighting
15:26up each individual tree just again a way
15:29to sort of build these larger
15:30abstractions and Shea
15:31rather than focus on the individual
15:33components another very tiny thing of
15:37sort of make patterns break patterns
15:38thing is how the when we have this
15:43little slopes in the grass here these
15:45are way patcher than the rest of the
15:47grass right we have these very calm
15:49areas that sort of marks the main areas
15:52of navigability the plains but then you
15:55have these more patchy areas that sort
15:56of draw your attention a little bit more
15:58that mark out the places where you can
16:00navigate between the different levels
16:01there's an easy way to sort of mark up
16:03those slightly more important this is
16:05also a thing that we are it doesn't look
16:08exactly like that in the released
16:10version of the game but this is coming
16:11out in a patch quite soon sort of being
16:14a bit more consistent in how we use
16:15grass and not grass to to communicate
16:20navigability yeah and of course another
16:26interesting like as I said the a very
16:29like in the borders not texture thing we
16:33have the smooth gradient sharp outlines
16:35so there's a bunch of different ways our
16:37clients are getting generated here
16:40there's a bunch of different techniques
16:41to do outlines but one thing I do for
16:43the trees here for example and this is
16:44also a thing I added very recently is we
16:46just turn off the voxels base you'll see
16:53that the like here in the perspective it
16:55looks a little bit weird but you see
16:56that each tree has like a little bit of
16:57a shadow behind it so that's that means
17:00it's drawing each tree twice and the
17:03first time it draws it it draws a darker
17:05pushed back a little bit and one pixel
17:07further out and then it draws it again
17:09in the right place and one pixel further
17:11in in the right color so that means that
17:14the outlines of the trees they will kind
17:16of form a common outline sometimes so
17:18here you can I don't know how Oh
17:19actually I think I implemented a thing
17:21too
17:22does that work yeah it kind of works so
17:31you can see that the yeah the treat like
17:33they form a common outline here and you
17:35yeah they form more of a yeah of a
17:37common blob more than individual blobs
17:38because I push them like that yeah and I
17:43mean obviously here we see also all the
17:45kinds of other
17:45techniques I use for like for the island
17:48itself I use this very I mean it's a
17:50basic very old-school technique where I
17:52just draw the mesh twice once inside out
17:55and once right side out one pixel pushed
17:58out the first time in one pixel pushed
17:59in the second time obviously when you
18:02have the pushed out as far as this this
18:04looks very artifacting you get these
18:05weird like spiky triangles at the other
18:07edges but if you pull it back to just be
18:10one pixel effect it looks it looks good
18:13especially with some interior icing on
18:15it and even more so this is a trick I'm
18:18quite proud of - you see the actually
18:21put a separate slider on that one you
18:23see the little like the shadow under the
18:26grass so that's also all done in shader
18:28and it uses a a trick like it's a it's a
18:32shader function called dvy which
18:34measures how a pixels value changes in
18:38the vertical axis of the screen and you
18:40can use that like to do things like put
18:43a tiny bit of shader underneath a tiny
18:45bit of shadow underneath something in
18:47just one pass in the shader without
18:48drawing drawing like the shadow in one
18:51polygon and then the the grass on top of
18:53that on another polygon that's also the
18:55like that also breaks up a ton around
18:58the edges of the triangles but you don't
19:01really know is that if you pull it down
19:03to just one pixel like that this is
19:06quite neat so it produced an outline but
19:07the outline gives also gives a
19:09three-dimensional feeling to it
19:10a lot of the a lot of the art in bad
19:13North tries to blur the line between 2d
19:16and 3d and that's one way we do it quite
19:20technically also for the forests and
19:23also for the units in the game we use
19:27like they of course billboard which is
19:29also something you don't really notice
19:30because they're so subtle in themselves
19:33but they are just billboarding planes
19:34and we do that billboarding in the
19:38vertex shader instead of doing it in
19:40c-sharp because that would be that would
19:41be quite slow but doing in the vertex
19:43shader is very very fast yeah and so we
19:49have the sharp lines and of course what
19:50produces most of the soft gradient in
19:52the game is the lighting which is a
19:54quite a specialized way of doing
19:57lighting that birds really well
19:59if you have a voxel-based game
20:00especially if you have very low res
20:02voxel-based game so this is what it
20:04would look like without lighting there's
20:06still a bit of normal lighting going on
20:08but I think if we because we can put in
20:11a bit of an overcast and some cloud
20:13coverage then you see there's almost
20:15there's still a tiny bit of normal
20:17lighting happening but it's very little
20:18but if we add the lighting to it like
20:20there's a nice bit of AO to it the way
