Transcripts
1. Introduction: When it comes to realism, there are really
only two main things that you need to worry about. Lighting and texturing
for lighting, there is a separate miniseries that you can have a look at, but here we'll see how
real life textures just propel your renders
to absolute realism. When you talk about models
with real life textures, there is no better
way to achieve them, but by using photogrammetry. We'll make the scorpion APC
using photogrammetry and also see how to correct the nuances that arise
during the process. During scanning, you don't
always have ideal conditions. This three part series
is all about making the best possible assets in
less than ideal conditions. We'll talk about how we can
quickly gather the data for photo scanning using video from your phone in a
matter of minutes, Process them to be ready
for photo scanning. And importantly, how to process these extremely
dense measures in blender to make them ready
for you in your scenes. Stick around and you'll get a free high quality
three D model as well that I've
made for you using the same technique we'll
be using reality capture and this technique is
gorilla photogrammetry.
2. Mesh Generation - Reality Capture: All right. We'll use our
phone to shoot our video in overcast conditions
at four K 30 FPS, which most devices these
days are capable of. We will move slowly around the subject to
reduce motion blur, and capture every part
of it that we can. This is the Scorpion APC. We'll move around it in
an onion peel fashion, which means we move
around the object in the first pass and then moving subsequently closer
with next passes, making sure to get closer to individual parts of the object. We can see that we have
a real life limitation here that we cannot reach
the top of the object. And this becomes
problematic when you do not have a texture
data to project. Later in the process,
we will see how we mitigate this using
texture painting during the processing
part in Blender. We'll take three round passes and this will get
the job done mostly. We'll bring this file
into after effects, Speed it up so that we have around 600 frames for an
object the size of a vehicle. About 600 frames. Ensure that we have adequate overlap
between those frames. This will help aligning the images in the
reality capture. In the next step of the process, we will bring up the shadows and decrease the
highlights so that the image is flat and
lighting is neutralized. You can shoot in log
footage as well, which will give you some more
wiggle room to play with. We will then export it
out as a Jpeg sequence. We've now converted the video
into frames 617 in total. We'll select them
all and then we'll put them into reality capture. There we go. The first step
is to align the images. Setting the image
down scale factors one and click a line. We wait and we wait. All the images have
aligned and we've got ourselves a sweet point cloud. This is so fascinating for me when all of it comes
together quite nicely. Fairly early on in the process, we'll now examine it to see how the point cloud
has come together and see if anything is lacking. We'll define the
ground plane now to align it with one of the axes, Now we put two for the top
view and then we'll rotate it, align it with one of the axes. This will help
define the area for mesh generation, which
is the next step. We'll see how it looks
from different angles and then go on to set
reconstruction region. We'll clear the current region and we'll define a new one. Set the rectangular boundaries as close to our
subject as possible. There we go. Then
just inspect it in perspective view to see how everything binds together
at just the height. See how it's looking. Top view, side view, make minor adjustments
if needed, Backside again, some
minor adjustments, and we'll make the bottom
margin a bit tighter as well. In the photogrammetry process, inspection at every step
of the way is paramount. So you'll see us
inspecting our model from time to time to
see how we're doing. At that particular step, everything seems
to be on point and we can go to the next
step of the process, which is generation of the
mesh from this defined area. For mesh generation, an
image down scale factor of two is a good
balance between quality and management and
reconstruction in normal detail is how
you should be doing it. We'll wait for it to calculate
the mesh for us perfect. So now the mesh is calculated inside the region
that we defined. So we'll just go over and
see how the mesh is looking. Just observe it from
every direction. See what parts have been
reconstructed well, what parts are lacking, how the mesh is looking. See if any nooks
and crannies have been missed out, any
holes in the mesh, any part that didn't
generate as you intended it to in
objects like these, oftentimes it's the underside of the model that is
undergenerated because we mostly cannot physically go underneath it and capture
the necessary data. We'll see how we address that in the blended processing
part later on. But now we will get rid of the extra geometry by selecting
it and filtering it out. So using the rectangular selection tool as
the first pass, crudely go over all
of the geometry to broadly select all
of the extra bits. And then we'll refine it later on by the lesser tool which will be a bit more precise in micro adjustments
in the selection. Pressing control, while you
select the new geometry, keeps the previous
selection while enabling you to
select new parts. By pressing control
and dragging, we select all the extra
bits that we can. This part of the
process is important because we do not
want to be projecting the textures onto areas that we just simply won't
be using later on. So it's a good practice to
save ourselves from unneeded. Unnecessary texture
and mesh information. And it will also save you
memory down the line. So we'll just keep
selecting crudely as much as we can inspecting
the model as we go, which is always a good practice, and see if you're not
leaving any areas behind. Just selecting as much as we can using the rectangle tool, Go to the other side,
select that as well. Now generally the selection
process is fast and simple. But because this is
a track vehicle, we will need to spend a
tiny bit of extra time. I switch from rectangle to the lasso tool quite
early on here. Let's see how far we can just take the rectangle
selection tool. Another tip is to double click
by your left mouse button while you're selecting and that moves the rotation
pivot of the model. This will enable you
to rotate around any specific areas of the
mesh that you intend to. We'll go underneath
and see if we can select some pass
from down there. It's always great to have as little as possible to
clean up in blender. Reality capture handles these extremely high poly
measures really well, but if you do it in blender, it will bog you down a bit. So ideally, clean up as much as possible In
reality capture. Now our first selection pass is complete and we'll
do the filtering. Now it's crudely gotten rid
of that extra geometry, but it still needs that micro
cleanup Before texturing, we'll select the
lasso tool and then we'll start the micro
cleaning up process. We're setting up the
rotation pivot by double clicking our left mouse
button like I said earlier. And now we'll trace
out the object as close as we possibly can. This process does leave a bit of jaggedness and jagged edges and we will fix those when
we go into Blender. This is a bit of tricky
selection because we're making our way through the
teeth of these tracks. This lasso selection tool is extremely good in
situations like these. Blue triangles that
you see underneath are non manifold geometry which have risen as a result of
our previous pass. They are not a part
of our mesh and they do not form a closed mesh
on their own as well. We would want to get rid
of them sooner or later. We can see that some
of them are being selected because they are
coming in our way of selection. Let's quickly select this, making sure that we
select as many of these straight
triangles as possible. We'll go underneath
the track once more. Thankfully, not in person selecting as closely
as possible. Once again, remember
