3D Week 09 – Assignment 3: part 4

Hello!

This is the final update on the final assignment of the 3D1 course. I’m really pleased with how it turned out, and by comparing this with my first assignment (the sci-fi crate) I can tell that I have really learnt a few tricks. The modelling is so much faster and the final mesh is much more optimized. The texture is more appealing (and more fun to create) as it is hand drawn instead of using photos from the internet.

Well, here is how the final character turned out in UDK:

  

Here is how the diffuse look like:

And the specular:

And the Normal:

---

 

And that’s it for the first 3D course. This course has been the most interesting and fun so far form me. Everything has been new, and the learning curve has been going straight up. It feels like every week a hundred of new things was though out and that’s really cool.

Until next time; Take care!

-Lui

3D Week 08 – Assignment 03: part 3

Hello there!

This post is just a visual update on the Assignment 3.

Here is how the final body mesh looks:

And some details:

Here is how far from the concept silhouette I have modelled. Notice that the front pose concept art is not symmetrical and therefor the differ on the left side:

And last a comparison side by side with the turn around:

As you can see I have tried to stay as much as I can to the original silhouette, and then added some more interesting lines to it. My client (Seamus) is totally cool with this and he says that he likes how the final mesh turned out, And so do I.

That’s it! CYA!

-Lui

3D week 08 – Facial modelling

Hello!

This week has been about modelling faces. We talked about different methods of creating facial meshes and theoretical about edge-flow and topology for facial projects.

One faster and simpler ways of convert real life object to digital 3D meshes is by scanning them by simply photograph the object from all the right, necessary and different angles. The photos need to cover all of the faces, edges and angles that you want to have on your 3D mesh later on.

When all photos are taken, upload them on to your computer and open them in a 3D scanning program. We used the Autodesk 123D Catch as it is a great program that will scan your photos for free. What the program does is that it takes all of your photos and arranges them in a 3D space which then creates a 3D mesh of the photographed object. It also gives you a texture map for your object created from the photos as well.

At the motion capture studio in school we took photos of our own faces for converting them to 3D meshes by using the scanning technique just described. Unfortunately the process did not work for me as my photos was a bit blurry so they did not transform in to a correct mesh of my face (my head was all distorted and only half of my face did make it to the mesh, so I could not really work with it) so I had to use an already existing scan of a face for the anatomy parts of this week.

The anatomy of a face is rather simple actually. The key parts for the edge-flow grow from the mouth and eyes areas and meet across the nose and on the chin. Around the eyes the edge-flow goes around in circles then meets like a mask that connects both of the eyes (I have only modelled half of the face as it can be mirrored, so you’ll have to imagine the other half of the mesh being identical). For the mouth it gets modelled the same way as for the basic eye edge-flow, that is circles around the mouth, and then reaches out to meet up with the rest of the mesh.

Here you can see the edge-flow for the eyes and the mouth:

Here you can see the edge-flow of the face and how it connects the two parts:

Note: the software used here is a trial of 3d-Coat.

After reading the article on ‘Photogrammetry’ in ‘The Vanishing of Ethan Carter’ I can only say “wow!” these guys really knows how to use photo-scanning as the tool for what it is, and they also really have the time and skills to do it. When we were making our models from photo scanning we used about 9 images for the process (only for the face) while they were using everything from 26 photos up to hundreds or even thousands of photos which resulted in that they got a much higher quality mesh of the object.

Also when we were creating our objects we did it from a living creature (lol. yes I am still alive ;p), and as they state in the article it’s really hard to capture moving objects:

“Photogrammetry gets confused by moving subjects. If you want to scan human face, even tiny head/facial muscles movement between photos means errors in scan and lots of manual cleanup and resculpting.”

But with the right set up, the right settings and the right equipment a perfect shot of a moving object is still possible. So I won’t give up on this technique, it actually really cool and I want to try it out for more complex objects with even more shots trying to capture that perfect  mesh of a real life object.

And that’s it for this post, CYA in the next one!

-Lui

3D Week 07 – Assignment 3: part 2

Hi!

This is just a really quick visual update post on the assignment 3 showing my work in progress of the base mesh and the edge flow. At this point I have really sticked to the original concept when modelling, but as I have some troubles with the turnaround I received from Seamus I will do some remodelling, mostly on the sillhouette to get a more appealing resoult. I will also redo some of the topology so it will become a mesh that is better suited for animation.

