Beyond Authentic Pixel Art in Unreal Engine | Unreal Fest Europe 2019 | Unreal Engine

>>Chris Wilson: Hello everyone.
Thank you so much for coming. I know you have got
a great choice of talks, and unfortunately
you are at mine, but I will try my best. I am going to talk to you today
about our pixel art game. Specifically,
how we have gone from it looking like this
a couple of years ago, to looking a bit more like this, and finally, we are
in this kind of space now. Specifically what I want to
cover is, how we have taken a bunch
of classic influences and how we have turned that into
something that looks authentic. How we built that basis. Then, how we have used UE4 to go beyond what we
considered authentic but, in a way that is still kind of, you know, respectful
to the source material. Finally, as not many people
are using Paper2D, I really don’t know why,
just a couple of tips that I wish we had known
before we got started. First, a little summary
just to give you some context about myself
and the company. I am the Design Director
in our six-person studio, Cardboard Sword. Day-to-day that kind of
means that I am responsible for maintaining the vision,
doing the bare level design, gameplay that sort of thing. But, if you speak
to the artists, I am really there to make their
life as difficult as possible. I do not necessarily disagree. I have been
working independently for over
three years now, I am still alive,
and I am still able to eat. I do not think
it is going that badly. Before that, I worked in AAA
for about nine years, the last thing I shipped
was Forza Horizon 3, a large open world sort of
simulator-ish, racing game. I am not working on
anything like that anymore. I am working on The Siege
and the Sandfox. Now, The Siege
and the Sandfox is what we are calling
a Stealth-Vania, and it is a phrase we made up just to quickly
explain the game, but it really has stuck. It really just refers to the two
main pillars of our game. Those pillars
are obviously stealth, there is a focus on taking
the long, the scenic route,
not just rushing in. Taking time
to consider your options, being stealthy, being smart.
All the mechanics support that. The other side
of exploration, exploration like you
might remember from classic Metroid games or any Metroidvania really,
it is a common genre. We have this
large contiguous open world, progress through it
is gated through a mix of just players mastering
the Park/Hold controls, which have a fair degree
of depth to them, and also the classic sort
of unlocking abilities as time goes on.
It is also a pixel art game. Probably should have mentioned
that and in theory that is why you are here. But, yeah, I sometimes
forget after all this time. Now a common misconception is
that the artists on the project have worked on lots of
pixel art games before, that is something
they specialize in. Actually,
that is not true. They too came from
a AAA background, they worked on things like,
Unreal Tournament 2004, Unreal Championship,
and Elite Dangerous. If you know modern 3D sci-fi
hides hard surface stuff, anyways,
nothing like this at all. But like me, they fancied
doing something a bit different, and that is kind of why
we all came together. Why we decided to make a 2D
style vania is for another talk. But, let us just say, we really liked
Mark of the Ninja, and felt there was space
to make something like it. The first point to start
when we all got together was to assemble
a list of influences, classic games
that we liked the look of and we wanted to break down
and see if could use that to basically form the basis of
the art direction for the game. We started with one
that keeps coming back, which is a Batman game
by Sunsoft on the NES or N.E.S., if that is how you choose
to live your life. I have never heard
of this game before, I have still never played it,
don’t tell anyone. But, the artist always uses this
as their primary example, and in my opinion actually,
it is quite easy to see why. If you look at this
one screenshot here, Batman is three colors. But you can still recognize
that he is Batman, and nothing else in the scene is really more than
a handful of colors itself. You know, the NES was
very limited in that regard. Also, the other thing, is that
nearly a quarter of the screen is completely black,
just one color. That is obviously
a technical limitation, they could not fill that
if they wanted to. But here it is being used
with purpose, its negative space is drawing
your eye toward the detail where they have spent it,
which is obviously on Batman. You can see this in more
recent games like Zelda on this NES,
recent is a relative term. There is no pitch black
in this image, but there are large areas
for the grass and water of just solid color,
and again, the same principle. Not much detail there, draws
your eye towards the center, where they spent the time,
attention and the detail. That is something
we wanted to do. The same thing with Metal Slug. I do not think you can
really talk about pixel art without talking
about Metal Slug. We do not have any specific
thing that we like about this, just the general high level
of attention to detail. Everything is so polished, the animations
have so much character, this is kind of a high bar
particularly, just the way the character
bounce up and down, it is just so charming. What were the common themes
among those things we just saw? Well, I would say
that they embrace the technical limitations
that they had. They did not fight them;
they saw what they had, and they worked with it. That mostly includes the
restricted color palette again three or four shades at most,
and they use this combined with, you know,
lots of negative space. Lots of areas where
technically there is nothing, but it is pointing towards
something more interesting. We felt we had
enough information to actually start putting
something together. But before we rushed
into using a game engine, we wanted to start by
just mocking something up in Photoshop.
