Wednesday, 1 June 2016

FMP: Post Mortem

In 20 weeks I set out to create 3 vehicles.

An accurate study of a 1963 Corvette Stingray, a 1940s COE Truck and a vehicle of my design.

I completed all 3 of these tasks within the tri budgets and texture budgets.

I believe that my skills clearly develop throughout this FMP and I am now a much stronger vehicle artist.

 1963 Corvette Stingray

Choosing a vehicle, purchasing a model and sculpting in the initial forms.

This project was my first foray into Sub D work flow so there was allot to learn.  After getting to a point where I was happy with the Shell of the car I could begin to cut out various panels. I could then begin to add bumpers etc. 

I now know that many forms should not be modelled during Sub-Division, It can be allot easier, more efficient and it can yield better results to wait unto after Sub-D to model in many of the smaller forms. Sub-D should just be used for the initial sculpted shell of the vehicle. During this project I did not know this and I tried to model the entire vehicle whilst Turbo smoothing. This held me up greatly when it came to tidying up the vehicle and general retopologising

Further dressing of the vehicle

Throughout this process I am altering many elements of the mesh to fix shading errors.

Once I had finished the car and it was shading correctly I then began texturing. My mentor did some quick renders of the car with a highly reflective surface to point out some reflection errors at this point. At the time I wasn’t aware of how dramatically Brocken reflection could be on a mesh that appears to shade correctly. This was the hardest part of the project. Getting accurate reflection requires consistent quaded, neat and flowing geometry. This is difficult to achieve when you have complex forms within which are cut-outs. 

Here I am baking down many areas of the vehicle as well as fixing some of the reflection errors. The last screen shot here shows the many different variations of mesh I went through in my attempts to fix some of these errors. 

Finished Vehicle

I am actually very happy with the final outcome. As my first vehicle project of this type I am very pleased with the forms of this model. I feel that I have accurately captured the lines and proportions of a 1963 corvette Stingray. 

I can see issues with certain creases which a far too sharp. Also I do feel the interior is poorly modelled in that it isn’t as accurate as the exterior. Also the interior is extremely poorly textured. All these issues are due to the unscheduled faze of retopolagising to fix reflection issues. I don’t feel that this was wasted time as this is an area I really needed to build skill and understanding in. I still have a lot to learn in regard to the relation between functioning accurate reflections and geometry, and how to create this kind of geometry.

This issue actually is the cause of other issues with the final vehicle. Although I spent allot of time trying to fix the reflections, areas of the car still have issues. This is why the finale paint texture isn’t glossy realistic car paint. I have deliberately made the car unreflective.

Although I do like the surface quality this has given the final renders.

I have learnt allot during this project. I can now much more accurately capture the forms, lines, proportion and creases of a vehicle. I have learnt how to then create different body panels, model details into these forms, include an interior, fix shading errors via editing normals and geometry, fix some reflection  and This is the first project during which I have used Marmoset.

If I was to do this vehicle again now I would model it differently from the start. I would collapse the Turbo smooth modifier allot sooner and then model in the vents. I would also adjust the mesh as little as possible as to try and keep as much of this original geometry as possible. This should help with getting better reflections. This though may result in a much higher budget car. So in regards to creating a vehicle with the same budget I would have to simply apply less multiplications of Turbo Smooth and maybe forgo some of the cars complex details.

I would also research how to create realistic car paint shaders.

I still have allot to learn in regards to geometry and how this affects lighting and reflections.

1940 Lifted COE Truck

I began this project by buying a model of a 1940s 4 wheel drive military truck. I also studies modified leaf sprung off road vehicles. This was so that I could accurately create a working and period appropriate chassis. Also as I was going to further alter the stance of the vehicle in order to make it much more aggressive, I could now realistically add my own modifications to the chassis and suspension.

This vehicle was very much a mechanical study.

Finished Truck

Again I am happy with this vehicle. I completed what I set out to create.