20:22the AO is done is we do a sort of ray
20:26marker but a very very simplified ray
20:28marcher from a bit a little visualizer
20:30actually yesterday for it so if we
20:33sample from a position here you'll see
20:36how it's sort of a tree going out from
20:40that position so it's actually not ray
20:43marching in sharp in straight lines but
20:45it doing it in a tree structure so if it
20:47stops early it doesn't have to keep
20:48branching out from there so it doesn't
20:50produce perfect lines but it's perfectly
20:53good enough for something as soft as the
20:55shadows we use in bad north so that's
20:58also a way we're like the technical
21:00aspects of it and the technical
21:01limitations of the specific game we're
21:03making sort of tie that into the the art
21:07in a way that doesn't yeah that sort of
21:09makes them all come together as one so
21:13you'll see if I put it put that one down
21:14here you'll see how that area is darker
21:16and it is darker but because this sort
21:18of tree structure gets blocked and that
21:21also gets baked down into an AO texture
21:25so this is like a 3d texture so these
21:27are like the slices you can see the
21:28shape of of the island here and you see
21:32it's like darker here because this like
21:33underwater this is the lowest level and
21:35then it gets brighter and brighter and
21:37fewer and fewer obstacles as you as you
21:38go up from there and having having that
21:41lighting information in a very low res
21:433d texture also mean that it's really
21:45cheap to sample the lighting in all
21:47kinds of things apart from the static
21:50environment so in like the unit's
21:52walking around I can just sample the 3d
21:54textures to get their lighting the
21:55particle systems can sample this all the
21:58little rasp it's can sample this so it's
22:01a very it's very fast to to generate
22:04because it's so low resolution and it's
22:06very it's very fast to calculate as well
22:09because you just have to sample that and
22:11then map it to like an inn
22:13the value and then you have the lighting
22:14so that's one way to produce very soft
22:17gradients that give you a sense of the
22:20three dimensionality of the island
22:22without taking a lot of attention from
22:24like sharp shadows and weird shapes
22:26keying getting created by that yeah and
22:32now I thought we were gonna take a look
22:34at the units of the game it's of course
22:38a very very important part of the
22:39gameplay here we have the three or no
22:42the four basic unit types of the game so
22:44it's like you can hear a little bit get
22:48them walking around - so we got some
22:50archers we got some pike men we've got
22:52some sword and shield guys again they
22:56there's a very basic thing here tying
22:58into the idea of abstractions where the
23:00units move around in groups and they
23:04stand around in squares so it's all
23:06about trying to make the unit's read as
23:08a group and not as individual units most
23:11of the time though a lot of the time
23:13they do break up in individual units
23:14fighting together and that's sort of
23:16when you start to lean into the more
23:17chaotic aspects right so when
23:18everything's going well for you your
23:20strategy is working your defense is
23:21working your units stand around in nice
23:25squares work in unison like the the
23:27archers fire in synchrony stuff like
23:30that but when your defense strategy
23:32starts to break up like and I mean
23:33individual units are like all fending
23:35for themselves in all kinds of different
23:37part of the maps and it becomes quite
23:38like hard to read the battlefield so
23:40that's a fairly central thing of the
23:44basic drama but also to of course the
23:46design of the units is just like what
23:48the trees are they're supposed to be
23:51very simple of just a very simple
23:53silhouette so that when they stand
23:55together in a group they like blend
23:57together as a square there also because
24:01this was like making the units was
24:03actually one of the early on and like
24:05doing the basic R&D for this game
24:07figuring out a good workflow for making
24:09the units was actually one of the bigger
24:10hang-ups but it was quite difficult to
24:13figure out how to do that in a way that
24:15doesn't require a lot of work doing a
24:16lot of different animations require a
24:18very sort of labor-intensive export
24:20pipeline where you like have to do a
24:22bunch of different animations for
24:23different styles of units and export
24:25them that important
24:26unity mark up the sprites and I put the
24:28sprites into different animations stuff
24:30like that and also take like a lot of
24:33texture space doing all the animations
24:35wanted to do so we where I came up with
24:38a what I think is a fairly clever
24:41solution there were all these units the
24:45sprites they use all look like this and
24:48they are that low resolution so there
24:52are three different like animation sets
24:55being used for all units in the game
24:56it's units with no shield no sword like
25:02the guys with my archers do actually
25:05have yeah my archers and there are and
25:07my Spears and there are units with a
25:09sword in the back hand so those are the
25:11guys have a shield in the front hand and
25:12there are the units with just disorder