Now as discussed earlier, the artist had never
made pixel art before, so I want to stress it is not
a level like a final scene. It does not match
any of the reference we have seen in terms of detail. But you can see
the beginnings of the game. We can see, we have got
the negative space, way too much of it
but we can see it is there. We can see we have
the limited color palette it is way too limited.
It reminds me a lot of Quake 1 or Thief or something
but the principles are there, and we can see it
is probably going to work. I have to show you animations because the Metal Slug
influence is clear. I love that these are still
charming animations even though we have got
much better ones now. We actually started
with Gamemaker, because we started this game
a long time ago and at the time, Unreal Engine
was still subscription-based, and we just wanted to feel
it out a little bit first. This was the first level
we threw together. Now this looks a lot better than
that reference image thankfully, but I think you can see
that it is still adhering to all those guidelines
we put together. There is still a large amount
of negative space, it is drawing your eye
towards the detail, and while the scene looks
much more colorful overall, each individual element
only has a handful of colors. More than Batman, but still
you know severely limited. You can see
this most obviously when you look
at any of the gradients. We have very few gradients,
and basically no translucency because by our own limitations,
we cannot do that. We did this,
and we felt pretty confident, we felt that we had something.
But the problem really was that, Gamemaker was not
the engine for us. It is a fantastic engine, but obviously looking
at our AAA past, we already have experience
with Unreal Engine. We have aspirations
to make 3D titles in the future, imagine that whole
third dimension. It just was not
going to work for us. You will be shocked to discover;
we did move to Unreal Engine. We basically
took the Assets that we had from
the Gamemaker version, threw in some
very rudimentary lighting and put it together
for a Dev Grant Pitch. It ended up
looking like this, which looks dramatically better
considering it is exactly the same Assets
from the previous shot, just with some lighting. The lighting was
a pretty easy win, we pretty much just turned it on
and it looked a lot better. Some of the shots are more
dramatic like the early one, some of them are
quite subtle like this. I admit, you can barely
tell the difference. But that is okay, because we were wanting
to build upon the pixel art rather than do something
that disrupts that. We often got asked
even at this stage, “How have you got this to work?
Because officially, Paper2D and the materials
you are provided with, do not work when they are lit”.
I will be honest. Pretty much, just turn it on
and it did work. There are some problems,
we did not get shadows and there were
a few little issues, which I will come
to later. By and large, we got
what we wanted from it without too much trouble. Come on kooky thing,
there we go. Lighting was as I said
an easy win. It pretty much just worked, and it made things
dramatically better. But we had to
kind of be careful here because now we are
in a situation where we have a lot more power
than we had before. We can apply all kinds
of crazy effects, but we are really going to start
ruining what we have if we start attacking
this foundation. We needed to sort of, needed to move beyond what we
had actually used to start with. We needed something, you know,
a bit of a stronger rule set. The way I wanted to
handle this was to look at things
in terms of authenticity. My best example of authenticity,
I think, actually comes from my days
at Playground Games. Now, it is an open world
racing game and it is broadly a simulated,
but not exactly. I would argue that really, all of those cars
handle authentically. How you expect them to,
not realistically and I will give you
a specific example. If you fire up the game,
open a Ferrari F40, a notoriously hard car
to drive in real life, I am sure most people
who play that game, people who do not even drive,
genuinely feel that that is what it would be like
to drive that car. It feels authentic. However, if I done handed them
the keys to the real one, it is not really
going to work. I say that is what we are
doing with the art here. What we are saying is,
we want this to look like you genuinely believe
those old games look. But actually,
if you went back to them, that is not how they look, and it is a kind of fine balance
to get that right. Our starting point
to do this, was to respect
those core limitations we have talked about.