to keep pressing control on your keyboard
while you select it. We'll go as far back as we can in this view while
tracing the model. And then come back to
complete the selection. Let's select this left away
island here. There we go. We'll go to the other
side to see that we're not taking any part of the
mess that we want to keep. That looks good. So now we can
move on to the other side. Let's trace this break. Now I can understand that
this can be a bit boring. Believe me, the time
spent here will save you hassle later
on in the process. So take your sweet time, put on some music and
just get on with it. The isolation of the ground touching the side
of the tracks is easier relatively
because we have a defining margin between
the tracks and the ground, so we will just
easily trace that. It's helpful to follow
these natural lines in order to prevent a sharp cutoff when you develop the model. Later on you'll notice
that while selecting, I'm constantly changing
my rotation pivot point. This ensures that we are only rotating around the
point of our interest. Inspecting or
selection as we go. And moving on to the inner side to clean up what we
can at this stage. On the other hand, what
we can't clean up, we'll just use billions
in blender to do our job. Now we go on to
the back side and selecting this ill defined
topology as much as we can, This is pretty much
what we can do at this point without
wasting any more time. We will go on to the next step. Just give it a quick lo and we can go on to filter a selection. It will take a
second or 22 process and now we have our mostly
clean up model here. Let's see how we're doing
in terms of topology. Give it some time
and it shows that we have a few defects
in our topology. Let's now fix that
these topology issues arise as a result of
our clean up process. But fortunately, reality capture has this important and
very handy feature where it can detect bad topology for the most part and
then try and fix it. The software tries to get rid
of non manifold geometry, fix any inconsistencies
with the geometry. It's a good automatic first pass to help you down the line. Now the model that we have
is 5.3 million tries, which is arguably quite large. So we'll try and simplify
it to around 500,000 tries, which is a bit more manageable. If you have a low end machine, you can use this 500 K model. I would recommend using
about 5 million tries, because the texture that
will get projected on it will be much more crisp
and better quality. Later we'll go over
how to transfer that high quality texture onto a model version with just 20 K tries out of the 5.3 million
that we've made here. In the next part, we'll
actually be working with the 5.3 million tries model. However, before we do that, we'll need to unwrap the model for the texturing
part of the process. The unwrapping
looks fairly dense, which is always great to see. And then we'll click Texturing. Now the textures
have been projected based on the UV
unwrapping that we did. We'll again quickly scan
over the model and see where are the discrepancies between the texture
and the mesh. It looks fairly fine
except the parts where it didn't have the data to
generate that texture data, we couldn't capture the top of the vehicle as we had
real world constraints. And this is what this
series is all about, making lemonade out of lemons. After texturing,
just move around your model to see what
areas of the model would need to be worked on
just to get an idea of where you'll need
to focus in the post. In the next part,
we'll head over to blender and make it
into a usable asset. If you've got any
questions or suggestions, be sure to let me know and
we'll sort it out for you. Until then, farewell.
3. Mesh Cleanup - Blender: In the last part, we exported our model from reality capture. In this part, we'll
see how we'll treat this model within blender to
make it production ready. We've imported the
model here in blender. Let's see how it
generally looks. See if there's any straight, we see any straight
geometry around it. We'll see the model in rendered
view as well and see if there's any discrepancies
in texture geometry, what things we need to work on, and just a general overview
of how it looks like. We'll go into solid
mode again and we'll rename this to UHP, which is for ultra high poly. Let's turn on the statistics to see the number of polys
that we're dealing with. So 5.3 million roughly. Let's now move a model to
more centralized position. Rotated to align with
the views of blender. So the front of vehicle to
the front view in blender. And then move it to a
centralized position again from the side view
as well as the front view. Nudge it down a little
to that ground level, see if everything looks fine. And then we'll set the
three cursor as the origin. This ensures that it
will scale and rotate from the ground level where
this vehicle will be sitting. We've started adding
a cube underneath the vehicle where we
couldn't reach earlier. And the geometry was not constructed in a way
that we wanted to. So we'll knock out this
degenerator geometry. The way we knock out this
geometry is by using billions. So this box that we're making is essentially 1 billion cutter. We'll scale that box
in a way that it conforms the geometry
that we do not want. But at the same time, be careful that it does not include the geometry
that we want to keep. So we do want to keep the
inside teeth of the tracks. So we'll make sure
that the box does not cover them before executing
the Boolean function, we see that the box covers
what we wanted to cover, select the box and
vehicle as active. We press control minus
this cuts out anything that's within the box and we did not need before applying. We see the mesh once
more and we click Apply. The modifier is now applied and this box has done its job, so we don't need this anymore, so we'll just delete this. We keep on inspecting the
model every step of the way. Now the second step
is to get rid of the geometry that is not
attached to the model. We go to edit mode,
deselect everything, press L over the model
to select the bulk of fit and see what parts
have not been selected. These are the islands of
geometry that we do not need. Once that is done, we press control I to invert
the selection. And let's just also select the stray vertex that's left
by the Boolean operation. We'll press X,
delete the vertices. We'll see what more extra
geometry we need to delete geometry that was
left attached to the model. The geometry that
did not come within the Boolean operation or within
the invertletion method. With a little more inspection, we find out that
these pass underneath do not form a part of the
model and are not necessary. These were left behind by the reconstruction bounding box. We'll just take our time
and select all of them. There we go, and
we'll delete them. The other side does not have straight triangles like
these. That should be fine. Let's inspect our
model once again, shifting into object mode, see if the model is ready for the next
stage of the process. Everything seems to be in order, so we'll go to the next
stage, which is sculpting. In the sculpting process, we will get rid of these
coarse surface features, which are not originally
part of the object, but they arise as artifacts from the photogrammetry process. A good question is why should we get rid of these
imperfections, especially if the model looks fine when we saw it
and rendered view. Well, the answer is when
we are baking normal map, as subsequent part
of the process, this artificial bumpiness will become a part of the normal map. Because these are artifacts, we do not want these to become
a part of our normal map. However, we need to be careful to not over smooth and overdo this because we then run into the risk of losing
the mesh detail. Just as we did here. We'll switch to
the flatten brush, which will give us
some of the form back. Especially when you're
dealing with flat surfaces. I mostly tend to switch between smooth brush and
flattened brush. Smooth brush for meshes which
have more organic curves. And flat brush when
the models are more faceted and hard surface. An important difference