25:14and those get reused like throughout the
25:17game and you see that the colors are all
25:19weird here and they have like a big
25:20square blobby heads that's because the
25:23colors here are actually UV coordinates
25:25that I then used to look up a deed a
25:29more detailed texture that contains
25:31stuff like the yeah the helmet here and
25:35so the specific style so these units are
25:38all like using the same basic animations
25:40but then I swap out what texture I'm
25:42mapping to those UV coordinates that
25:44also means that I can have quite high
25:45resolution helmets and specific things
25:50that are like unique for each character
25:52but I can have very low resolution
25:55especially since they're so blobby in
25:56the shape I can have very low resolution
25:59sprites that I need to animate and of
26:02course the ones I have to animate I have
26:03to export like a bunch of different
26:04versions of so that that's really
26:05helpful yeah and also since they are so
26:11blobby in the shape we actually were not
26:13doing we're just animating them in two
26:15different directions or in one direction
26:17and then mirroring them or not doing
26:19like towards the camera and away from
26:22the camera not back in front as you can
26:25see them like as you turn the camera you
26:26see them you see these guys flipping
26:28their swords like that which is like if
26:31you are looking for it you do know is it
26:33happening but when you're playing the
26:34game you don't think about it and I've
26:35had a lot of people playing the game and
26:37when I tell them the units are too
26:39they were like okay I didn't know is
26:41that so that's quite neat because they
26:44read more as a group and the group is of
26:46course three-dimensional because there
26:47are like units and spread out in
26:49three-dimensional space in the group and
26:51because some of the units like these
26:53shield guys here also have some parts of
26:56them that are not two-dimensional but
26:58there are three-dimensional you see the
26:59shield's here just floating in front of
27:01them you're sort of tucked into them
27:03kind of clipping them but that's also
27:05since there's just one color you don't
27:07really notice that they don't have an
27:09arm properly or stuff like that and
27:11that's also because the like just the
27:14silhouettes of a guy with a sword in one
27:16hand the sword like the guys kind of
27:18symmetrical the sword is kind of
27:19symmetrical if you mirror that one you
27:21don't really know is it especially if
27:22you're standing in a group but stuff
27:24like the shield that's very very
27:26important in again sort of trying to
27:28play to the right abstractions
27:29expressing the sort of directionality of
27:31the entire group like where's the group
27:33looking what kind of enemies are the
27:35group caring about how are they moving
27:37stuff like that so that's why the the
27:40it's much more important that the three
27:42that the shield is three dimensional and
27:44the character is three dimensional and
27:45that's the same with with the arrows and
27:49the bows to like the bows are 3d because
27:54the like with the sword you can I don't
27:57know if how many of you have animated
27:58mellow things but with a sword
28:00you can get away with a lot like you
28:02just do a quick swipe in the air and it
28:04doesn't really matter in what direction
28:05and then a guy gets hit by that and you
28:07kind of buy it but with a thing like a
28:09bow if you aim pick up the bow aiming
28:11one way and release it and then the
28:12arrow fires that way like it looks it
28:14looks really weird but if we have a
28:16three-dimensional bow then that solves
28:18that quite easily
28:19another more extreme example of that is
28:22of course the spears which we've done
28:24with a separate sprite for the spear
28:26itself so we can aim it in different
28:28direction with the spear is like it's
28:30even more difficult than a boat because
28:32the bow is kind of like a gun you just
28:33need to aim it and shoot it but a spear
28:35has to have a lot of weight to it to she
28:38like aiming it in one direction and then
28:39if the unit comes from another direction
28:42you don't want it just twitching around
28:43with this pew you want to keep it in one
28:44direction you want to point in the right
28:46direction before you hit and keep
28:48pointing about the direction afterwards
28:49stuff like that so that's why the spear
28:51is a separate thing but not the sword
28:54this approached units comes with one
28:56very interesting limitations is I
28:58haven't actually figured out how to do
28:59good looking axis in the game which
29:02might sound like a word thing if you're
29:03making a very Viking themed game like
29:05you might want axes in there but I and a
29:07lot of people have said that I've tried
29:09several times but it's been really hard
29:10making good looking axes work because if
29:15you just put them into the sprite like I
29:17do with the swords they it becomes super
29:19obvious that all the accents in the game
29:21are either pointing that way or that way
29:23and you really see them flipping around
29:26you could do it in 3d possibly probably
29:28but then like when you animate things
29:31that are three-dimensional you need to
29:33put a lot more work into them looking
