Friday, 19 December 2014

Crimson Rouge is back in action!

Ah the holidays, nice to have some time free to do some of my own stuff, there are quite a things I wanted to work on over the break. First thing is to redo one of my odd characters from years ago: Crimson Rouge, a samurai character from the 1940s.

The character as I imagined her back then was some girl that ended up getting trained with a group of mercenaries in martial arts, with her energies focused on speed and the samurai arts. Without going into backstory stuff that is.

First here is an old design from early college;
back when I had no idea how to design clothes, characters or perhaps most obvious here,their faces:

Old piece, please don't judge me.
My goal in this little project is to create a new version that doesn't fixate on this old design but follows my original idea while using all the iteration techniques we have learnt here at DMU. Hopefully I can come up with something a lot cooler and maybe a bit less... rubbish.

I started on some things all the way back last year where I would randomly add cool things to a pinterest board for her, from things such as 40s style coats and jackets, RAF pilot gear, kimonos and samurai armour and weapons. So just before we made a proper start on the sentry turret project a couple of months ago I made work on creating a proper moodboard and doing some random thumbnails and things.
Quite untidy, but with these moodboards I usually move things as I go start to develop the concepts, treating the photoshop document like a pile of loose images to draw connections between.

Ok now for some concept-y things, starting with some thumbnails, then followed by a bit of photobashing:

And here is a really basic colour drop, with some vibrance applied. Take about 5 minutes so nothing special, this is just used as a point of reference going forward for some sort of general colour scheme.

Sunday, 14 December 2014

Dynamic Duo III - Final Models and Postmortem

So from the last post I had my design finalized, all I had to do was create some simple orthographics so I could get going on the model. I only wanted to model one of the characters with my stretch goal being able to rig aswell so I decided to go with the woman as she appeared to be the stronger of the two designs.


I learnt from doing the gladiator project last year that character modelling is no easy task, topology can be a hard thing to grasp sometimes and requires a lot of fiddling with the model to understand how it works. One thing I didn't do as much last time however was gather together a load of reference just for the topology so this time I created a board with a load of images and printed it out (this helps as it means I don't have to flick between the reference and the model all the time trying to remember the reference) 

Topology and modelling reference board - Printing this out along with my other moodboards and concepts aswell allowed me to see everything without flicking between windows on the screen. I found this also helps in making sure my model still looks like my concepts aswell as seeing all the concept art in front of me helps me keep focused on the design.
The head was quite interesting to model as I used the on the far right of the board as the base for the topology, this method required box modelling which was good but it meant I had to be careful in terms of how I rounded out the forms otherwise the woman would've had a very jaggy looking forehead. Box modelling makes it easier to define the forms of the face but for making sure the edge flow is correct strip modelling might be better suited.


Texturing was quite a hard thing to balance as on the concept there wasn't a lot of details on things like the clothes folds because adding too much detail would detract from the original style of the concept which is more minimal, some amount of detail had to be applied so the character had some textural interest though and the hard part was striking a suitable balance between the two.

In the labs one day I was just randomly looking through the programs we had on the lab computers just to see what we had and I came across 3Dcoat, which allowed me to paint texture straight onto my model using the base texture I already created, this allowed me to quickly iterate on the design without constantly resaving and checking 3ds max to see what my texture alterations look like. This was helpful aswell in fixing those horrible texture seams I got as I could just paint straight over them with hardly any fuss.

Using 3D Coat allowed me to quickly make alterations to the texture, adding a blue gradient to the lower part of the hair to separate the glasses and painting the details on the hands were very easy things to do compared to the usual method of painting to a texture sheet,
I also had a debate with myself over whether I should use PBR for this character as she seemed to look decent without the glossiness which I first imagined on her clothes. The flat shading look that I got from consistent colours seemed to work well so I created the other PBR maps and compared a lit and unlit model in UE4 to see which I preferred. And here is the final result:

Lit and Unlit versions of the final model - Which do you prefer? 


I was initially worried about this project as characters aren't my strong point but when I got my Pirate Radio concept down everything became a bit clearer. The first review of my project went well but it was only after when I received much better critique on my designs and then went back to focus back on the original 'cool/thinks they are cool' dichotomy that I came up with a design I was happier with. 

This however lead to less time creating the model and by extension not completing my stretch goals I set on this project which was to rig the model and pose it in a manner suitable for presentation. I will have to react to critique quicker from now on and be extra careful to stay to the original idea more so I don't end up in a stupid cycle of redesigning the same thing over and over, more brainstorming and experimentation in the early stages of future projects could be a good place to start.