that you should and must know when using flat and smooth brush is the nature of how these
two affect the textures. When you use the smooth brush, it sort of warps the
textures because it's averaging the height of the mesh detail within
its circle of influence. The flattened brush, however, does not cause any
warping of textures. Because it just flattens out
everything in one plane. I therefore primarily use and
like the flattened brush, but if something's not
working out the way it should or I don't
want it to be faceted, then I'll tend to use
the smooth brush. Another thing that
the smooth brush does is that it shrinks
everything together, sort of slimming them. In a way, you can avoid the slimming by using
flattened brush. Or if you have to use
the smooth brush, you can switch to inflate brush to make up
for that shrinking. You'll also notice that
while I'm using the brushes, I'm not using them at
100% of their strength. I'm using them at about 35% 45% depending on the influence
that I want from them. This part will be
fairly sculpting heavy. So if you don't enjoy cause part of the meshes
being smoothified, you can skip to
the last third of the video and carry
on from there. In the last third of the video, we will be baking
texture maps from this ultra hy poly
to a high poly mesh. For those who want to stay, we will be talking
through the process and our thought process behind
whatever we are doing. So a good way to conduct this smoothing
process in sculpting, it's good to visually select one part of
the object and then make your way from there successively to the
adjoining areas. Currently, we're flattening
out the back side of the vehicle Using
the flattened brush, I generally do a course pass over the intended
area and then just zoom into smaller areas which need a bit more
finesse with the hand. So this bottom back
plate is done and we move on to its
top adjacent area. Switching from smoothing
to flattened brush, we move carefully
in order to not flatten the areas that
should not be flattened, like this curved
ridge right there. So in this way, we're making
the model more defined, keeping the ridges and eliminating the
blobby elevated bits. You don't need to be very, very precise with it. This process can be
very forgiving so you don't have to break
your backs over it. Another advantage of this
flattening process is that at a later stage when you
are decimating the mesh, the resultant mesh that comes
after decimation will have less polls in these
flattened areas than if we had not
flattened the surfaces. So the decimate modifier assigns more vertices and
more geometry to the areas of the
model which are more curved in order to
preserve that curvature. But the areas that
are flat will not get that much geometry
because they don't need that much geometry in order
to hold themselves together. If you do not have
these areas flattened, which are supposed to be flat, the subtle changes
in height will be picked up as curvature by
the decimate modifier, and these areas will end up
gathering more geometry. Ultimately, the
geometry distribution does not end up where it should. This gives us a low quality
remesh after decimation, which we obviously do not want, just taking care
of the back here. You'll see that
while we flatten, we do get rid of
some of that detail. But that's not worrisome because we will get that detail back from the texture and the normal map that
we generate from it. We keep flattening the facets, keeping in mind the plane, if they lie, flatten
the bit as well. And then going up top, we will take care to
not dissolve these X shaped ridges
because they do give a fair bit of detail
to the back side. We keep avoiding those ridges and flattening out the
areas between them. If you're not good
at sculpting or even if you've not tried
it yet in blender, please don't be intimidated
by the process. It's straightforward.
In this process, we use very limited
amount of brushes, so you don't really need
to know a ton about everything in sculpting except a few brushes that we're using. Anyway, like at every
step in photogrammetry, we are moving quite a
lot around the model to see how the model is looking
at that particular stage. We continue doing that. I want to drill this down
in you because it is very important in every step we are done at the
back of the model. So we'll move on to
the side over here, we'll align to the
plane and we'll start flattening and
flatten a bit more. Once we're done with local area, just give it a little
look how it's looking. Then this little area, we can flatten this out as well. After that is done, we will
compare it to the other side, like go back to the other side and start
working on these tracks. Now importantly, we'll use the
smooth brush to smooth out these wheels because we
don't want the curvature of these wheels to get
flattened by the flat brush. You see the raised
and depressed area in the central
part of the wheel. We do not want to flatten
that in the sprocket. Same thing again, and now when we're doing the outer
rim of the wheels, we do need them to be flat. Switch to the flattened brush, Decrease the size carefully outlining that circumferential
area of each wheel, so as to give it a little
bit more definition. We've adjusted the strength to 50% because we only want
to give it one pass over the area in order
to not introduce any more facets while we're
using the flattened brush. In these uncommon situations, you do need to be relatively
more careful while you're tracing out the topology because there is a
chance of messing it up, especially with circular
shapes like these when treated with these
sculpting brushes. So just about 02:00 position
on this last wheel here, we see that there
is a dent and we do seem to be struggling
to try and get that out. The flattened brush flattens
out the topology based on the viewing angle that
you're viewing the model from. You should align the
viewing angle by the plane of the surface
that you are flattening. We've gone out of norm here
to shift our viewing angle, to fix that one key
bit of topology and it does seem to do
relatively fair job. This Brocket is relatively
straightforward, just flattening out
between the pins here. Here, just around here. Then we'll just
complete it all around. Let's examine where
we are going next. The topology
underneath the tracks which was gotten rid of by filter selection in
reality capture is a bit jagged and we left
to fix it in blender. Now is the time to do it. We use the smooth brush to just smoothify all of these
saw shaped jagged edges. You don't have to
be precise because these will be touching
the ground anyway. But just for the sake of
clarity and cleanliness, we'll just quickly
take care of them. This moving out of word seas will also help if you want to select all these bordering
words and give them a face, but we won't give
it a face here. So we'll just die if there are
any uneven, unkempt edges. And we'll shift on
to the other side, we'll pick the front
of the vehicle as our next side of action. While we are flattening
this side of the mud guard, we see that the flattened
brush is creating an artifact. So it is a good time
and a good example when we switch from flattened
brush to smooth brush. So we'll do that and
that takes care of it. Same brush at the front
with 20% strength so that textures don't get warped up and we'll just cover
this front hole quickly. No rocket science here go underneath because oftentimes
we tend to miss that. And then we'll cover this
little island here as well. You can see that with 20%
strength of smooth brush, we are retaining some
of that bumpy detail. And that will become a part of the normal map that
we bake later on. Some level of
bumpiness is desirable to keep the detail in
the mesh level as well. In effect, normal map drives
the micro details and this macro bumpiness drives
the macro non uniformities. If some part of the mesh
is particularly malformed, the point of no
rescue and you know, you can't salvage it
by mesh manipulation. It's easier to just get
rid of it outright. So delete that part
of the mesh by using Booleans and
then you can model that part yourself and place it exactly in the position of