---

Well take care!                                                                                        

-Lui

3D week 06 – Assignment 3: part 1

Hi everybody!

This week we started to work on our assignment 3, which will be a humanoid character (it doesn’t have to be a human but as long as it has two arms, two legs, a head etc. it’s okay).

When creating a character first of you will do a concept of what kind of character it is. Like if it is a human or a monster or whatever, if it is male or female, if it’s young, old etc. a kind of background check so here we go for my character.

Obviously my character concept is inspired from the Pokémon Squirtle. Squirtle lives in a world that is populated with humans and fantasy animals. The animals/Pokémon’s have different ability’s according to elements such as earth, metal, air, fire, water etc. Squirtle is a water type Pokémon so he would live near water or in water. According to the Anime I would say that the world takes place in current time. The people in this world have “smart devices” and a pretty high standard technology very well comparable to what we have in our world in present time.

About gender and age, my Squirtle is a male in his younger teenage years. Compared to human age he would then be around 13-14 years old boy. The differentness between a female Squirtle and a male Squirtle can’t really be spot and they are unisex in their appearance so there is really no definition for the genders. Squirtle is like a koala in size, around 70- 80 centimetres in height and weights 10 to 15 kilos. His skin is of amphibian type, scale and dry skin that is.

The Humans are the leaders in this society while the Pokémon’s are more like pets to them. The wild Pokémon’s often want to stay free and not being captured by humans so whenever they are being approached by a human they often stay aggressive and defensive, and will fight for their freedom. The Pokémon’s doesn’t seem to mind though once they have been caught by a human (if this isn’t Stockholm syndrome then what is? and no one seem to think it is weird lol). The captured Pokémon’s often becomes best friend with the human that caught them so once the dramatic kidnapping is over everything is very peaceful and happy. My character is still a free Squirtle though.

The economy for humans would be something compared to what we use such as coins and bills. The Pokémon’s use more of a trade form in their economy. They trade food or objects for food and objects so the Pokémon’s don’t use an actual currency when performing trades.

Well I guess that’s it for the background of my character. So here’s the turn around the model should be made from made by me using illustrator.

 

---

But as in this assignment we’re not going to make our own characters I’m not going to model my own Squirtle. The guy who is making my concept is named Alexander Westerdahl (you can find his blog here if you want to follow up this project or just check out his work, which I think you should as he is a great 3D artist). I am going to make a concept from Seamus Newman (blog here, and you should really check him out aswell) which I will blog about next.

And until then; take care!

-Lui

3D Week 06 – Assignment 2: part 3

Hi everybody!

This is the final update on the Assignment 2, just to show everyone how it turned out.

Tris count: 993

Texture size: 1024 * 1024

 

Hey ho, let’s go!

These are screenshots of the dagger/knife in UDK:

This is how it looks in 3DS max:

This is how the diffuse map looks like:

And this is how the specular map turned out:

And this is the normal map:

That’s it! Moving on to the assignment 3. See ya’ there!

-Lui

3D Week 05 – Assignment 2: part 2

Hello!

This post is just a visual update of the assignment 2 for the 3D1 course. The screenshots is from my second attempt on this model as the first one was kind of sloppy done.

The new model has a total poly-count of 536, and a tris-count of 997. The first model had a poly-count of 799 and a tris-count of 1509, so there is a lot less of polygons in this version and this one is much more optimised. When I was making the first one I did work on it aiming at a much higher resolution on the topology than this version. But as the knife is a rather small object, there is no need to go as high as I did then. Then I did also change the shape a bit more and exaggerated the edge-flow to make the cartoony art-style more obvious.

As I did change the model I did also need to redo the UV layout. Here is how it looks (the red area to the left). The image also shows the knife with the UV-checker on to show so that there is very little distortion both on the top and the bottom of the model.

That’s it for this Update. There will be one more post on the assignment 2 next week.

Cya!

-Lui

3D week 05 – Assignment 1: part 4

Hello!

This is the final post about my crate and my first assignment in 3D graphics 1.

A 3D model often consists of at least four parts. The mesh which is the actual 3D model, the diffuse map which is the texture that holds all the colours, the specular map which tells the secularity of  different parts of the object, and last the normal map which contains information making details look realistic and not like “stickers”.