We used those as our foundation, we are not going to compromise
those wherever possible. Anything we use on top of that, is going to enhance something
that already exists. We are going to use it
with intent. We are going to say,
“Why are we using this effect?” The answer cannot be, “Because I can just press
a check box and it works,” because that is often the case. The biggest limitations
regarding what they could do with pixels
in these classic games, or more accurately,
what they could not do really, they could not scale them
or change the size of them. You are always
going to really count the pixels
across the screen. They’re always going
to be one size, you cannot arbitrarily scale
them for the same reason, it just was not
really possible. We are not talking
about Mode Seven for anyone old enough
to know what that is. But you could not
rotate them either, they are all things
you could not do. Easily summed up with this
one image from the game. On one side of me, we have got
the painstakingly hand animated version
of this water wheel with 16 frames of animation,
it took a fair amount of time. On the other hand, it is just
an image that has been rotated. I will be honest, a lot of people probably
can’t tell the difference. I will be honest I will have
a look on a screen of this size, I can tell the difference.
Hopefully some of you can, but it does not really matter
if you can or not. It is a very subtle thing and as you will see
going through cumulatively, all these little things
add together that make the game
feel the way it does. I think honestly,
even for players, subconsciously they
will appreciate this even if they cannot
point to it as a thing that is actually
making a difference. The same with a limited
color palette, if we bring up Batman again,
he is three colors. I will give an example;
this is even applied to me. When I was preparing
this presentation, I thought, “Oh, he is about six
or seven colors.” He is not,
it is literally three. Even in my head over
the space of a couple of months I am misremembering it. We are just
kind of trying to do that over a longer period of time. You look at this, our character
is about 16 colors, but he is still in both of them.
You have got these areas of like you know flat
shaded single colors and you will think,
“Yeah they look about the same.” But when we put them
side-by-side, they absolutely do not. It is just enough there
to evoke the memory of those classic games
without being, you know, restrained too much
by what they can actually do. We extend the same thing
to our tile set as well. We have got no more than four
to six shades of any color in there
and the whole lot is no more than 80 or 90 colors
for the entire world. Just by limiting yourself
in that manner, that necessitates the way
the artists are going to work. You know,
it is not full of gradients, there is no fake translucency, because we literally
cannot afford to do it. As you can see in the tiles,
there are lots of them, maybe 50% or more black
and that it’s just fine. Another obvious one,
but worth pointing out is, we use an Orthographic camera
so do they, why not? Even at this stage
I should point out, lots of people told us
this was probably a bad idea, but we ignored them.
We will come back to that later because they were probably
wrong to do that. Another thing that has become
popular with lots of, in the pixel art games
is to apply normal maps. What a lovely normal map that is
and we thought about it. We did strongly think about it
for quite some time. I think the opinion
among all of us and the artists
was that really often it is just destroying
the sort of classic aesthetics of those games. It really,
really removes the charm. I think the obvious exception
is anything is a bit more hard-edge, in sci-fi,
like traditional normal maps can add a lot to that.