the part that was deleted. Join the newly modeled part with the original mesh by selecting both and
pressing control J. Then when you bake the textures
in the lower poly mesh, your original textures will be baked onto the new
geometry that you made. We won't be needing that
technique here because we've obtained a
fairly fine geometry, but that's a neat trick
to have up your sleeve. This part here has a texture
of straight grill lines. We'll be careful not to use
the smooth brush to prevent those straight lines from curving as we go
further and further up, we have less and
less texture detail. Because of that real
life limitation that we had during the
capturing process, we're not too concerned about not using the
smooth brush here. But flattened brush still
gives more defined edges. We'll still continue to use that flattening out
this top part a bit, keeping into account all the
ridges that we encounter. Preserving those and then
nicely moving on to this front. You can see how the flat brush defines these edges
quite nicely and gives back that angular nature of the transition
between the two facets. Taking a step back, having an overview and
taking care of those tracks. On the right side, well, left side of the vehicle, but right side for
simplicity, forgive me. There we'll continue
working our way up. Remember when I told you that smooth brush has the effect
of shrinking the mesh? So we'll be gentle when we use the smooth brush here on
the front of the barrel, since this is a flat area and bumps on it do not
really make any sense. So we'll just pick
the flat brush. And just flatten
it out completely. We switch to the
smooth brush again, to just go over the
rest of the barrel, especially the
underneath and the back side of it gently
covering what's left, making sure we do not become
too aggressive with it. We already have the power set at about 20% Slowly but
surely we see that the model is transforming
from being bumpy and lumpy to a more planar,
well defined model. As we go through
this sculpting pass, the edges of the model
will be more defined and this will help when we light the model
up for rendering. The light will nicely
blend in around the edges, giving great looking
edge highlights. If we put the model in retraced engine without
this sculpting pass, these bumps will cast
some amount of shadows and this will affect the
texture integrity of a model. The extent of sculpting and treatment can vary
depending on how clean the data was initially that was provided
to reality scan. But here I deliberately selected this model
to teach you how to optimize suboptimal measures and still get great results with it. So we know we could not reach this top part of the object
when we captured the video. Inherently, this part of the
mesh is relatively blobbier. But just by using smooth
and flattened brush, it nicely defines
the geometry and reinforces the creases
between the facets. While we are using an
alternating between the same tools that we have used to fix the other
parts of the mesh. I want to take out some
time and speak to you in a broader perspective
of why even bother with collecting
these photo scans. You see you should not
be fixated on creating a single model or a
scene or some animation. What you should be striving
for is creating systems, creating an animated
or three D scene. A set repository or a library is essential to increase the
productivity of your workflow. Every major studio has their own asset library
in one way or the other, whether it's made in house or it's bought from
somewhere online. So in your everyday life, whenever you come
across an object that particularly
catches your attention, it can be a hard surface object, it can be a nature object, it can be an everyday
object in your home. Anything that makes you go, hmm, I should make this a part
of my three D scene. Take out your phone, scan it, and make it a part of
your asset library. Over time, these assets will accumulate and
before you know it, you will have an
entire repository that you can use as drag
and drop objects. And you'll have your very
own personal asset library. And believe me, there's
something about scanning your own assets that's so satisfying, it's
almost addicting. Do see your scanned
assets within your three D scenes that you have captured from real life. It's just a joy
at another level. This way you can quickly create your three D scenes by just dumping the models
that you've scanned, throwing in an HGRI, Add any additional lights, if you may set the
composition of it and you can call it today. However, be mindful
that your job is to be a grand collector and not a
grand holder. The difference. Well, collectors collect only the things that
they absolutely, really, really like orders. On the other side, they'll
just put their hand on anything and everything you want to make your asset
library a high quality place to get
your assets from, not a garbage bin. Photo scanning will also
give you the opportunity to step away from your screens,
go out in the open, in the nature in the museums, wherever you want to take a time out or if you're
feeling like it, take your phone out
and start scanning. It only takes two or 3 minutes. If you do the scanning with the video method and it
yields phenomenal results. You'll start to casually
scan whenever you go out, but then you'll find yourself
taking time out, going out, especially to Photoscan,
that one object, when all, if it comes together, I promise you,
it's all worth it. Moving over to the other side, you notice that we did not go over very aggressively
over the grill part of the object because we do want to preserve this
surface information. Had we used that
flattened brush, it would have just wiped out
that grill surface detail. It's like we treated the
wheels on the other side. We'll treat the wheels here
in the same way as well. Using the smooth brush to smoothen out these curved
parts of the wheels, smoothing out the
area between them. Just gently going
over this Procket as well and then switching to flattened brush like
we did previously to just flatten out the
circumferential area of the wheels. We're just giving one pass with the flattened brush over
the circumferential area. But if you feel that your
mesh is particularly blobby, you can use multiple passes of the flattened brush to
optimize that area. I do recommend to not use
the flattened brush at 100% strength because that
gives an all or non approach. If you have the strength
to about 22:40 percent, you'll be able to preserve
some surface detail. If needed, you can have a second pass which
will add on to that 20 or 40% Now these
left tracks are done, we'll keep eye baling, we identify This break
needs a bit of smoothing. So we do that then
using the smooth brush, we quickly fix the jacket
edges underneath the tracks, giving it a quick, smooth
pass on the outside. There we go, we move
on to the inside. Just quickly going
over that area because it won't be much apparent
in the final renders. Now we eyeball the model again, quickly inspect what
areas need some revision. This is a good time to look at the deeply seated areas of the model which are
likely to be overlooked. So we identify this
deeply seated area, which is the transition
from the body of the model to the barrel. These areas can be
relatively more undergenerated
because if you're not careful during the
capturing process, you might end up
capturing less than ideal information in
these deep areas. A good way to mitigate this
undergeneration is during the capture process to capture enough data for the
software to process. You can adep this
by moving closer to these areas and
honing in on them. During the capturing process, be sure to capture some
amount of parallax so that the software has enough data
to generate a viable mesh. At this point, we're
pretty much done with the sculpting
stage of the process. Before moving on
to the next step, we'll quickly see if
everything looks fine. The next stage is
to set up the nodes for texture maps, but
before we do that, we'll switch to
render view to see how the textures are
looking at this stage, see if there's any warping
introduced by sculpting and use of the smooth
brush all looks fine. And now we'll start to set up the nodes for