UE4 Orphos.png
Orphographic view in UE4

As with the Sentry Turret, an important thing for me to focus on is the actual modelling stage which takes me forever. A bit more speed modelling practice and 3d studies will benefit me here as this seems to be my weakest area at the moment despite having a bit more 3d background before I joined this course. My next project involves me speed modelling a Westland Wyvern fighter plane which should be the perfect thing to help me push my modelling skills.


+Texturing I thought turned out well and matched the look of the concept, also using 3D coat to paint out those seams is another new thing for me and will work on this method of texturing more.
+Head turned out well and features well defined
+Topology seems correct, used reference to create deformations, rigging might help to identify issues
+Orphographics were created quicker and more efficiently, considering my difficulties in creating them previously

-Long concepting period due to not reacting quick enough to critique
-More thorough brainstorming at the start to avoid losing focus of the design, would help save time down the line
-Didn't achieve stretch goals in the end due to redesigns

Final Model Spec


Radio Girl
Tricount: 6763
Textures: Albedo, Roughness, Metal, AO (All 1024x1024 Targa) (Combined Roughness, Metal and AO)

Sunday, 7 December 2014

Dynamic Duo II - Not Cool Enough

After gaining some desperately needed critique on my designs I came to the conclusion that a little redesign was in order. Comparing my ideas for the duo, my sketches and the final design I realized that I lost the original 'Cool and Uncool' dichotomy I wanted to convey so I went back to my sketches and redid the female character:

First I looked at my moodboard again to pinpoint exactly what look I was going for and then expanded on it by extra research into 60s mod fashion and design. I then used these as a basis for some shape exploration.
Extra shape exploration to find an imposing, stronger character.
The character I had in mind I wanted to be cool but the design before came across as more quirky, so I thought by giving her more of a stern, rigid look as opposed to the roundness of the guy would help sell the difference between the characters more.

The next process I used was overlaying value onto my selected thumbnails and refining it, I did this by using transparency lock and painting, a method I should have used in my initial concepting as it's quick, fun and it gets the job done well.

Value separations done within 45 minutes ish, could get them done even quicker with more practice.
Some refinement of the thumbnails; trying to work in the important things like the arms which can be handy.
More refinement, messing around with cutting forms and seeing how much I can add before it starts to detract from the design.
Next I worked a bit more into the guy character after receiving some feedback about his clothes, I created a moodboard like I did for the woman but used this as a basis for sketches which could then be worked into the final design.

Extra moodboard for the guy character
After some tinkering and experimentation I came to the final designs:

Final tweaked designs, ready for modelling. 
The guy in my eyes looks more friendly here and the woman is exactly like I wanted her; her clothes really make her look like a hard-faced bitch and it's great. The Dichotomy is a lot more along the lines of what I was imagining here.

And for the next part: Modelling process and Conclusion

- James

Sunday, 30 November 2014

Dynamic Duo I - All the Dichotomy FTW

New project again! This time though we a centering around creating characters, two of them in fact. This project is titled the 'Dichotomous Duo Project' and we have been tasked to concept two characters that work as a duo but contrast in different ways using shape language, colour and values. We also have to take forward one of the designs through to a textured model within a reasonable tri budget.

The contrast of the characters is the most important part of this project as we have to make sure that they have personality and look like they have storytelling potential, in a nutshell it's making sure they are different enough while still making them look like they belong in the same world. Some good examples are Mike and Sully from Monsters Inc, R2D2 and C3P0 and Batman and Robin.

I started by looking on Pinterest and through my reference image folders to put together a load of stuff I like and see if there was anything cool that jumped out, I also looked at some fashion and a lot of designs from animated studios to see if I can find anything that's a bit different.

I written down a list of opposite keywords and did some other brainstorming in the hope that I can narrow down my options of what I liked, I tried to see if I can make any connections between these different themes and cut out anything that didn't work. I ended up with 3 things I wanted to research: 60s, Pirates, and pulp science fiction.

Going by a piece of advice from Mike Kelly I created a moodboard for each theme and then created a mega moodboard for everything that I used to draw connections between things and filter out even more of the clutter until I got a more refined idea. Eventually the idea taken shape as doing 60s era 'Pirate Radio' characters that could be radio presenters, with the sci fi element as a possible backdrop. What would it be like if a duo of 60s radio presenters were broadcasting tunes down to earth from a rogue orbiting satellite? How could I make it work?

Joint moodboard 


I began by just doing some loose sketches and some shape exploration, I did a load of thumbnails aswell and tried using the shapes and size to help come up with a duo of silhouettes that work well together.