But it was not working for us. But then saying that,
it was easy to try, and the artist gave it a go
using the traditional method where they were using it to add,
surface detail to things in the same ways
they would for 3D games. However, the surface detail just
came across as surface noise and for the reasons
I have highlighted we just didn’t want that. Basically,
we got rid of all of them and the artist
deleted it so much so that there is not a single
image of it that still exists. That is why you have been
staring at this terrible placeholder, sorry. Just to summarize
those simple rules, the main one really is that we
want to maintain pixel purity. We don’t want to distort
the pixels or resize them or do anything drastic
to the whole scene. The more you do that,
the less it feels like one of these classic games
and the more obvious it is for someone
to point to that thing and go, “No, that didn’t happen
in those games that I remember.” Likewise, with a limited color
palette, we are not as limited
as they were, but you can count
the different shades of colors. We still use the same camera and we still just avoid
the normal maps because it just wasn’t
working for us. But even at this stage
there is always going to be some opportunities to bend
the rules and be flexible. Our rules were basically that if there is a clear intent
behind what we are doing, we could see
what the advantage was, and if it was just supporting
an already existing feature, it could be considered. The first thing we tried
was to try and solve a mechanical issue
we have. Because obviously, we have
the general lighting system and things
are lighter and darker. But also, as a stealth game, it is really important
to telegraph to the player if they are visible
to enemies or not. That is more of a binary state,
you are, or you are not. We did this using lookup tables,
very simple stuff really through the arts,
but obviously, not something they
would have done at the time. I would say this is really only
a bend because, while it is using
more colors overall, each set has the same limits, so it does not really cause
too many problems. Just a simple way that,
in a modern engine, it made life
a lot easier for us. Something slightly more
controversial at the time, is that we were
in some instances, actually distorting the pixels.
This kind of snuck through, mostly because it is
incredibly localized, it is in a very small area, so we are not deconstructing
the whole scene. But also, these lights
we used everywhere, and it was just supporting the flickering effects
that already were. It was not adding something
completely new. However, this did give us our
first window into the troubles we might have with
the orthographic camera. This affect again, not crazy complicated
by the artist on a mission, but they assumed they would be
able to do it in the same way they used to in the past,
using a ray of… refraction techniques.
Unfortunately, those do not work with the Orthographic camera
as we discovered. While this is not
a huge problem, it just meant they had to spend
a little bit more time thinking of another way
to do it. Again, super simple. A high like this, not because
it is incredibly complicated but just because it was
an example of a situation where something we assumed
would just happen actually took
a bit of work to get around. But the effect looked nice and we used it
for some more situations, some other limited,
localized distortion. I think looks pretty nice. If you want to see
more detail on that, we did actually do a blog post
on it a while back now, so feel free to have a look.
Now, outside of the art, the other thing
that is obviously going to really contribute
to the look and feel of these is how
the levels are put together, which unfortunately is my job. The biggest most obvious thing
is it is tilemap based. We have a series of tiles;
they are all 32 x 32 pixels and they are all aligned
to a grid. This is super important
because obviously that is the same
limitation they had, so we never deviate
from that grid. Except when we do.
If I give you this example here, while I have got some props
in the background now, they are highlighted. They look okay they fit neatly
into the background. But it just looks so much nicer if we just move them
around a little bit, it just feels a little bit
more organic. These guys are sat
on a 16 x 16 grid. again, bending of the rules.
They are both on a grid, they don’t just share
the same grid, and again, I think this all goes
in the category of things that are so subtle I don’t think
people will notice. But together,
they kind of compound and make things
feel more polished. To give a quick one for of
how different layers work, we will start with obviously
the most important one, the design layer. This is basically what
determines all the collision and it has some metadata markup
for if tiles can be seen through by the player or enemies and also some of the things
that park or running up walls, holding onto things,
that sort of stuff. Then the artists used that
as their guide to paint over the foreground,
that is pretty simple. Then they add some detail
to the background and then most of the extra
sort of little polish comes in the foreground or the super-foreground
as they call it. Like the most close
to the camera as possible. It still has that classic look
but, you know, we have added more layers
then you might have had at the time
and the half tiles as well. Then, once the tilemaps got
everything in it that we need, we just put it in a standard
Unreal level file and add in everything else
that we might need. Lights, animated Sprites,
post-process, gameplay entities,
that sort of thing. That gets us to here,
which is sometime in 2017, which seems a long time ago. Hopefully you can see
that it is still there, like the classic thing, the classic aesthetics
are still there. You can still count pretty much all the pixels
across the screen, you still got the areas
of negative space, but everything looks
a little bit nicer. The lighting is doing a bit more
at this extra depth. However, it is still
a little bit on the flat side, so maybe we can
actually move beyond this. I would say that
we got to this point where everything feels authentic,
everything looks like you honestly
remember those games look, even if they do not.