our texture maps. We'll zoom into the model in an area where we can see
the texture details. So let's select this front part, over to the window
on the right side, we press shift F34 times to
go to the shader editor. Over in the shader editor, let's remove these
prepopulated nodes that the shader already has, accept the base color. We'll use this base color
to make a normal map. First, we'll do this
by using a bump node. We'll introduce the bump node, plug the base color
into the height, and plug the normal into normal. We'll keep the strength
at 100% and use the distance as the driving
factor for the bump. 0.35 is a good
number to start with and you can dial it down
according to your model. We'll change the
string to see how it looks with and without bump. And to be honest, bump gives much more detail than you would get otherwise with
just the texture. Next we'll move on to
make the roughness map. Let's introduce a color ramp. Plug the base color
to the color ramp. Invert the colors and
color to roughness. Now if we switch to the
output of the color ramp, we'll see this distribution. The areas in black
will be shiny and the areas in white will
be absolutely rough. At this point, we'll dial the roughness in
the shiny areas. We do this by increasing the black amount to a bit
closer to being white. It still has some gradient, but it's generally rough. We can also bake
in a metallic map, but since this model will
be 100% metallic anyway, so we can just pull this to
100% in the principal shader. Seeing how this model
is looking from every direction with
this simple node set up. We can see here that bump node is doing its
magic really well. We can see that on top of
the model where we did not have access to
shoot the video Earth, we do not have good
texture detail. We'll fix that in the next part. After baking, after inspecting
with the node set up, the model is a bit contrasty. To begin with, we'll introduce the brightness contrast node
to decrease the contrast. One stop looks well for baking. All sit to solid mode. Duplicate the mesh by shift D, right clicking to release the duplicated mesh to the same position as
the original one. Rename it to high poly hide the ultra high poly
mesh from the viewport. For now, we'll put down a
decimate modifier on it, aiming for about 100 to 150 K tries from the
original 5.3 million. If you divide 150,000
by 5.3 million, the answer comes at about 0.03 Setting this
will now apply, we immediately see that a lot of microsurface detail
has been lost. However, the triangle count
has come down significantly. We'll get those surface
details back by normal map. Now, because this decimated
model has new topology, we will unwrap it for
better texture baking in edit mode and smart UV project. And now we have an
automatic unwrap. Arguably, the packing of the
UV's can be a bit better, but it does work fine as it is inspecting the model one last time before we commit to baking. And now let's give it a new
material in which we can plug in the texture maps that will be generated after baking. Call it Scorpion High
poly Now will unhide or ultra high polymesh make sure it lays on top of
or high polymesh. And now for baking, we'll use an add on called Simple bake. For baking, I cannot recommend
this add on highly enough. It generates all the
texture maps in one click without having you to set up everything
on your own. We'll select PBR bake
in the settings, I've made a preset
four photogrammetry, but you don't need to do that. You can just follow
along as I'm doing. Select the ultra high
poly as the bake object. With bake selected objects
to target objects checked. Select the target
object as high poly. We're baking diffuse
roughness and normal map. So make sure these are
checked and everything set up at the back
end, which it is. We don't need any other passes, so we'll skim over that. In terms of texture resolution, we're baking at four K, so 4096 by 4096. And we'll set the export
path as a subfolder called texture within the
parent folder for organization. This will be in the sub
folder of High poly four K, set the calculations to
happen in the foreground, so it uses all the resources
that we currently have. And then we click Bake. It takes some time and our
texture maps are baked time to import them and
place them within our high poly shader to import all of the
texture maps at once. Press control, shift, select all of the
textures. It's okay. And all of the textures
will be imported, plugged in into the shader
to see how it's looking. We'll switch to Matt Preview. It looks a bit washed
off because we do not have metallic
turned all the way up. When we do that we have all of our textures with the details
of the ultra high polymesh. We can see that the
normal map has preserved all of those micro
texture details that were within the ultra
high poly mesh and projected it onto the
lower poly version. In the Matt preview,
the bump looks exaggerated in the
EV so we'll switch to cycles to see how the textures really look
in the natural light. Seeing from all angles, it looks like it's coming
really well together. Fantastic. In the
next and final part, we'll see how we fix the blurry and inadequate
textures on top of the vehicle. I will introduce you to some
more tricks in the process, so stick around and
enjoy the process. Until then, farewell.
4. Texture Optimisation - Blender: Welcome back. In the last part, we worked on optimizing
the model and we baked texture maps from ultra high
poly to a high poly model. In this part, we're
going to repair those textures in the
areas where textures were not generated in the
first place because of the restricted access that we had while we shot the
video of the object. We'll then save those
new texture maps and bake them onto a low
poly model so that the low poly model has almost text detail as
the ultra high poly one. Stay with me till
the end and I'm sure you'll pick up a few
tricks here and there. Let's begin. This is
where we left off. We'll do some organization
before we begin. We'll delete this ultra high
poly as we don't need it. We'll delete this cutter
collection as well. And this default collection
is not needed as well. So nice and simple,
justrahighpolymdel. So we'll switch to Matt Preview
and go to Texture Paint. This workspace is where we'll
do the bulk of our work. Let's switch to flat
lighting because we only need the diffuse map without
lighting affecting it. Let's now identify some unwell
areas in terms of texture. This back part needs some doing. The side looks all right and particularly the top
needs a lot of work in order to make it
closer to the rest of the texture quality from
the panel on the left side. We'll use the fourth tool, which is the clone stamp tool, as our main driver
of texture painting. This will paint the areas
by sampling the part of the texture where the
three D cursor is placed at. We press left click to
designate the area that will be sampling and then we start painting. It's
simple as that. At least on the surface, we'll keep changing
the position of our three D cursor so
that we're sampling the part of the
texture that lies in the vicinity of the area
that's being painted on. If you worked with a clone
stamp tool in Photoshop, this should be a
walk in the park for you but with a twist. Because we're
working in three D, there's one important thing that you need to keep in mind about the sampling site
while you are texture painting the clone
stamp tool clones, the area under the
three D cursor based on the perspective
that you're looking at. If the perspective that
you're looking at, the sample area is
skewed a little, the area that you paint on will inherit the same
skewed textures. It's something that
you might want sometimes but
sometimes you don't. An important thing
to keep in mind, that's the reason we'll
mostly be painting in orthographic views of top, bottom, left, right,
front and back. An extension of the same thing
is that you need to keep the area under the three D
cursor within your view, only then you'll be
able to sample it. Another thing that needs
mentioning is that the sample area needs to
be in the same model. Cloning textures work if you have your three decursor placed on a different object. This means that you cannot clone a different texture