What has surprised me is how the design doesn’t actually quite work. I believed when choosing this vehicle that it would be intrinsically cool. I didn’t do any design I simply found some good ref for a 1940s American COE truck, this in itself was an awesome characterful vehicle, I then though that lifting it and adding huge off road tyres could only add to its presence. Sadly this hasn’t actually translated into the finished renders. I believe this to be down to the trucks silhouette. It’s extremely simple. With game art assets we are used to extreme overly designed vehicles, even though this truck would look awesome in real life when standing buy it. As a render of a digital model your comparisons are with the extremes of the gaming world, and it falls short.

I am happy with modelling and texturing. I believe during this project I developed allot in substance painter. I believe the wheels in the renders are to a very good quality. This is down to creating a high quality low poly, which I made in Z-Brush and then Decimated. The high poly was then baked onto this. This in conjunction with the average normals work flow has created what I believe to be a high quality tyre

In regards to the presence of the vehicle I believe the inclusion of a floor plain could have really helped. The truck actual has allot of rake and this is hard to see in the render because of this.

If I was to do this vehicle again as well as the alterations I have already suggested I would physically model the doors. Physical geometry due to casting dynamic shadows/lighting etc, will always look better than normaling the information. Also I would model damage into the mesh. This is something which I will definitely take forward into future projects. Wear and damage up to this point I have always added during texturing, this doesn’t have to be and shouldn’t be the case.

This vehicle could also do with Decals/Stickers/graffiti to add character and a personal touch. This would also add colour/tonal variation.

Although the bed of the truck is missing this area could still be used to add character to the vehicle. I could have added personal belongings of the owner/ suggestions of its use and or the environment that the truck exists in. All these features are not on my truck because I never did a design pass, I simply recreated my ref.

Now that I have finished my FMP I now know that when creating your own vehicle design you have to approach it with the same incredible detail as a character artist. I should have asked some core fundamental questions what, where, when, who, why and then my design should be able answer these. This would create a vehicle with character and depth. 

Designed Vehicle

This is my 3rd and final vehicle. I have already learnt allot at this point and I believe this is evident.

This vehicle needed designing. I began by collecting a vast amount of reference. I had no prior perceptions of what this vehicle should be so this was anything that I was drawn to.

I then overlaid these images and then drew over the top. The example of this above is mostly many overlaid images of engines. I found this to be occasionally extremely successful and I would quickly create a few designs that I would want to take further. Sometimes this method would produce no real results.

I quickly found that I was drawn to a particular theme. This was to adhere huge amounts of complex, mechanical, ambiguous forms to a relatable streamlined vehicular form. Thus creating something extremely complex, relatable with purpose and yet alien.

After a period of time designing I decided to try designing in 3D space, in 3DS-Max. I picked my most successful design at the time and began to model and try to make sense of my extremely vague designs. I had begun this during the truck project, after putting time into this vehicle I felt like it was working and simply made it the finale design. This was going to be the finale vehicle of my FMP.

I began by making the cab of the vehicle as this was the most important form. This immediately appeared to be a really successful design.

I quickly decided that it would be important to ground the vehicle by attaching it to a real life chassi. I then put allot of work into researching 1930 hot rod chassis. I was then going to build the car on top of this. I really liked the idea of this hyper futuristic tech being bolted to this ancient 1920/30 tech. This is actually exactly what hot rod culture is doing today. It’s common with modern day Hot Rods to have the exact underpinnings as they would have had 60 years ago the main change will be a modern power train with modern induction and modern paint. My design is simply this idea but in the distant future.

I was really happy with the design at this point. But I was having a lot of trouble with the back of the car. I knew I needed something here I just didn’t know what and how much. 

It was at this point that I realised the relation between good vehicle design and character design. But I was well into this project so I couldn’t push the design too much and or get too detailed with my ideas. This though did at least make me think creatively when strapping the piece of unknown possibly alien tech to the back of the car.

Looking back now I believe this area could have been pushed allot further using the thought process above.

Now that I had finished the base mesh I had a hell of allot of high polies to make and allot of baking to do. I was determined to bake detail onto every single surface.

I began baking using X-normal and then learnt how to bake in Substance Painter.

Above you can see the extent of my baking.