29:36nice and feeling weighty
29:37than when you do two-dimensional things
29:38so it's possible that an axe could work
29:41like possibly with a big enemy like a
29:44boss character something like that could
29:46possibly have a 3d axe he drags around
29:47and does some animations with then it
29:49could be more gimmicky or stuff like
29:50that but they're very like axis for
29:52regular units has been it has been very
29:54hard so we've just skipped it we do have
29:55axe throwers though but that's easier
29:57because then the axes just throw flying
29:59in the air yeah another thing that kind
30:05of ties into this abstraction idea as
30:07well is that since the game is kind of
30:10zoomed out most of the most of the kind
30:14of animation or how the game feels when
30:17it moves doesn't really come from the
30:20animation of the individual units it
30:22much more comes from the behavior of the
30:25group and how they like from writing the
30:27AI I feel like I'm doing more animation
30:29when I'm writing the AI than when I'm
30:30actually animating the sprites because
30:33it's much more important like how they
30:35react to each other when they back off
30:36when they move forward how they like
30:38flow around when they walk around stuff
30:40like that
30:41then like how the arms moving or how the
30:44legs moving stuff like that because
30:46because the game is so zoomed out I
30:48think I think that's an a good takeaway
30:51in general that when you do animations
30:53for a game it's usually much like you
30:56can make great-looking animations but if
30:57you don't put a lot of work in to claim
30:58the right animation at the right time I
31:00like moving around in the right way it's
31:02probably not gonna look good so it's
31:03probably better to have that good
31:05looking good
31:06movement then good-looking animations is
31:08probably more important to have
31:09good-looking animations than
31:10good-looking characters in that order
31:19yeah so those are kind of some of the
31:23basic ideas going on here
31:26additionally the yeah the the water the
31:29way I've done the water here is it's a
31:32fairly basic it's I get a lot of
31:34questions about it but it's really
31:35simple it's just like a mesh here that
31:38place this sort of looping wavy shader
31:40all over again it's pretty simple shader
31:42and it also ties into the idea of
31:46expressing things in borders and not in
31:48like textures not in field so the what
31:51most of the water is just completely
31:52flat but then at the edge when something
31:55meets the water I highlight that instead
31:59so I don't need to tell you like over
32:01and over its water its water its water
32:03its water its water it's more like where
32:05these two materials meet there's a
32:07coastline that's it and I thought we're
32:10gonna try and actually get some enemies
32:15in here see if that works
32:19I think it did work see what the game
32:22looks like more like in motion and how I
32:26don't know if this is gonna be a
32:27difficult level or an easy level and see
32:31more of there's also of course this yeah
32:32I mean and you get very much this
32:34symmetry idea to with like when the game
32:36actually starts playing so you have this
32:37like quite isolated level in the middle
32:40is usually kind of tall in the middle
32:42and then flatter towards the sides and
32:44you have the enemies coming in from the
32:45side yeah you see them trying to burn my
32:55house down okay this might actually be a
32:58very difficult level down house there
33:00and it also ties into the the order and
33:03chaos kind of thing where your units are
33:05like restricted to moving in grid units
33:09and defending yeah working in grid units
33:12whereas the enemies they just flow
33:14through freely and they your units are
33:16just standing standing on their grid
33:18squares where the enemies like coming
33:19from all over the place
33:20with like ships angled in all kinds of
33:22direction stuff like that the houses
33:24you're defending they are all kind of
33:27aligned to the grid whereas the enemy
33:30ships are like pointing in all kinds of
33:32different directions and of course as
33:34you keep playing the game the the island
33:36turns from this like quite orderly
33:38quaint looking nice
33:42yeah idyllic Island to this very messy
33:46bloody chaotic place so even if you win
33:48the game it still it still looks quite
33:50tragic it looks like like you won but
33:53you know it was still a pretty horrible
33:54battle and a lot of your units died and
33:56so it's been so it's been quite
33:59important this for me to like leave the
34:02corpses and bloody the level as you keep
34:04playing the game also as a way of
34:05environmental storytelling okay I think
34:10I'm going to lose this game I didn't pay
34:12attention really also a lot of the
34:16abilities in the game try to write so
34:19these units here are probably going to
34:21die pretty fast but the the abilities
34:24also I tried to make them look quite
34:26orderly so they feel like you regain
34:28control for a moment so they're like
34:31trying to hold the ground there but if I
34:33send them there they get to do this very
34:36sort of okay now it was just one guy
34:38doing it but they get to do like a very
34:39synchronized motion together where they