on a different model, on top of another texture
on another model. If you don't get it, don't worry because we will see the
practicality of this shortly. Essentially, what we're
doing here is that we're copying
textures from areas which are crisp
and then painting those texture patterns to the areas which are not as
crisp as we would like. There are a few tips and tricks
that will go over during the painting process in order to achieve a
fairly good result. It also comes down
to skill practice. So you'll get better as
you get the hang of it. Like we did in the first
part, during sculpting, we chose one part of
the model and then we progressively moved to the
adjacent parts of the model. We'll do the same here. We've selected the back side of the model and we'll
work our way forward. In the first pass, we'll have the strength
of our brush to 100% and we'll place the
three decursor over the area, the texture of which closely resembles the area that
we're painting on. Now we keep moving the
three Dcursor for it to prevent looking homogeneous and for painting over the same area, we're using different
places as our samples. This will mix up the textures sourcing from different places and the texture will have a much more heterogenous
look to it. Now as I said earlier,
the sampling sources, the textures from the view angle that you're looking at it. And there's only so
much texture from the same viewing
angle that you can clone before it
starts distorting. And sometimes you want
to clone a texture from the other side of the model that you can't see
in the same view. The solution to this is that
you go to the edit mode, select the entire mesh, and duplicate the entire thing. Now you can rotate this
duplicate to any side. We'll put the right side up to source the textures
from that side, see if it aligns well, and see that the
texture that we're about to source is in the same Viewing angle, we'll go back to the object mode and
then you can place your three decursor
on top of the duplicate and source
the texture from it. This part will again need continuous examination
of the model to see where we can
source the textures from and what textures
will be appropriate. Where once the
texture is cloned, we'll see the general pattern of how the texture is
looking at that point. So we got rid of that
originally cloned line because it didn't make
sense at this point. So once the area has got
a general texture from a different area of the model
to refine and outline it, we can use the same
adjacent cloned texture like we're doing here. The slightly blurred outline
is good in the way that it will help us to blend
in the adjacent textures. The storage bin on
the APC should have the same texture as the
storage bin on the other side. So we'll source our texture from the other bin and we'll try
and match as best as we can, changing the source
to the middle part and then continuing our painting just like
Sir on the edges. You have to be a bit careful in order to not bleed
on the other side. We'll zoom in keeping the
cursor in our view and we'll carefully paint the
boundaries that is done. We examine the area and see
what other area needs doing. Adding some texture here. Now, this area is more
homogeneous than we needed to. First we'll define the
boundaries and then we'll sample a texture that
introduces some variation. Like so, we can always experiment with different places for sourcing the texture. And during this refining phase, we'll turn down the
strength a bit. We'll do that in order to mix the new textures
with the old ones, but when we're
painting a new area, we'll set the strength
back up to 100% Here we're trying to paint the top of the storage bin by
sampling the side of it. As we can see, we can't get a
convincing paint over that. We'll try different
sampling sites, but that does not seem to
be working in our favor. When that happens,
you can partially paint and then leave
it for that time, and we can come back to it at a later time when the
adjacent parts are painted, so that we can source
textures from there. So we've done somewhat
of a passable job here, but we'll move on
to the next area, which is the central
part of the turret. Let's inspect the model and identify the area where we
can get the texture from. For this part, we'll use the light brown back side
as the source of it. And for that we will move our duplicate mesh near the area that we're
going to paint on. So place our three
cursor there in the same view and
paint a part of it. We'll then move the cursor
to a different area. Use that as a source to
add in more texture. After we've cloned this texture, we can source a different
area for the crevices. We can try different
areas to be hetrogenous. You don't have to be perfect, but it should make a
little bit of sense about where you're painting
and what you're painting. This area at the front of the turret does not have
good textures as well, so we'll take care
of that as well. Sourcing from the
adjacent areas. Mixing and matching using
different powers of brushes. Cloning textures
from different parts of the mesh by duplicating it. These are some of things
that you will end up using quite a lot
in this workflow. You want to be sourcing
the that is not flat looking but has some sort
of texture detail in it. Here we're trying to make
a smooth transition from one color to the
other painting at low power as we approach
the meeting point and bringing the power to 50% in
order to mix these textures. These are extra features here that we don't
need at this place, so we'll get rid of
them using the texture on the front plate as
a source to paint on the top and sampling
nearby areas to get rid of these
dark unwanted areas. Mixing the textures
from surrounding areas and moving on
to the next part, again using the front
plate as the source, cloning the texture on top
of the turret and the side. Inspecting the model where
we need to paint next, the crevice between
the front plates needs a bit tidying up. We'll use the bottom plate as a sauce and we'll
paint the top one following roughly
the painted area black and the brown area
with the brown sauce. Tidying that up and leaving
behind some imperfections. This area shows some
grime deposition. So we'll leave it and we'll
just tidy up from the top the texture on the night site balance
weight is a bit blurry. So we'll sample the
texture around it and paint on top again,
trying combinations, mixing and matching until
it looks just all right, we'll save the work
on the texture that we've already done
by image and save. Very importantly, whenever
you work in texture painting, be sure to save your
textures periodically. Oftentimes it gets
overlooked and you've put in half an hour of work and it
gets lost and it's painful. Make a habit of pressing alt. S. Painting. Whatever you do, it gets saved as you do it. There is another crevice
between the grill and the body, which did not receive
any texture information because we could not get to it during our data acquisition. Now is a good time to address
the textures of this area. We can clone texture
from the front of the vehicle and we can
quickly go over it. We're not too
bothered with areas that are within
crevasses and within confines of between two
surfaces because these aren't areas that get seen
very often in the renders. Nonetheless, we should make
an effort to make the texture look as better as it can
be here in the front. These radiator lines
could be made better. We can clone from the bottom
part, which is straight, and we'll do the same on the
engine intake grill as well. Being a bit careful there. All done. You can also go a step further and paint using
an image as a stencil, but this would require having an accurate image of the
model from real life. But getting a suitable
image might get tricky, so we'll stick to our
trusty texture painting. Now we'll paint the front
right of the vehicle and we'll see where we can
source the texture from. Let's move the duplicate mesh aside to give us
some working space. There we go, and we'll identify a place where we
can source the texture from. Again, filling in
those crevices where textures are not