A handful of thumbnails
Handful of sketches
Next I tried to develop the thumbnails using the sketches as a basis for the details, I realised this approach didn't quite work though as I lost a lot of the original energy which was in the thumbnails. So I created value buildups instead so I could rely on the separation of values to create details.

The drawover designs which didn't quite work out
Value buildup from thumbnails, this helped create details while still keeping the energy of the thumbnails.
From these I created a basic line drawing with using the supporting sketches and reference to improve the design.

Value finished plus line drawing after tweaks
I then did some basic colour dropping on my design with some value adjustments to fix it, I then picked a colour scheme from these.

Final colours before feedback
Next up, my response to feedback, and reflecting on my designs. :) Oh and maybe some 3d.

Tuesday, 18 November 2014

Sentry Turret III - FACEMELTING!

Ok I imported my turret into engine and set it up with the blueprint template which was provided. It was good but there was something missing... the lava. At the moment it was just firing cubes so I had to create a projectile mesh and an emissive material that I can use to emulate the lava. However this still wasn't quite enough to make it seem like the turret was shooting blobs so I created a lava particle effect using the same projectile mesh and attached that to the projectile blueprint aswell as a pointlight so it lights the ground up underneath while it shoots.

Some other things I added:

  • 3 Extra Barrels (4 Total)
  • Changed rotation speed and firing rate with testing
  • Added sound but then later removed because it would not play correctly with the firing rate and the sound did not fit the blobs as it was ripped of the first person controller.
  • Ejector socket and blueprint removed to fit design
  • Created a room that better shown off the emissive qualities of the turret and particles
The results were pretty darn cool.

Rapid Fire Action! Fun for all ages! Facemelt your neighbour's cat!
Here is a shot showing off the firing in action:

Action shot of the blobs in flight.

The hardest part of the engine was trying to get the lava working properly in the tanks as I experimented with different ways to get the blobs in the tanks animating like a proper lavalamp. The material was also another pain in the ass as I wanted the colour to change as the lava rose to the top of the tank but after a lot of experimentation I couldn't get it working like I wanted, the best I could do was get the colour to change slightly over the Z of each individual mesh.


After a lot of trial and error on different things I came to a result that looked decent and reflected my original concept. If I were to do something different I would say I would spend less time on concepting, which would have given me more engine time aswell as spending more time on the high poly model which caused a lot of problems. The issue with the high poly model is the bake generated seams at the edges where I wanted the nice chamfering between faces but it didn't work out, also a lot of detail was lost in the normal from some of the smaller bevels which I think is due to to the details being too small for the bake to pick it up. Something to take forward I guess.

When talking about the engine I will say that I love the blueprinting and particle system. I had a lot of fun tinkering with all the options and actually the engine stuff is my most favourite part of the projects so far, I really want to expand my knowledge in this area further because UE4 is awesome.

So with that, heres a promotional poster. Facemelter 3000 is coming your local toy retailer in time for the holidays!

Should be posting a proper post-mortem and self-evaluation later this week aswell, So stay tuned as always.


Sunday, 9 November 2014

Sentry Turret II - Final Concepting and Modelling

Hi again! Continuing my concepting work from last week I created some more refined designs and made some value buildups based from them. I then used these to create some basic blockmeshing.

Refined Thumbnails with multiple iterations
Value buildups, here the turret is beginning to take more shape
For the blockmesh I imported the value buildups onto a plane in 3dsmax and created some simple geometry and then kitbashed some different ideas again from the bottom row, the iteration is strong in this one. :D

I got some opinions from other students on which design  looked better, I found that the bottom right corner design kitbashed with the second along from the bottom left looked best and I taken this to colour.

Colour Comps

The colour stage was an important one on this project as I wanted to have that awesome, 'space age' 80s plastic toy feel, plus it was a personal goal of mine to make something super colourful for once so I drawn from from a tonne of colour palettes to get something cool. 

You better like colour comps because here is the motherload... 

Colour Comps I - First draft of colour schemes, at this point just trying different things out to see what would work and what I liked.
Page 2, just variations on the previous page - messed around with inverting the colour placement and using Adobe Kuler to correct the palettes.

More of the same, some of the colour schemes here reminded me of the kind of colours that retro video games consoles used to be (especially the top left and bottom left)
After these I created on more page where I gathered together the different variations I liked and compare them to the idea I was going for, I was originally going to go with glow-in-the-dark plating but then decided the light on the lava might be enough as a highlight, also I wanted to give my sentry turret a retro Nerf gun kinda feel with the decals and colouring so this allowed me to focus my colour design down.