34:42are also involve also you're like in
34:43control for a little while until it all
34:45falls apart again yeah I'm definitely
34:49going to lose this
34:54yeah there you go yes so those are kind
35:00of the things I had them prepared we can
35:04just take a look at a few more Islands
35:07and I would be happy to take some
35:09questions from the audience and since
35:12I'm in unity you can all ask all kinds
35:13of things and I can like poke around and
35:15answer thanks if you if you want to are
35:28the islands on the world map
35:29pre-generated for that map and then they
35:32look the same with a procedural
35:34generation in 3d or is it just sort of
35:36abstract in the map and then whatever
35:39they look like no they look the right
35:40way on the map so you can you can
35:42definitely see especially if you've
35:43played the game for a bit or made the
35:45game you can like you see this really
35:46weird thing going on here if we enter
35:49that level we will definitely like we
35:51will see that shape here right so know
35:54that they're correctly they look the
35:56right way the way we do it is we Ghent
35:58like we have this cloud which also has
36:02changed look now in the campaign screen
36:05here you see right so this also again
36:09place into the like the border is quite
36:10detailed but then the this is quite flat
36:13so these are like the behind this are
36:16levels you haven't discovered yet of
36:17course you know you've played the game
36:18but we generate like we try and generate
36:21ahead of the so like the next levels
36:23will have been gotten generated now and
36:25when we generate the level we take us
36:28like a snapshot of it from above baked
36:30that down into a texture that we also
36:33then save to the same file so whenever
36:35you start up and you start up or load a
36:38game it has to generate all the levels
36:39again but if it has already generated
36:41them once then it'll they already know
36:44what they look like and they also know
36:45they also have a seed that will work
36:47with the generation so actually we will
36:49be pushing fairly soon and update to
36:51what some of the tiles look like and a
36:53few new tiles so then if you load an old
36:56save you'll see some of the islands
36:57changing shape on the world map as they
36:59load
37:09hi
37:10so the the island smoothing effect is
37:14that the verdict are you pushing the
37:16verticals around or is it the shader
37:18effect I am pushing the so the thing I
37:21did now with a slider that's something I
37:23prepared yesterday so I am actually
37:25encode like when I because I am merging
37:27all the islands into one yeah into one
37:33mesh and then I'm looping through the
37:35looping through the vertices and all
37:36pushing them the same way but it's not
37:38enough to just push the the mesh itself
37:41because there are all kinds of other
37:42things that tie into as well right when
37:44I run the grass placing algorithm for
37:46example that has to run on the same like
37:48slightly modified world and then the
37:50navmesh of will the grass runs on the
37:52navmesh but the navmesh too has to be
37:54modified in the same way as the island
37:57so here I just did like I do that but
38:00then I also saved the offset so I could
38:02then in a shader like reapply there the
38:04old offset but yeah sorry I'm taking
38:08another question are you using this both
38:10in that mesh or your own no though I
38:11built my own it's it's even like you can
38:14actually see if you look at the
38:16different like the green part here in
38:20the meshes that's actually the nav mesh
38:23so I build that in like I build the nav
38:25mesh pieces in Maya just like I would
38:27build the regular mesh then I asked when
38:30it's already built when it's built the
38:32level it takes all the navmesh pieces
38:33and type together and build a nav mesh
38:34out of that actually writing the nav
38:36mesh is really fun and I can recommend
38:38writing your own nav mission it's much
38:39more fun than you even think it is a lot
38:41of very very small math problems how
38:47much work in precision has to go into
38:50making each like module to make them fit
38:53together yet to make them victor yeah
38:54good question so there is some there's
38:58different ways you could do this but the
38:59way I do this is I from side I generate
39:02a hash value from like the positions of
39:05the edge but then I round them somewhat
39:07so it doesn't have to be perfect
39:10precision but since it's a rounding
39:12thing like if I end up if I have two
39:14things that are almost exactly the same
39:16but they just happen to be on opposite
39:18like sides of the rounding then they
39:21won't fit together so that especially in
39:25the beginning it was a little bit of a
39:27hassle but now I have a if you just do
39:29the work right in Maya and you just like
39:31if I build a new piece I copy a bunch of
39:33old pieces that have the right sides and
39:35then I keep those edges right and then
39:37tie it together and then it's actually
39:39very easy but you have to keep that in
39:41mind and that's also why it's so
39:42important that I have this like
39:45visualizer here right so I can if I've
39:48if I build some new interesting pieces I
39:50can actually make sure that like yeah it