that good fidelity. For starters, the
surrounding areas are a good place to source
the textures from. If that doesn't work, we
can look for other places. You'll notice that we're
not spending a ton of time perfecting a
single area because this process is
all about churning good usable assets in the
shortest amount of time possible while tackling
the real life limitations that we can have during
the acquisition process. We'll inspect again which
areas need our attention. This front bit seems to be a bit off and we can't
seem to paint over it. We'll take care of it shortly. For now, let's paint
out the flag so that the vehicle can
be used universally. Again, using the surrounding
areas as source. Whenever you see that the
painting is a bit splotchy, you can again sample
different areas to mix it up. Here we will draw
a continuation of the bright streak
of paint that you see coming in from underneath. And on the left side, the darker paint can be
given a continuation. Once we've painted the area
from different regions, we'll zoom out a bit,
decrease the strength, and we'll give one or two
gentle strokes. There we go. Now let's address this
unpaintable area. And there are a few
triangular areas on the left side of
the model as well. If you go to edit mode and
select this part of the mesh, we can see that the mesh
is actually intersecting. When we move it out of the way, this solves the problem. On the left side of the model, the geometry is actually and we can see through
these triangular areas. What we can do is we will
create new geometry for this. We'll switch to vertex mode,
Select these vertices, press that gives it
a face and we'll do the same thing for the other big
triangular mesh defect. Select the vertices,
press give it a face. There are a couple straight
triangles here as well. We'll give them a phase two, then let's try and
texture paint on them. We switch to texture paint mode thredcursor
on the source, but we still can't
paint over those face. The reason is that
this new geometry does not have any UV's
for us to paint on. We'll make UVs for it. We'll identify and select
the newly created faces. We'll do this by going to the phase select mode
selecting one phase, shift selecting the second, and shift selecting the third. Press U projection. Now we have the UV's
of these triangles without affecting the UV's
of the rest of the mesh. We'll scale this down to match the texture resolution with the similar tries in the mesh. And then we'll move
them to an area in the UV space where the
textures correspond, similar textures that are present in the area
surrounding these triangles. We can see that the
textile density within this triangular area
is still too large. We can either scale
them further down, or we can find a different
area in the UV space, an area which is relatively blurry to match
the surroundings. But we resort to
scaling them down even further to match
the textile density. At this point, this
seems just about right. And we'll move them
to the side over the brown area so that they'll show up with
a brown texture. Rotating them a bit to match
the grain of the texture. Deselecting. If you don't look at it too closely, that looks all right. Alternatively, what you can
also do is that you can place those triangular
UV's in the black area, in the texture map, that's where there
are no other UV's. Once placed there, you'll then be able to paint
on top of them, just as we were doing before. Either way, this is to give you multiple ways of
troubleshooting your mesh and how to go about in
the face of problems related to the
mesh and textures. What we're essentially doing is we're sealing texture from one part of the mesh and
painting on the other part. And this reminds me of a lovely book that I
recommend you all read. It's called Steal
Like an Artist. It's a small book that tells
you in palatable chapters, how do you think creatively and sustainably in an
artistic mindset? One of my favorite ideas in the book is productive
procrastination. It's the idea that you're
procrastinating on something, but you're working on one
of your side projects. Something that you're
really passionate about, something you're
really interested in, and in our case, photo scanning, and that comes with
the added benefit of building your asset library. You can be working on multiple productive procrastination
projects at one time. So if you get bored
of one project, you can hop between the projects and in the end you'll
have your library. There's another idea
for artists who are struggling with
idea generation. So if you struggle to generate ideas for
your three D renders, know that every
idea is just a mash up of previously existing ideas. What this means is you can copy ideas from different renders. This is basically
what referencing is. Then you can add in
your own touch to it. The direct derivative of
this concept is that you should have an
Inspiration folder with you in your hard drive, whatever, three D renders, animations, models, you come across, screenshot,
it, download, whatever, keep it saved this way you'll have
a bank of references. What's good is that this bank is tailored to your own likings. And you'll have an
inspiration folder, whatever you came across and you particularly liked in the past. This will also help
develop your niche. In three D, you
will find yourself gradually gravitating towards
certain rendering styles, animations, certain sort of cinematic styles, color grades. And you'll want to replicate that look in your own renders. And that's how your niche slowly but surely
starts developing. You'll add your own touch
to it and lo and behold, the most original and
authentic render of its kind. And on top of it bank of
inspiration and references, you will be a massive
idea generation machine. Now coming back to our model, we're painting the front
part of the weakle, just behind the radiators. The texture does feel
a bit homogeneous, but remember this
is our first pass and we're just crudely populating the texture to cover
every blood texture area. Once these blurry
texture parts are covered with some sort of
detail in the texture, we can then move on to introduce some variation
in the texture. It can be with a
different color, can be with different
texture pattern, whatever. That just breaks the texture up. In this instance, we're adding a different color that
breaks the texture up at some variation and has a similar degree of
roughness and grime in it. We've sampled this from the
back side of the vehicle. And this reiterates that no matter the size of the
source of texture, if it looks like it belongs to that part of the
object, it will work. Notice again that we've decreased the strength
of the brush to mix the textures up on the
top side of radiators, the outlining can be
made a bit better. So we'll sample the bottom
side and just clone it. The top, that's one side, that's the other makes
it a bit more defined. The area above the engineer
intake is not defined. We'll put our three
decursor there, we'll clone this as well. Inspecting again like we
do every now and then, we see there is a streak that we can just get
rid of right here. We'll tidy up the area around the black **** here to paint
out the spilling textures. To add a bit more detail
to the front part, we'll identify the areas
in the duplicate mesh we can grab the texture data
from. We'll look around. The gunner's hatch here has just the right amount
of detail that we can add to the front
part that is done. The texture at this
point might look much less detailed
than you wanted to, but keep in mind that this
is only the diffuse texture. When you add in the normal
map and roughness map, it will give that
micro detail to the texture and it will all
come together really nicely. We're now looking for an area
where we can clone texture from for the generator
compartment in the front, you can expect that painting in the crevices can
be a bit tricky, especially between two
acutely angled facets, just as we have here, between the turret and
the body of the vehicle. While painting in
these crevasses, keep in mind that you should not paint with bright colors because dust and grime
naturally tends to accumulate in the interface