Now, we have been working with Unreal and 2D Engine
for quite a while, and we were confident
that we knew what would look good
and what would not and maybe it was also time
to re-evaluate some of the decisions
we made in the past. The first one to revisit
was normal Maps. We were still
pretty much convinced these did not really work
for our kind of game. But then Dead Cells happened, and Dead Cells
does a fantastic job of maintaining that chunky
pixel art aesthetic, while at the same time,
having normal maps. We had a go at it,
and these were done by hand. They were done in
a different way that they might approach
normal maps through a 3D game. Rather than trying to add
more detail to the surface, instead though we are using
the normal maps to sort of sculpt
the broad shape, so when the light passed over it
enhances the chunkier shape rather than trying
to add more detail. Now as I said, this was
something they did by hand, as a little step-by-step there, and there is a full Karma Sutra
article about the thing, linked at the bottom. I will be honest,
as a non-artist, I get about halfway down and it starts
to look like this to me. But if it is all good for you,
you can do it. It just takes quite a long time. We often refer to it as
sending them to Rainbow prison because it was
very time-consuming. Now you may be thinking, “Well, okay
that is very time consuming, but can you give some automated
process like these ones?” Obviously, there is
and shape recognition. We did try them and if you see
these in isolation, the light has an impact. It is doing something,
but I would argue that the hand-painted one
looks much better. The artist who made
the normal map made the Asset. They know in their head,
you know, what this is supposed to look like for a 3D object,
and I think that comes across. Now I should point out that
they could have spent more time trying to refine
the source Assets used for the grayscale conversion
and the shape recognition. But their very strong argument
was, “If I am going to spend the time
doing that, why do I not just make
the normal map myself?” and I did not really have
a good counter argument, so that is what they did. I think it is pretty clear
in that one, but I think it is particularly
obvious in this next example. If we look at the other two
methods there, it is doing something, it is
adding something to the scene. But if you look at the wheel
in this example, it is just so much more defined,
the wheel shape. You know,
the shape of the object in a way that only the artist who made the Asset
is going to be able to do. But it does not have to be
that dramatic. In this example, that’s before, that’s after, you can barely
tell the difference. But that is okay,
particularly in a static image, we are not looking
for it to scream in your face that we are using normal maps. It is just all adding
to the subtle things that are enhancing that
pixel art already there. You know, we can combine it with
particles, and it looks nice. This is basically
trying to emulate sort of the first
sandbox example we had. That is fairly subtle I believe.
This is less subtle, but we feel like after a while
we kind of earned the time, earned the ability
to push it a bit harder and obviously by the stage
none of this is stuff that would be possible,
I mentioned at the time, but we feel that we have
people along for the ride now. It is okay to occasionally
do things like that. Something else that we revisited is the way that we were
shifting around colors. Before this was done
with a lookup table, which was used
for these two versions, to say if you are visible
to MPCs or not. But it was not that flexible, if you wanted to do it
for lots of characters and sort of mix
and match it was time consuming. Also, someone like me should not
be messing around with those and potentially damaging
the source material. It was a bad idea all around. What I really needed
was something like this, some way that I could just go in and just change the colors,
see what I felt light, see what I thought
looked good in real time without ever really touching
the actual pixels. I cannot ever ruin anything which is
the artist dream really. This allowed us to do that. The way we did it is that we
would take the color image we converted to grayscale,
a specific set of grayscale, and then we would use
a material instance to bind those different
shades to different colors. Now, the problem is,
a very minor problem, but again sort of shows how we had to keep doing things
the hard way, is that obviously, you cannot
just convert to grayscale. Because, if you do that often, they just always end up
the same shade. The compromise really was that
the artist would open up in Photoshop,
open the source Assets, manually convert it
to a predefined set of colors for a single frame. Then they would have saved that
as an action script and go back through
and just do all of the frames. Those are saved
for all the characters. Now, the specific
set were chosen, so that we could go through
each step and generate some masks for each
of the different sets of pixels. That means it is incredibly easy
to make all kinds of weird and wonderful
color combinations. The two in the middle,
we used for like tutorial ghosts and last known position. The last one
is horrendously ugly and is literally hitting
the randomized button. But just an example
of we can do anything and that made things
much more flexible. Again, you know,
it is something only possible in the modern engine,
but the end result is something that does not look out of place
in the classic pixel art game. That takes us
to sometime last year, where the differences again,
they are much more subtle, but we have not compromised
on the core there. I think it is still
pretty much the same game but it just all feels
a bit more solid. There is one more thing that
we did need to look at again and that was
the Perspective camera. Now, for a long time,
we just kind of got on with it and lived
with all the weird foibles and just stuff that you would
expect to work and would not. Every now and then, we would
speak to people at Epic and say, “We are having some trouble
getting this to work in the orthographic camera.”
