Final Major Project Blog

Me and Luc have teamed up for our FMP.

Check out the blog for our development for it:

http://dystopianslumsfmp.blogspot.co.uk/

New portfolio up

Last few weeks ive been putting together a new portfolio so i can start seriously applying for some jobs.

Take a look:

http://danhargreaves.com/

Character Project : Part 3

Conclusion

I’m very happy how this project went. I was partly apprehensive because half the project had to be done over Christmas, which I knew I wouldn’t have a lot of free time, but I managed my time well. I learnt new techniques and bits of software which I plan to use again. It wasn’t all plan sailing though as I did run into a few problems along the way.

Looking back onto my concepts I feel like I followed them pretty closely and the end result looks very similar to it. But my original plan was to have 2 holsters with their contents, one to carry the bow and another to carry the arrows. But as time progressed through Christmas I realised I didn’t really have time to produce another one along the arrows and bow. I feel like I could have produced them but as a consequence I would have rushed everything else making the overall character have less polish. But I think I made the right decision to have less accessories but more overall polish.

With previous character projects I always felt like i fell down when rigging, so I wanted this one to have good deformation. Along the modelling phase I tried to keep how it would deform in mind, such as adding extra loops on the shoulder pads (which paid off in the end). But I did have a major issue with the skirt. If you look back to my concepts you’ll see that the skirt actually goes all the way down to the ankles, but when presenting this idea I was told it wouldn’t work well when rigging. So I made a slight change and brought the skirt up to the knees, thinking this would be enough. But after rigging it and animating the legs moving apart the stretching becomes very obvious in the center. In hindsight I should have but some extra loops down the center of the skirt.      

One thing I was a bit iffy about and something I was told about when presenting my idea was in my moodboard the warriors in their armour looked very wide, which wouldn’t translate well on a female. But I think I did a good job, the character looks feminine. I did this by exaggerating female anatomy through the armour – having wide hips and a narrow waist.

Im very happy about everything else though. I really felt like a nailed the surface properties (metal looks like metal, leather looks like leather etc.). Im glad I attempt the shells technique for the fur, I think it came out pretty well.

Character Project : Part 2

So its been a while since my last update – ive just been super busy making everything I didn’t get round to doing an update. But its all finished and handed in now so I can carry on where I left off.

So after I was happy with my base anatomical sculpt I moved onto dressing her up in her cloths and armour. The best way I found to do this was import my anatomical into Max where I would build a basemesh of an item of clothing say a boot around the anatomical model.

Modelling the basemesh for the boot

I imported this base mesh into zbrush wear I would sculpt in all the detail. This is the same process I went through with all pieces of armour and clothing for the model.

Finished sculpting the boot

When it came to the stiched on the leather armour I had 2 ways to approach this. I could either sculpt them in zbrush which would be the quickest option, or I could model the stitches and place them one by one in max, this method is longer but would give me better bakes. I opted for the later.

Placing the stitches on the armour

Below shows the final highpoly:

Highpoly

Next step was the low poly. To do this I imported lower subdivisions of all my highpoly sub tools to model around, I call this my ‘Trace models’.

90% of the lowpoly was made with standard box modelling or plane edge extrusion techniques.

Modelling the hand using plane edge ectrusion 

For the head I looked into one of zbrush 4r6’s new feature called ‘Zremesher’. Which basically produces a low poly mesh from your high poly with a click of a button, you can also paint in guide lines to effect the edgeflow, which produces a better quality mesh.

I was very impressed with the results and how quick and easy it was. But I felt some of the edge loops weren’t as good as they could be in some parts, plus the silhouette wasn’t as strong as I wanted. So I decided to model it from scratch using the same method as the hand. I felt by doing this method instead of letting zbrush do it I had greater control over everything.

The results from Zremesher on the head

To make this character as realistic and feminine as possible I created alpha planes to use as eyelashes

Modelling the eyelashes

Below shows the final low poly model. Which has a triangle count of 11,513. The budget was 15,000 but I felt I nailed everything and adding more polygons to it would be adding it for the sake of it.