Final Colour Comps - Here the colours are more narrowed and I picked my final palette for moving forward, I got a piece of advice from my teacher about not going to saturated with all the colours to it doesn't overpower the palette, this is done by staying within an imaginary circle within the colour picker so the saturation isn't too high on all the other colours.
Next I created some Decal comps on the colour scheme I chosen, I referred back to my moodboard for this stage to create some designs that scream '80s toy' and tried to replicate some of the iconography on 80s posters and toy boxes.


Next I used the blockmesh in 3ds max to create a multitude of render passes for the concept paintover: this included the creation of a 'clown render' which is just flat colours applied to the object which could be used to mask out different materials when painting over in photoshop. I rendered out this plus a lighting pass, a flat value pass, a mask and a ZDepth pass which I can use as a mask to control fog effects or DOF in photoshop.

Here is my final paintover, plus a little embellishment text. :)

Final Paintover concept.


Next stage was creating the model. A stretch goal I had for this project was to create a high poly bake using turbosmooth in 3ds max so I can add on really nice smoothing to the plastic and nice surface details like bolts, cables and vents. Most of the work was already done on the model from where I did the blockmesh so all I had to do was cleanup the mesh and add details from the paintover. 

Then I used the low poly as I base for the high poly... which was extremely hard, so far on this project the high poly has been the  most difficult and time consuming thing. I just hate how turbosmooth affects topology so much, though on the flipside it does help me learn what good topology is meant to be like. :D


Texturing was also easier as I could just colour drop from my concept and copy my decals straight over, the later thumbnails were based from the blockmesh which made transferring the decals super easy. Most of the effort was taken to creating the emissives and to the roughness maps, to which I added a darker value towards the edges of the metal areas for a more worn effect.

Wireframe model with texture applied. 
At this point my turret is ready to move into engine, apart from some some work on the texture and the high poly. Now time for the UE4 cool stuff where I can facemelt things with lava...

Next issue! Engine time...

Wednesday, 29 October 2014

Sentry Turret Project I - Arrrggghhhhhhhhh the colours!!!

New project time! This will be our first solo project of the year and the brief is to design and model a sentry turret that can be used in a Unreal 4 blueprint system which is set up to turn to follow the player, tilt and fire projectiles.

First I created a Pinterest board gathering a load of cool random things which could work on a sentry gun, starting off with basic mechanical bits and bobs and then going onto some cool sci fi weapons and a load of retro goodies from Dark Roasted Blend. I then created an initial moodboard of all the images I gathered then did a bit of brainstorming by the side:

First ideaboard plus some notes
I had a couple of ideas floating around before and during this ideaboard: one idea was to have some sort of briefcase that folds open into a sentry turret, I also had another after collecting the images of having some sort of projector or camera that has been adapted into a disintegrator ray but I still wasn't committed to the idea.

I played around with one of Mike Kelly's techniques about randomization which helps which direction to go in, I had a couple of ideas in mind and randomization just helps move things forward by picking one randomly and going for it. After doing this I ended up with going with an 80s theme (It also helped that I came across some really great images that persuaded me to do the theme)

I started to find a load of really colourful 80s imagery and it brought me to my next idea which involved one of my favourite things: Lavalamps. I was thinking it would be cool to have some kind of turret that could shoot out exploding lavalamps or massive lava blobs of death which would really give me the opportunity to play around with colour and the particle engine in UE4.

So yeh colour, lots of it:

*I will not be held responsible for 80s-induced seizures*
This second moodboard is grouped (loosely) into colour reference, retro computers, weaponry and other random 80s bits. From this I began to develop a bash kit from the silohuettes for some simple thumbnail creation.

basic bashkit for thumbnails
Then I used the bash kit and photoshop lasso tools to create a load of thumbnails for the sentry turret, I coloured the lit bits separately so they are distinguished from the rest of the thumbnail and makes it easier to read.

Lava lamp death machine
Stay tuned for the next part of the concepting where I will do some basic blockmeshing and some refined thumbnails. (and maybe a bit of intense colour dropping)

Bye bye.

A look back at the film room project

So we finished and presented the film rooms project on thursday and now seems like a great time to reflect on how things went and show what we ended up with. Since my last update we created models, textured them and taken things into engine aswell as meeting up round my flat to create a physical model...

Film Rooms Project - Asset creation

Assets I created for the tent plus engine shots

After reconstructing what the assets looked like with the concepts and using the blockmesh as a base I went to create the models I was assigned for this project. I started with the record player and attempted to do a high poly version of it by baking turbosmooth down but I couldn't solve the topology that needed to be fixed in order for it to work properly in the time I had; so I went and modelled it normally and focused on getting the details right on the low poly until later on.