They would say, “Why don’t you just use
the Perspective camera?” We just think,
“What do you know?” Then we realized that probably, they probably knew
what they were talking about and we should probably listen.
However, our concern was always that we would end up
with something like this. This is drastically exaggerated but we would end up with all
the layers swimming around and it would just look
very strange and just, we were so worried
that that would happen. We just overlooked
all the trouble we were having. We never got shadows to work, a lot of the more advanced
rendering things we wanted to play around
with just didn’t work. None of our walls
could really act as solid things
that would block lights. One day, one of the artists
in frustration just went, “Should we just try it?”
and it was pretty much fine. We did have to squish down
some of our layers and depth, it was all very flat
in that dimension now. But other than that, another thing
that pretty much just worked. Again, it looks pretty subtle,
the change here is really that when you look at
all of the rooms, each different area, the walls are actually
blocking light, imagine that. They look different colors, it just makes everything feel
a bit more solid, it is maybe slightly
more obvious in this shot. It just made everything look
a lot nicer. We did start thinking a little
bit about performance here. Not that we were
too worried about it, but we had never thought
about it at all before because obviously, the sort of lights
we were using. Now we were a bit more cautious
with it, only lights that are going
to cast interesting shadows are set up to be movable. A little hack we had to undo
from our war torch earlier, is that to simulate flickering, before it did
physically move around, which obviously is not
a good idea for performance. We locked that down and instead now
just that the radius changes, and I did not even notice
the difference. Oh yes, I forgot,
there is a good picture of that. Speaking of performance, I will talk a little bit
about it only because people do ask me
about it occasionally, and I think perhaps
they need to recalibrate from what their expectations
are from AAA, because there is just nothing
to worry about here. In terms of our materials,
very credibly basic. That is not because the artists
are not capable of making weird
and wonderful materials. It is that, you know,
our general thought processes do not do anymore there
then you need to, the pixel art should
speak for itself. That is probably one of the
biggest materials that we have. We have a handful of those, a handful of instances,
that is it. In terms of the normal maps
for the tiles, two 2K textures covers
the whole world, so it is nothing really. As discussed,
lighting is minimal, no more than 5.10 in a scene,
everything is neatly spaced out. In terms of the whole world, again by actual modern game
standards, it is tiny. We changed the Unreal units
to be one pixel for obvious reasons. The world is about 30,000 pixels
wide by about 12,000 tall. That is… I wrote it here
because I knew I would forget, that is about 300 meters
by 120 meters, nothing. It is microscopic.
Honestly, our only performance concerns
were much more related to AI, because we are leaving out
a lot of the existing sort of quite
Advanced AI systems for Unreal, but that is obviously
a talk for another day. In terms of graphics, there are
no concerns there at all. In terms of starting in Paper2D, there is not too much
to worry about, but these are just genuinely
things I wish I had known because we spent more time
than we needed to. The first one is something
that came up from day one and actually in our first
ever picture of this game, we left this in,
but you end up with this weird tearing
because by default the UV Padding
is not turned on for the tiles. Now, it is an obvious thing
to fix, but not when you do not know
that in Unreal it is called Border Margin and for some reason in Paper2D
only it is turned off. Two seconds to fix,
but easy when you know how. Another thing is that if you are
going to want animation State Machines or notifies,
which you probably are, you do not get any support
for that out of the box. Our only solution to it was
a horrendous hack which I would never tell anyone how it worked,
but it was not good. Luckily, someone solved
that problem for us. There is a studio called
Critical Failure Studio, do not let the name fool you,
it is actually great. This will give you
the auto-notifies, this will give you
the State Machine, it is great. If you think you are
going to need those, do you go and buy it.