39:52fits together with the old things like
39:54it's supposed to yep if you want to get
39:59into procedural generation where should
40:01one start good question
40:04well the so I showed in actually the
40:08algorithm that built the thing in the
40:11middle here that's a pretty good place
40:12to start
40:13it's called well this one's 3d so it's
40:16called marching cubes but the 2d version
40:19is called marching squares it's just
40:21like it's a very simple algorithm to
40:22place tiles on a grid or in a voxel
40:26space it's very easy to understand and
40:30very easy to start working that's not
40:31the algorithm I use in bad north though
40:33the pieces look very similar I'm not
40:40have seen wrong but when you show the
40:42this scene and you see the island it
40:46looks like you flipped the island as you
40:48get that reflection effect yeah water
40:50yeah is that done entirely in in shaders
40:55it's actually a model underneath the
40:56water I mean it is like it would be the
40:59same which way I would have done it but
41:01it is it is a model and well it's both
41:04actually it's because I whenever I
41:07create like some things are marked up
41:08that they're supposed to have an outline
41:09or outline or they're supposed to have a
41:11mirror and then if they're supposed to
41:13have a mirror that I just like create
41:15another instance of that game object
41:16that has the mirror and then I in the
41:18shader it then flips that upside down as
41:22well like flipping it is obviously quite
41:23easy but it also does some things too
41:25like it simplifies a lot of the it just
41:27it runs the same shader but there's a
41:30lot of places where it skipped some
41:31stuff
41:31that aren't needed when it does the
41:33mirror and of course it applies like an
41:35almost uniform color to it as well but
41:39you're correct in that it is a flipped
41:40foot mesh um hi so you may call the
41:51model components and you do all of the
41:52coding as well would you consider
41:54yourself a coder who does art or an
41:57artist who does coder very much an
42:00artist that does code because I come
42:02very much from an art background and I
42:03don't really do much code outside of
42:05code that in the end will produce some
42:07kind of art but I'm pretty far down the
42:09rabbit hole now so yeah hey awesome by
42:21the way it was super interesting and I
42:23know there are a lot of students here so
42:25I'm gonna ask the very helpful student
42:28question what's your workflow and can
42:32you think like everyday work day and
42:35contribute yeah sure
42:37so like as I said we work in a very
42:40small team I start the day with well I
42:45don't start the day like but when I've
42:46been up for like three hours I do a call
42:51with my colleague Richard who's just
42:53woken up and then we yeah we do a
42:56stand-up meeting that's fairly this
42:59first standard I've done that in all
43:01places I've been at so like just go
43:03through like what did we do yesterday
43:04what are we doing today are there any
43:06news stuff like that then the rest of
43:09the day like as I said I work from home
43:11so I have a pretty free schedule it
43:15really depends it can be very very
43:17different I like some some days I spend
43:19the entire day in visual studio writing
43:21code sometimes I do a lot of art things
43:23it's it's pretty much all over the place
43:24yeah it's very something that's like a
43:27lot of people ask me working from home
43:29is it hard to be disciplined but it's
43:31more the opposite it's very hard to get
43:32away from the computer
43:33like I I usually sleep quite badly
43:36because I like go straight from the
43:38computer to bed which is a very very bad
43:40idea don't do that
43:45to follow up on that that you're working
43:47from home do you have a need like
43:49techniques to stay working yeah
43:53don't have any other hobbies than work
44:01no but that's that's basically true like
44:03I haven't really played any video video
44:05games for a long time which I probably
44:06should do but it was a long time ago so
44:08played a video game and all my other
44:10hobbies required like friends or
44:12physical activity so I can't do them all
44:14the day anyway so when I'm at home
44:16that's like the only thing I get like
44:17when I'm done with my choice that's the
44:19only thing I can do do you do you in
44:27your team have a specific goal for the
44:29game other than just making a successful
44:31successful game that creates brenigan
44:34I mean yeah from for me it was very like
44:40for me it was very much also just
44:43designing a whole the game is very much
44:46just designed around what I think is fun
44:48to do and that's been I think that's a
44:51good especially if you're a very small
44:52team I think that's a very good approach
44:54to because to have so because it's very
44:57easy to especially when you're
44:58discussing something with someone else
45:00to fall into things that seem
45:01objectively defensible which would be
45:03like yeah how do we make more money or
45:05if you have like idealistic goals that
45:07could also be a thing but it's harder to
45:08defend like no I want to do it this way
45:10I think it's more fun to do that kind of
45:12work but I think it's very important
45:14because if it's fun to do the work