A diffuse map is basically a 2D image that gets wrapped around the mesh to give it the basic pixel colour from the bitmap used. A good diffuse map is made without a directional light as those are made whit the specular and normal maps. It should only have the basic ambient occlusion by getting darker in cracks and deeper areas. When working on your diffuse map you mostly work with the saturation levels and “raw” colours.

The Specular map is like the diffuse map a 2D image wrapped around the mesh, but instead of containing the colours for your object the specular map is a grey scale of how shiny your object is. The brighter the value, the higher specular the surface gets. In real life, one of the first thing the human eye notice is the secularity of a material, sometimes even before the colour of it. So to get the secular map right is really important to make a legit model. For example if you’re making smooth metal that is shiny you want to use a bright value to simulate that glossy surface, but if you’re making a dull surface like textile you will make it much darker as textile reflect almost no light at all. Take a look at your clothes and you will see that almost no light from other objects gets reflected from them, but for a mirror (a very smooth metal surface) almost all you can see is light already reflected from other materials.

The normal map is used to give your mesh a much higher resolution look than it actual has. A normal map replace what otherwise would have used a lot of more polygons. You should use the normal map whenever you want details on your model that is not interfering with the silhouette too much. What the normal map actually does is faking light and shadows on your mesh creating an illusion of that higher poly mesh for a much lower performance cost. Interactive normal maps can be used for creating effects when interacting with them, for example if you shoot at a wall the normal map will change so that the hole doesn’t look flat by getting proper shadows.

---

This is how my final crate turned out. I’m pretty satisfied with it as it’s my first model ever, and it even got a diffuse map on it. I have only played around with 3D modelling a very few times before and most of the time I had no clue of what I was doing. So this is my first finished 3D asset and I think it turned OK. (the image is rendered in UDK)

 

And this is how my Diffuse turned out (you can check my latest post about how I did this):

 

This is my normal map:

And this is how my specular map turned out:

Hope you enjoyed reading!

-Lui

3D Week 04 – Assignment 1: part 3

Hello. It’s time for an update on my sci-fi crate, so here we go!

The inspiration for my crate is all from the Halo and Portal games. Well for shape and model I was looking more at the Portal’s “companion cube” as I hope you can tell from the rough and hard edges I have in my crate. When it comes to texture mostly of what I aimed for is from Halo I would say, as the crate I am trying to create is a bit more serious than those from the Portal’s series.

So for the diffuse map (the map that holds all colours, icons and so on. think of it as a textured piece of paper that is wrapped around your model.) I chose the materials metal, rubber and carbon fibre as they are rough and almost industrial, just like the Spartan suit from halo is. For getting the right appearance of the materials I will have to use a specular map to get the reflection and normal map for the texture of it but that’s for next week. As the Spartan suit’s is themed with one high saturated colours of either green, red or blue I thought my crate would have one of these as colour theme. I chose red simply because I think it looks better than the others. (And maybe a little because I didn’t wanted to do blue just because “sci-fi” is supposed to be “blue”, wanted to try if I could make it work with red.)

The colours I used (well, it was only red) I made pretty hard saturated. The colour is distinct and even darkened a bit to compensate for the brighter metal scratches and polished parts inside the spheres. The rubber I made with zero saturation as to make contrast with the redness so they are basically all dark grey. To give them a little bit of wear and tear to fit the rest of the texture I highlighted the edges as when rubber or plastic gets damage they tend to brighten a bit. It also adds a bit to contrast so the grey parts aren’t all grey and boring.

The colour I use is warm. As sci-fi tend to have a bluish tone to it, it is often a cold felling in a sci-fi image overall and as I stated before I wanted to see if I could do the opposite of the “stereotypes” of sci-fi. My crate also have a dirty look to it, it looks like it have been out in the sun and hot weather for ages, maybe in a dessert. At least it doesn’t look like a new, directly from factory crate for sure.

This is my crate with an “UV-checker” I got from Nataska, our 3D teacher. It is used when doing the UV map. By using this it is much easier to see how large or small the resolutions on the different parts of the UV map are. This is all so that your crate won’t look blurry on some spots as they have too few pixels, while some will have “too many”. You would want a diffuse that is as even as possible. You can also use it to see so that there aren’t any big distortions on your texture and if there are you might need to re-cut your UV-map and relax the tensions creating those distortions and warps.