Lowpoly

Unwrapping was a quick and easy process because of the brilliant and free unwrapping tool called Roadkill. After unwrapping I baked out all my maps using XNormal.

Next up was texturing. After my bakes were finished it was time to move onto texturing. I overlayed my AO and cavity maps to get a good base to work from.

Texturing was a straight forward process of looking at references of real materials and trying my best to emulate them within photoshop. 80% of my textures are hand painted as I feel I have much greater control in the feel and look of the texture than just pure photo sourcing. Brushes such as scratch and grime brushes help quicken the texturing process. I only use photos to get that nice surface variation and micro details which is very difficult to achieve with hand painting.

Below you can see me adding slight wear and tear around the armour plates on the body armour to give the texture more character as well as making it realistic.

Early stages of texturing

After all the main texturing was done it was time to focus on the alphas for the fur.

One of the techniques I read about over summer I wanted to attempt was the ‘shells and fins’ technique which was used heavily in “Shadows of the Colossus”. This method creates realistic and volumetric looking hair and fur.

This is done by creating an alpha texture full of dots which represent a strand or clump of hair. The polygons the texture is applied to (aka shell) is then duplicated, moved forward and scaled up a tiny bit. This is done a number of times. The more shells you have the more realistic the hair, but this become very intensive for the engine, I settled with 10 ‘shells’.

Below shows this effect in action better. As you can see as the camera rotates round the head the fur will follow it. Giving off the impression of nice volumetric fur.

How the fur works on the hat

After texturing I rigged my character, posed it and took it into Marmoset 2.

Character Project : Part 1

So we've been given our last professional brief before we move onto our FMPs.

This brief is to create a female warrior from either India, Africa, China or Japan. She can originate from any point in history up until the present day.

I see this brief as an opportunity to create one of my original FMP pitches, which was a medieval character in full battle gear, focusing on several material types and using a few techniques I researched over summer. Most notable the shells and fins technique used in the “Shadow of the Colossus”, which creates a realistic volume of fur . The only difference between my character FMP and this is that it has to be female and of an Asian ethnicity.

I looked into the Mongols, and really loved there style of light amour of leather pads with metal plates and all the different layers of clothing they use to keep warm – use of  fur, cotton, leather and sheepskin. The Mongolians are famous for their use of archery, so I wanted to include that into the design as well.
Below Is a moodboard I put together to help flesh out some concepts.

Moodboard

Focusing on the armour design and other items from the moodboard I created several designs to give me a stronger idea on what direction I wanted to head in.
I really like the 5th one (bottom right) as it has a strong silhouette with lots of interesting layers and materials.

Designs

Next I create an orthographic shot for a basic female figure. This was used to create the base mesh for sculpting. Creating the base mesh was an easy process. It was tricky to make sure it was made up of uniform quads throughout the whole mesh, especially the more complex areas such as the hands and face. For these areas I used the edge extrusion method as this gives me greater control of the edge loops.

I spent a bit longer than usual creating the base mesh, making sure it was anatomical correct and had a good edge flow. As the more accurate and better I make the basemesh now, the quicker it’ll be to sculpt an accurate highpoly and add items such as armour.

Basemesh

Using a handful of references I sculpted the character, as she was an archer in the army she had to be fit, so I sculpted a slightly muscular body. Certain muscles on the back (such as the traps, delts and teres) would be more defined as these are the muscles which work when you pull the string back on a bow.
Even though the only parts of this sculpt which would be visible on the final version was the face and hands I sculpted the entire model. I did this as its more practice (and practice makes perfect!) plus it gives me a better idea of how clothes/armour will sit on her, and how folds in the fabric would flow around her.

Highpoly

Next I will be adding all the other parts to the highpoly, such as the armour and clothes. Check back soon.