Sadly, I am not on commission, because I think I would have
made a fair amount by now. Right, and oh yes,
if you really want to fight the orthographic camera,
little tip. By default,
in many ways in Unreal, the orthographic camera
is not considered to move. If you have started to build
a large world like we are and you are moving very far away
from the origin, you will find that the lights
just start fading out even though the power is
actually the same in the Editor. That is just
a level of detail thing because it thinks they are
so far away, they are irrelevant. If you pop into
the Project Settings you can just change
the Min Screen Radius to 0.0, essentially disabling that, and for a game of our scale
it was never a problem. But seriously though, just do
not use the orthographic camera. The games were so small versus
what we lost it is not worth it. To sum everything up,
really the focus should just be making something
that is authentic. It does not matter
what that means to you, it is not going to mean
the same to everything, but you need to know
what it means to you and be able to align that
with your chosen influences. Inevitably, once you have done
that, you will discover that most of
the shortcuts you thought you were going to take to
get there are not going to work. Resist the urge
to compromise on those. Whenever people ask me
about the game, it is specifically about the way
it looks and how we did it. If I read between the lines,
I think really, they’re looking for this menu. They’re looking for a button
they can just flip, to tighten the graphics
to make it look good, it does not exist. You are just not going
to get it. As a photographer,
it reminds me of the situations that are so frustrating. When I take a beautiful picture
and I am very happy with it. Someone sees it and they go,
“Yeah, that is a great picture, you must have a really
nice camera.” I have to strongly resist
the urge to punch them, because it does not matter
what you use to take it, the tool really doesn’t matter. It is how you put it
all together. Saying that though,
why not use Unreal Engine form. I heard it is pretty good
at that sort of thing. In all seriousness though,
it still blows my mind that people just do not
seem to think it is possible. To the degree where,
a couple of years ago, we had this game on Epic stand at GDC below the world’s
biggest Unreal logo of all time. People would play the game
and they would look at me and they would say,
“Yeah, that looks nice. What engine is it made in?”
I would just step to the side, and they would realize
what had happened and there is no reason not to. There is nothing
that is missing from here. I hope that I have shown you
that really this game, it is just a combination
of looking at classic games, appreciating those aesthetics, and working how you can
sympathetically use them on techniques to enhance those. In other words,
it is mostly hard work but that is a good thing.
Because anyone can work hard, and anyone can use
the exact same tools that we use in Unreal Action
to make that. With that,
I would like to point out that you can watch me do this
on a regular basis and struggle
with them on Twitch. Also, if you have any other
questions, please do consult our website because we regularly
make blog posts about these sort of things.
With that, thank you very much. [Applause] ♫ Unreal logo music ♫

4 Replies to “Beyond Authentic Pixel Art in Unreal Engine | Unreal Fest Europe 2019 | Unreal Engine”

  1. why would one use unreal for 2D when Paper2D has no focus, no further development, and there are Godot and Unity? I mean I love Unreal, but in 2D its just way behind in every aspect

  2. I hope Epic one day get in a negotiation with the guys who made PaperZD so they can provide updates to all the 2D aspect of the engine

  3. which wheel was suppose to be the good wheel? I liked the on the left because it reminded me of PS1 graphics.

    and, doesn't using a perspective camera just invalidate the entire point of using tilemaps and a grid? don't you end up with different pixel sizes at the edges of the screen than at the center?

    and, and, I'm using grayscale sprites with a color lookup material too. you said it was tough work and had to be done by hand. why? I do all my sprites in normal color first, then I run a script on the folder and gimp converts everything exactly as I want it in seconds.

Leave a Reply

Your email address will not be published. Required fields are marked *