For my crate I chose to use the same area of the UV map several times so that my result would be of higher resolution in the texture as we were limited to 512px X 512px for this assignment. For my crate that is symmetrical in all dimensions this isn’t a big of a problem as the symmetry already is obvious and the details are so small they are hard to see if being repeated compared to an unsymmetrical crate of a different material (like if you see the same planks of a wooden box being repeated they are much easier to spot). Also, this crate has six sides, and no matter how you try to look at it you will only see a maximum of three sides at the same time, so why don’t repeat the texture on those sides that you can’t see at the same time? This saves both time in creating all of the maps, but the crate also gets a higher resolution by reusing the space on the UV-map.

This is how my final diffuse map turned out. As you can see I reused a lot of space. If I haven’t done this the resolution would have been around 6 times smaller than what I managed to do now. But as I said before, this only works for a few types of situations such as mine.

This is how my crate looks in the Unreal Engine (using UDK). Note that this is only with a diffuse map. The specular map and the normal map will be made next week, and hopefully this crate will look much nicer than it does right now.

That’s it for now!

Hare gött!

-Lui

3D Week 03 – Assignment 1: part 2

Hi!

This week we have been working on our 3D-crates and how to optimize the models for game usage. First of we have learnt that a 3D-model or mesh should only contain polygons with a maximum of four sides otherwise they won’t be handled well and the mesh might end up distorted and not looking like your original model.  Any polygon with a number of sides greater than 4 is called an N-gon (I guess the N stands for any number, like in mathematical formulas, might not be correct though).

When 3D-modeling for games you might also want to consider the amount of polygons you put in to your model. A higher amount of polygons results in a higher “resolution” on your model but drains a lot more performance than simpler 3D-meshes. So depending on what your model or mesh is going to be used for and how close to it the player is going to get you want to consider the amount of polygons. A good rule for cutting down on the poly count is that anything that doesn’t break the silhouette of the mesh can probably be removed from the mesh.  Examples for what doesn’t break the silhouette would be small details and parts of the mesh that is negative extruded into the mesh. This information is added in a later stadium using something called normal-maps or bump-maps.

---

My Sci-Fi crate was optimized and corrected by Simon Brundin.

The problem with my Sci-Fi crate was that there were some N-gons (marked red on the top two screenshots) in it, so they have to be fixed as they had more than four edges. Another thing that isn’t as critical as the N-gons but can be improved is the poly count. By reconnecting and arranging the vertexes in the spheres he did some savings on the poly count without braking too much of the shape. I guess as the spheres isn’t actually breaking the silhouette at al they might can be reduced even more, you might also have to make sure that the appearance of the holes is corrected by using a normal map in a later phase. In this case I’m pretty satisfied with the poly count, so  next time I would’ve just make sure that there aren’t any N-gons in my model.

---

Next model is my medieval crate, and is worked on by Rickard Folland.

The problem with the medieval crate was the same as for the Sci-Fi crate, which was the amount of N-gons (marked red in the two top screen shoots). The high amount of n-gons I get probably founding in that when creating my crates I use a tool called Pro-Boolean ( it can be found under compound objects and is different algorithms for how to objects shall interact with each other where they intersect). I use Pro-Boolean to easier and faster doing my model. By performing a Pro-Boolean the vertexes doesn’t connect to each other just the planes and that often results in polygons having a high amount of sides. By the using of Pro-Boolean the poly count raises significant as you have to fix all N-gons as well. Rickard didn’t finished working on this crate as there where so many faces to fix and he did spend most of the time working on making the topology better as that was something the Pro-Boolean did mess up as well.

---

My last, the cartoon crate I fixed myself.

The problem with this crate was the N-gons (again lol). This crate is way simpler than the others in one way as all lines are straight (no smooth or rounded surfaces or shapes) so this mesh can keep a very low poly count without messing with the silhouette at all. So what I did on this was basically making all faces to four or three sided polygons tops by connecting the vertexes needed or removing the unnecessary ones that occurred when I used the Pro-Boolean tool. Next time I think I’m going to skip the Pro-Boolean tool for some of the steps as it gave me much more work in the clean-up phase, which takes a lot of times. The reason why I tempt to use the tool is that I have done some 3D-modelling for 3D-printing. And when you 3D print the poly count is something you rather keep high than low, and your models need to be fully watertight. For 3D models you only use in digital form (for games and such) you want the poly count to be low and if it saves on your polys the model don’t have to be watertight. So next time I probably will use another method were I can.

---

That’s half for this week.I’m also going to post about the prop I picked at Gotland Museum for the modelling-assignment in school.

CYA

-Lui