Personal Project : Phone box

Last week after the hand in for the vehicle I was wasn't overly happy with the results. So to pick up my spirits I decided to spend the next few days making a small but detailed prop for my portfolio, so I wanted the standards to be very high.

I wanted to create something which would be applicable to the majority of studios out there and to have a bit of character and push my skills, especially my hard surface modelling. Which I found in an old 90s American pay phone, It seemed to fit the bill quite well.

My first step was research. I gathered a handful of references, ranging from destroyed payphones in the dump to brand spanking new ones, along with close up pictures of all the major parts so I had accurate reference to model from. Below shows these references:

References

I really like the idea of the payphone being from  ‘The Projects’, in places like Baltimore, Detroit etc. Where I can play with graffiti/gang tags and have the phone box destroyed/neglected. If you've seen The Wire then you know exactly what I’m going for.

First things first I made a quick whitebox so I could get a general feel for the proportions and scale. I place a biped with an average height to also get a  better idea of scale. Below shows the white box:

From the white box I produce the highpoly. I tried to have the cleanest mesh I could to avoid any pinching on the model which could affect baking later on. Most of the model was relatively straight forward to model so I didn't come up with any issues. But some parts required a little more attention, such as the holes making up the phone logo and the phone cable.

To produce the telephone logo made up of holes I first creating the housing part, inside and out. I then created a bunch on cylinders in the shape of the phone logo, then using the Boolean method I subtracted the cylinders from the housing mesh, this was done on both sides.

Next was the phone cable, this was a tad long winded.  I created one section of the cable which I then instanced over a distance of about a meter. Then using a spline I created a path of how the cable would twist/look going from the box to the phone. I then used the deform to path modifier on the cable mesh. This took a few times to get right as sometimes the spline wasn't smooth enough which created very jagged movements when the mesh was applied. But after a few tweaks I got it right.

Highpoly

Next up was creating the low poly. This was a straight forward process so I won’t go into it. Modelling the phone cable could have been an issue, but it wasn't. This was because I used the line path from earlier, this allowed me to get the correct shape while playing with the parameters to make the mesh suitable for a low poly version.

Next up was unwrapping. I decided to keep all parts of the phone box texture unique as I wanted unique details which would work when mirrored. Such as graffiti/stickers. The back parts of the pay phone was mirrored as these would not be seen.

I wanted to produce the texture as realistically  as I could. So  I looked at a lot of pictures to analyse materials and interesting parts they have, such as stickers half peeled off but It still leaves that sticky residue, or burn marks on the side of the housing and extra wear and tear around the metal edges. Especially so around the coin slot as when money is inserted into these slots it always scuffs the edges around it. I also used references to help me created realistic spec and gloss maps.

I also got hold of the latest version of marmoset. Which is very easy to use and has a range of parameters to play with to get the absolute best out of your model.

Below are the final Screenshots:

Final model - 1,449 Tris, 1024 D/N/S

Final model - 1,449 Tris, 1024 D/N/S

I got some really good feedback from my peers about tweaking my spec and gloss maps to improve the material readability (which have been tweak in the version shown). But I think I will take it further and tweak it some more. I also want to produce a LOD version, which I will do over the coming weeks.

This project went really well, im very happy with the results, the time frame it took me to produce and learning some extra techniques.

Vehicle Project : Part 2

After finalising my vehicle model I moved onto making the environment. The tri limit for the vehicle was 10,000. But when modelling it I tried to make every triangle count and not needlessly put in more edge loops just because I have more triangles in my budget to play with. This is why I have only used 4,068 tris. I believe adding anything more will not change the silhouette and I would be adding loops for the sake of it

My plan for the environment was a small repair garage/pit stop. The aim for the environment was to keep it simple as to not distract the viewer from the vehicle. The plan for the environment was to have one strong light source. This was in the form of several bright strip lights directly above the vehicle.