I did do something new during this stage however! I had some help from Jake on creating a ZBrush sculpt and exporting it for baking to a low poly in Xnormal for the bed which sorely needed some extra detail in the folds, baking is an important skill for creating realistic assets so it's nice to finally get into the basics of it. I am aiming to do more of this as I go forward into the next project. :)

Bed Sculpt in ZBrush


I went onto texturing next where I experimented a lot more with adding detail and variation to the roughness maps as during the asset swap project my textures didn't pop as much as I would have liked them to. I played around with using grungy brushes on the roughness map on a low opacity and building up subtle details which instantly made the materials look a lot better in UE4, I also did the same with the albedo in adding some colour variation.

The texture work required a TONNE of feedback and back and forth to get it looking right with our lighting setup as the light made everything look super orange but we got to a stage in the end where it looked ok, the roughness maps had a little bit of adjusting but not as much as the albedo did. The record was also a fun part to work on as creating the bands of value on the roughness really brought out the look of the vinyl disc.


Lighting screenshot in ue4
Lighting was a tedious process and required loads of tweaking, our scene was an especially difficult one to replicate it turns out because the light through the fabric was hard to predict and make it match the original shot. We also had to add more blue lighting behind the camera to pull back some of the oranges and we also applied some colour grading which fixed some of the issues. Next we got the emissive texture working for the disco ball and the globe which contributed to the scene's lighting and picked out the front of the record player like in the original shot. After a load more tweaks and adjustments we ended up with our final scene:

Final shot comparison with movie still


I am happy with what we achieved on this project, our final scene matches mostly with the actual shot and we managed to fit in a lot of detail but there are a lot of things that could be done better. I think part of the problem lies with the difficulty of replicating the lighting in this scene but also I would say more back and forth between with the shot image and the scene at every stage might have helped the accuracy as the placement of objects are a little off.

If I were to do anything differently I would say that I would try and get stuff into engine earlier so there is more time for light tweaking, we thought we had enough time allocated for lighting but on reconsideration I think even more time should have been allowed for it or even doing a lighting pass before the texturing with values to see if its accurate before going into details.

I feel like I have learnt a load more over the past few weeks compared to a solo project just because of the group factor, it has been an intense but rewarding experience this project and I am more confident in a group role as a result. The next project is a solo one (Sentry Turrets ftw) but I feel like I can take the project management and time keeping tricks I learned over this project and apply it to any others. Hopefully I can keep to it this time. ;D

Sunday, 19 October 2014

Cool Shit! - Tomorrowland trailer, Feast, and art wall update

Cool shit again! This week we had Mike Kelly do a presentation about making sure to feed our minds with awesome-y things so we can go and create better art; so what he did was show us a handful of links and places which get those creative juices going... he started with one of my favourite websites ever: Pinterest.

My Pinterest Page - Yeh, shameless self promotion I know
Pinterest I've been using since college now and it's the first place I go to fill my brain with all sorts of cool imagery. The great thing about it is that the content is all prefiltered and organized by the people that use the site, allowing you to easily find the kind of images you are looking for rather than spending loads of time sifting through google image search. The newest feature that brings up similar searches with tags is also extremely helpful as it allows you to really narrow down the search to the exact kind of image you are looking for without having to know what keywords to type. You can tell I like it because of my 3000 and counting pin count. :D

Mike went through a couple of other links aswell as suggesting that we should 'possibly' donate our bodies to science so we can go to places like Cambodia and Malaysia to get some travelling under our belt (and a truckload of reference photos) if our time last year in the abandoned warehouse meant anything it's that I'm willing to go to extreme lengths to get reference photos.

An abandoned, falling apart building.. this room used to house a karate class as there are a lot of posters and stuff on the walls as well as hooks for punch bags.
Speaking of trespassing, Mike also mentioned a site that features a lot of urban exploration picture among loads of othr things. Will be making more visits to from now.

Ok my turn for cool shit. This week the teaser for Disney's Tomorrowland came out and it looks INSANE. Yeh basically sci-fi stuff again but for anybody that knows me it's not surprising that I find this to be cool shit. The way the scene transitions is really sweet, you will see what I mean by watching. :)

Next up, another Disney thing. This time though it's the teaser for the short film 'Feast' by Disney Animation Studios who are again experimenting with binding together traditional media and cgi just like they were trying with their last short 'Paperman', here the result is different but still one hundred percent Disney. There is a really cool cel-shaded look to this that instantly reminded me of Wind Waker and similar games, I really hope Disney uses this style eventually in their feature length movies. 

And to finish this post: Another Cool Wall Update!