between two surfaces. We don't want to get rid of
that because that gives us that natural environmental
eroding aging process. In summary, when painting in crevasses sample dark textures. Now at this time
we're mostly done by the first and second part
of our texture painting. The third brief pass, we'll just sample a few areas randomly and just lightly dab
on textures here and there. This is in order to
further break up the textures and have
less uniformity. Be sure to be gentle at this phase because
you don't want to undo your hard work that you've put in the first
and second passes. You can easily get carried away by trying to add in more detail, but in real you're
taking the detail away. If during this
pass you think and see that the textures are
getting blurred again, you can always
press control Z and revert to previous state again. Make sure you keep saving
the image by pressing all S or going into
image and save. We'll paint the mouth
of the main gun black. And clicking, instead
of dragging to just sample one area
under the three D cursor, we don't need much
variation in this area because that's just a cover
at the end of the gun. We'll now re examine our model how it
looks at this stage. We'll keep the camel part on the generator area because that just gives another
level of variation. At this point, the duplicate
mesh has done its job. So we'll get rid of it. X, delete the war disease. We go back to the
object mode and we give a final look to see if there's any area that needs some work. We'll look around and we
can identify that there is a small patch of area on the top that looks
a bit blurred. So we'll shift back to texture
paint mode and just paint over using the surrounding
areas as the source texture. Just briefly going over them. We're not too bothered
at this stage giving another look and it looks like
our texture is now ready. Be sure to save the texture
by alt S. This is how the material looks in the
material preview mode with all texture maps combined, preview exaggerates the
normal map or bump a bit. We'll shift to cycles
view and we'll see how the texts are
looking in cycles. We'll look around the
model to identify any anomalies in
shading, modeling, topology, texturing, anything at all that we
can repair at this stage. Now if you remember, the
underside of the model was left open due to the
Boolean cleanup operation. Now's a good time to
address that as well. Before we prepare the
model for export, we'll add in a plane, bring it down to match with the
bottom plane of the APC. We'll reposition it such that
just sits in the center. We'll refine the position a bit. Once that is done,
we can scale it up to match the bottom
part of the vehicle. We'll make sure that it does not cross the confines of
where it should be. We'll take a view of how
it looks from underneath. Adjust the size a bit if
necessary. There we go. And then from the side view, moving to the center and scaling
it in the y axis, Again, refining it a bit just like so we won't scale it all the way because we will add
an area of transition as the next step for
that in the edit mode, let's go to select mode. Select the edge. Press all
Z for x ray view to see through what we're doing To extrude that
edge a bit higher, a bit angular, to meet
the body of the vehicle. We'll keep switching between perspective view and
right orthographic view. This is in order to calibrate the exact position of this edge. Let's switch back
from x ray view. We can bring it down a bit in z axis to match with
the body of the model. The edges are showing
up on the side. We'll scale it down on x axis. And that takes care of it. This looks all good. So now we'll go to the backside, select the edge, bring it a little closer to the back side. Once that looks good, we'll extrude an edge
from here as well. Move it up in the Z axis so that it lines up
correctly with the body. Look for any gaps
which we do see here. So we'll move it a
little back on Y axis, the edges are peaking out again. Similarly, we'll scale
down the edge on X axis and to give a
smooth transition, we'll press control B and leveled bot edges control B again and scroll for
the number of edges. Four would do just fine. Go back to object mode,
select the plane, select our APC and
press control J to join it all together. We'll now go to edit mode again. Press L to select the plane
and go to UV editing. Normally we would project it
and place it somewhere in the UV space for it to inherit textures like we did
with the triangles. But here by default, it looks like how we want it to. Therefore, we won't bother
meddling with its UVs. And also because the
underside of the model won't really be visible
in the final renders, it's just there to aesthetically
complete the model. As a final step, before we
convert the model to low poly, we'll look around and
see one last time if there are any incoherences
within the model, including the mesh, the shading, the textures, and all the maps. When that is done, we'll
duplicate the mesh and move it to low
poly collection. There we go, hide the
high poly collection and rename the duplicate to
denote that it's low poly. Add in the decimate modifier and then give it a new material. This will be based
on the new UV wrap of the decimated mesh. We'll name it Scorpion Low Poly. Now we'll proceed to
decimate the mesh. We'll aim for the poly
count 20-25 thousand tries, a ratio of 0.15 brings
us to that level. We have roughly 23,600 faces, which is right in the ballpark. Just for the sake of comparison, we'll give it the
previous material to see how it's holding up
with this decimation. We can see that there
are certain areas where texture just cannot hold up with areas
of decimated geometry. See, the line here in the
storage compartment is skewed. We'll fix these
issues with baking. We apply the decimate modifier. In the edit mode, we can see
how simplified the mesh has now become from the
original 5.3 million tries. We can unwrap this again for a bit more textile density or we can just go ahead and
bake our textures now. And we'll go with
the second option. We set up our simple
bake by turning on both instances
of the model bake, selected objects to
target object turned on. We select the target
object as poly, set the texture resolution to two K for this low poly version, for a good usability, edit the export
path to low poly, two K. All three maps are
selected and we click bake. It takes little time to bake
four K to two K textures. We'll now select
our low polymesh. Give it the material
that we intended to give disable Hi poly model from view control shift
with principle shader selected and select all
our baked texture maps. Let's go to Rendered View
to see how it's looking. Immediately you can see that the line is restored to normal. The other small
texture discrepancies as well and the model is
holding up really well. The model is made of metal, so the metallic goes
all the way up, tinkering with specular a bit, but it looks fine as is. So we'll just leave it be. We'll save our hard work
and congratulations, We have successfully brought in a real life
object into three D, facing real life limitations. And you can see that
it renders fast. And it's just a delight to see you can make it a part
of your asset library where it will live forever for any project that you want
to make it a part of your very own Scorpion APC to make it presentable will enable both high poly and
low poly meshes. Align them side by side so we
can see them both together. Set up a front view. So it's all nice and tidy. Just like that. We'll import the text that we wrote for
our free ferret model. Copy it. Paste it here, edit it. The Scorpion developed in 1970s for advanced
reconnaissance. It's also a light tank
if you want to add that, but we'll stick to vehicle here. Make sure that hy
poly and low poly correctly set up
the view angle to greet the viewer
with a nice angle When they open the file. Be sure to grab the
free resources. I've provided the video
data for you to practice with when you do share
your renders with me. I'd be absolutely thrilled
to see them with this. Thank you so much
for tagging along. I wish you well and farewell.
Hamza Meo, Limitless