For some of the more detailed environment assets I made a highpoly for them to bake down from. These assets where the shutters and the toolbox. The shutters were made very quickly as all I had to do was make one ‘link’ which is made up of chamfered boxes which are then duplicated to create a wall of these links to create the shutters. Below shows the highpoly assets:

Highpoly shutters

The environment was then modelled and textured. The budgets for the environment was 10,000 tris and 1 1024 D/N/S. The scene came to 3, 068 tris as I didn’t need any more. The texture budget was split up into 2 512s for the tiling textures and 2 512s for the unique assets.

Next stage was unwrapping the vehicle. Unwrapping was an easy process, some parts of the model I mirrored to save on texture space, but the more visible parts I didn’t mirror as I want asymmetrical detail across these parts.

Different texel density. More UV space for more visible parts 

Texturing was a simple process. But to create realistic layers of damage I textured the vehicle in a way that a real vehicle is painted.  Like this: base metal > redpaint > whitepaint > decals. This allowed me to mask off individual layers to reveal the layer beneath, creating believable damage/scratches.

I also looked into metal discolouration for the jet engine exhaust. This simple effect of adding a few different colours adds a lot of character to the model.

Diffuse Map

The brief asked us to produce a lightmap for the environment as the shader we were given was limited in the number of lights in the scene and lack dynamic shadows.  I made this by setting up lights in the environment where the actually light models were, rendered out a lightmap which I then applied to the second channel of the environment. This gave me a nice and accurate baked shadow of the vehicle on the floor.

Setting up lights for baking the lightmap

We also had to produce a cubemap for this project. It was one of the most important parts of this project as this technique produces realistic reflections of the local environment on the model.
It was a simple process as well. I set up 6 cameras, facing 6 different directions and placed them where the vehicle will be. From this I took a screenshot from each camera and stitched them together in Photoshop in a certain order.

Setting up cameras for cubemap

Cubemap result

Along with the cubmap we had to create fake HDR. This was done by painting what parts of the cubemap you want visible in the alpha channel. The way I did this was merging all the layers and changing it into greyscale, then tweaking it through curves. Doing this made my lights a bit darker, so I repainted them pure white to get the strongest highlights I could when using the shaders.

HDR

After this was all done I could start tweaking all the parameters in the shader to get the most accurate material properties I could.  After I was happy with the results I figured some parts of the model didn’t reflect  the model just came from a race – it still had that highpoly polished surface look. I looked at my spec map and realised that it was too white/little contrast, which gives very little surface variation. I solved this by overlaying a few different photo textures over the body. This small change gave a real nice surface variation, almost like loads of dirt being kicked up settling on the vehicle.

Before and after tweaks to the spec map

Below are the final renders within 3dsmax’s viewport using codemastsers car shader:

Overall I feel like this project came out ok. A lot of decisions i made turned out to be the wrong decisions. I feel like this is the first project in a very long time where im not happy with the results and given a chance I would completely redo the project. 

I think the first major mistake I made was to abandon my original idea of the Caterpillar 797F Dump truck which I had in my head for at least 2 months prior to the project starting. I dropped this decision because the project brief has a great focus on materials and I thought the dump truck didn’t have enough varying materials to show off the Codemaster shaders. So I decided to concept my own vehicle with different material properties.   

     

I think the vehicle came out relatively well, different material properties can be recognised. But I think the environment and lighting brings down the overall quality of the scene. The lighting within the scene is rather flat, making the environment look very amateurish. My environment idea was to have one bright light source coming from the ceiling which is how actual pit stops look. But this made everything overly bright and washed out. In hindsight I should have had another light source within the scene with some colour from it. Possibly have the shutters open so the sun would be another source of light, this would also give more interesting reflections on the vehicle as well as produce a more interesting light map. 

I think this project has been a major learning experience for me. Learning to use shaders used by actual game studios and all the little tricks they used to fake certain techniques has been great. I will take everything ive learnt from this project and apply to all my future projects.