45:15you'll do it better you'll do more of it
45:17and like you won't be stressed out and
45:19you'll you'll like it more so that's
45:21been a very that's been a very big part
45:22of bad North trying to just design a
45:25game around the kind of things I enjoy
45:27doing yeah yeah hi I haven't played the
45:34game but I'm just assuming here that for
45:37each island that you get to get an
45:39increasingly harder that my question is
45:41then is the procedural generation of
45:44every island like does the island
45:47geometry like get more difficult and how
45:50I'll have you like program that in that
45:52that you just tell the like the
45:55algorithm to generate a harder level for
45:57me like how does that actually work yeah
46:01so no we don't do any of that we just
46:03make them bigger because that's that's
46:06easy but no it's quite it's quite an
46:09abstract concept to figure out what
46:11makes it because the shape of the island
46:13does matter a lot for how difficult the
46:15art especially especially depending on
46:16your play style and what your loadout is
46:19like are you playing a lot with archers
46:21or with Spears or whatever but there's
46:25those things are quite abstract and
46:28they're quite hard to put it like first
46:29of all I don't know if we have we have a
46:31great grasp on that and even if we did
46:32it's not obvious how you would evaluate
46:35that like if you give an island like how
46:37do you break that down and evaluate a
46:39level and give it a difficulty score and
46:42Eve then if we did that we would also
46:44then have to figure out okay but how do
46:46we then generate an island that tries to
46:49achieve a certain level of difficulty
46:51like that's quite an abstract and
46:53difficult problem which like sounds
46:54super interesting but we've just not
46:56done any of it we do we do I mean we
46:58just switch up also so that like the
46:59earlier islands are less complex they
47:02use I mean of course because they're
47:03smaller but they also use fewer modules
47:06I have some modules that are like marked
47:08up so because they're themed right so
47:10you get some themes throughout the game
47:11that we keep switching up as well also
47:13like a way to yeah keep breaking up the
47:16pattern so yeah we have these I mean I
47:20think we've seen some we've seen some
47:23caves a level we played first had caves
47:25also here's like some some ruin style
47:27things with like some walls and stuff
47:29like that those kind of things they
47:31break up the gameplay a little bit
47:32because like the caves or you need to
47:34keep track of all the different weird
47:35paths the units can take yeah but it's
47:39mostly as an aesthetical thing yeah so
47:45we vary the difficulty just by sending
47:47more enemies and then hope that the
47:49randomness from what the ways the level
47:51looks aren't like too far out of bounds
47:53was there any conscious reason why you
47:56choose to develop the game in unity
47:58instead of another engine I don't know
48:02if I would call it a conscious reason
48:04but I've like I've been doing pretty
48:06projects in unity as well I quite like
48:08it I think unity is a very good way to
48:11get into coding that's how I got into
48:14coding because it or my understand I
48:19haven't worked much in unreal like I've
48:21done some student projects in unreal but
48:22my understanding is that the this sort
48:25of the border between what you do in
48:27sort of normal engine tools and what
48:29you're doing code is moved quite far
48:31ahead in unreal like and it's quite a
48:33different space on the other hand you do
48:35have the very like powerful blueprint in
48:38their scripting system which like is
48:39great but it's not as good as writing in
48:41c-sharp like you can't do ask like you
48:43can do a lot of things but you can't do
48:44as much especially not for the same
48:46performance right but then in an unreal
48:50you would then instead go the extra step
48:52after that and start rewriting the the
48:54source code in C++ which is like a
48:56bigger threshold so it's like it's
48:58similar thresholds but they're moved in
49:00quite different places and to meet it
49:02for my current expertise it fit me very
49:04well to write the code in C sharp which
49:07is like a fairly easy language to learn
49:14Hey
49:16so you posted on Twitter a lot of your
49:19progress and how you did things and was
49:22that part of your marketing strategy
49:24because it definitely got me hooked and
49:26made me buy the game so if it was it
49:29worked
49:30yeah I mean I mean it obviously it
49:33obviously was but I I didn't post on
49:36Twitter because I wanted to do marketing
49:38it's more like I posted on Twitter and
49:40that's also marketing because that's
49:42especially I really like showing things
49:46I do but I think I come up with
49:47something clever I want to show people
49:49and especially since I work from home I
49:51don't have colleagues well I can show
49:52Richard but I don't have that many
49:54colleagues to show so then I can post it
49:56on Twitter and show people there instead
49:57which is very nice when he's very it's
50:00very encouraging to although all the
50:02positive feedback you get and I find
50:04also a common pattern in my workflow is
50:07that I do something and then I sit
50:09around tweaking it for too long but then
50:11as soon as I make a gif and post it on
50:14Twitter then I'm done with it and then I
50:16move on to the next thing
50:26yeah
50:27so how what's the transition from
50:31working in yourself to becoming an indie
50:34developer and how can you use that
50:36experience use the Ubisoft experience
50:39yes yeah good question I mean it's a
50:43very big difference the biggest and
50:46nicest difference is that there's just
50:47less time spend in meetings because
50:50that's quite like you can be however low
50:52level you want you're still going to
50:53spend like a significant amount of your
50:55work within meetings in a big company
50:57like that and obviously it's way less
50:59trust like the turnaround times for
51:01everything is much quicker and obviously
51:04I don't have to like I can take the
51:06decision straight away I don't have to
51:08wait for a chain of command so that's
51:10much nicer but I do think I benefited a
51:12lot from it because I got to meet so
51:16many like talented senior people work
51:20like there's just such a huge reservoir
51:21of expertise in such a big company so
51:24many people to learn with you get to sit
51:26next to someone who's like been in the
51:27industry for 15 years and learn all
51:29kinds of things from them for me was I
51:31feel like it was a really good place to
51:33get started it also very good thing to
51:35have in your CV and like if this indie
51:37thing fails which it does for many
51:38people yeah but I enjoy it this way more
51:48so it was mentioned earlier just before
51:52the last question about your posting on
51:54your progress with that North on Twitter
51:57have you considered like setting up like
52:01I'm part of your website or something
52:03like just an overview of the progress
52:05and how each thing has been done or
52:09would that be like I don't know no I
52:14mean I haven't thought about that
52:15specifically I've thought a lot about
52:16like would it be good to gather some of
52:19these things into more than just like
52:21tweets I could keep sending over maybe
52:23it would be that's also probably a lot
52:26of work probably not I mean I don't know
52:29I'm thinking I'm thinking about like
52:31probably not for about north but I've
52:33been thinking about after this I'm not
52:35sure I will but I might do some other
52:36more just demo projects like the the
52:40house and the planet thing and doing it
52:42in a way where I consciously think that
52:45I will make like I will tutorial I some
52:47aspects of this I would like like maybe
52:51make sure the the code is quite clean so
52:53I'd be happy to share it if I would have
52:55to in the future stuff like that so not
52:57probably not with bad north but but but
52:59I do I do like explaining things and
53:01show us things off but it's also it's
53:02also a lot of work those kind of things
53:04so yeah maybe something in the future
53:18so I had one question about a solution
53:22true the exes but what a solution about
53:26the axes yeah because I don't know if
53:30it's about the how you've planned the
53:32aesthetic of I'm bad north but would it
53:35be blasphemy to put in double-sided axes
53:37yes okay all right I think I might be
54:02deep into the the dark zone of what
54:04you're want to say and I have to keep
54:09quiet but have you thought about what
54:11would be the problem of why people would
54:14stop playing your game and what solution
54:17would that be oh sure yeah I mean people
54:22would stop playing it probably because
54:23there should probably be a little bit
54:25more content in it
54:27we should probably wanted to put way
54:28more like items and abilities in there
54:30that we just hadn't had the time to make
54:32it's probably coming for in Compton like
54:33we have a big content up but we have all
54:36this on Twitch okay whatever
54:38we have a
54:39we're gonna do some content update and
54:42then we'll hopefully put in somewhere
54:43those things
54:44so that's one big thing obviously it's
54:48also there's a little bit of a
54:50disconnect between which I don't find
54:51problematic myself but there's a little
54:53bit of a disconnect between the game
54:54looks quite sort of cute but it is quite
54:57hard actually it is very easy to get
54:59into and learn the controls but it does
55:01get quite difficult and there's
55:03permadeath in it right so there are like
55:05definitely some people who get dragged
55:07into the dragged into it for the
55:11cuteness and then also find all this is
55:13strategy game that's actually quite fun
55:14with strategy but then like it does get
55:16quite difficult and quite punishing and
55:17some people definitely get surprised and
55:20maybe put off by that after a bit yeah
55:26all right I think we'll leave it at that
55:29yeah thank you very much
55:30[Applause]

About this transcript

This page contains the full transcript of Konsoll 2018: Oskar Stålberg - Developing The Bad North Look by Konsoll, generated from the public captions YouTube serves with the video. The transcript has 10,311 words across 1,396 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to translate the transcript into 100+ languages.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, translation to 100+ languages, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.