Now that I know how to bake complex meshes in substance painter I could complete this process much faster. This also means that I could also put more care and attention into the high poly details and thus create better designs.

This turned into an extremely complex substance file. Again I’ve come along way in substance and I’ve learnt the hard way about its capabilities. I was really trying to push my texturing, for example there are 4 shades of paint on the cab of the vehicle. Obviously the top of the cab is pink and then the underside is blue, but the extremes of the top of both these colours have a sun bleached variant. I did this by manipulating masks that used the world space normals. 

This vehicle is meant to appear as a ramshackle of unrelated and repurposed parts. And I believe for the most part I have achieved this. 

During what was meant to be my last day of texturing for this vehicle my substance file became corrupted. This was possibly as a result of the sheer size of it. I now know that I should have Brocken the vehicle up into areas of similar texture quality. This though was my first corrupted file of this degree (not including Unreal files).

Because of this I lost a day of progress and I also lost a day trying to recover files. This all happened in the last week. But I still managed to complete the vehicle, this if anything only altered when I would begin this post mortem. 

Finished Vehicle

I am extremely happy with this vehicle. I am mostly pleased the design of the cabs form and it’s texturing.

I’m also surprised with the success of the colours, I felt that I was taking a risk; initially I planned to just use dark moody metal tones with maybe some colour highlights. But I ended up using two extremely bright and playful colours on a large amount of the vehicle. I believe this may differentiate this vehicle from allot of vehicle art, it also grounds the vehicle as being domestic. It could have easily have ended up being some kind of military/dark underground cliché themed death car. I believe good design doesn’t need to have guns and or kill people to be cool. Because of this I did try and stay away from obvious use of weapons during the design process.

Again I believe this vehicle is in need of a floor plain for the same reasons as before. It could also do with more stickers etc on the cab.

If I was to do this vehicle again I would more solidly lay down some of the designs before I commit to them. The interior was made up on the spot and this was the first iteration of it. As a result it suffers. Some of the features of it are only thought through in the texture and aren’t present in the geometry. The area behind the seat is meant to be soft rubber yet it has a hard edge due to this. This is a case of designing on the fly, this method shouldn’t be planned in as it will never yield the most successful thought-out results.

I’m at a point now where I feel I have a vehicle art portfolio that is ready for applications. Although I can see issues with all of them I believe this will always be the case and if I keep waiting for a project to be “finished” I may never start applying for jobs.

Friday, 27 May 2016

FMP: Baking

 Here I'm just showing a hand full of the high polys that I made during the baking part of this finale vehicle. 

Initially I was baking using X-Normal. Because of this I was detaching small sections of the vehicle and working into these.

I then learnt how to bake in Substance Painter. Baking multiple object simultaneously is extremely simple when using substance. Because of this I began baking mulitiple sections of mesh at once.

Below are 3 examples of the grouping I was then baking. The only real limitation using this method is the poly count when creating the high polies in 3ds Max. The below meshes have extremely high tri counts.

I much prefered this method of baking. It's much more efficient. It's also better for design as you are now modeling in detail for multiple areas at once. 

This method does slightly alter the proces. As the geometry your baking now is allot more complex simply cutting out these areas manually before compositing in photoshop isn't really an option. This though is easily fixed by saving out the unwrap for the area your working on, and then using this to quickly fined the UV shells to then cut out.  This does now mean that you will have a file with many various UVW maps.

Another negative is that due to the poor export quality of Substance Painter exports the highest resolution map you could use is 2k. Substance painter will export at 4k but the actual quality of that export will be closer to 2k, this is why I wouldn't work at 4k with substance. X normal can export at any size so your not restricted in this way. 

Wednesday, 25 May 2016

FMP:Substance Crash

My substance painter file got corrupted.

 I only lost about a day’s work but then I spent all of today (Wednesday) trying to get the work back.

The substance painter file as of this morning was completely unusable. I can still rotate the model but that is all. I then tried to export all the textures, at least then I would get the work up to this point. This caused substance painter to crash, but without any visual prompts. Nearly all the crashes that I have experienced today (Which are the first for this project) were impossible to distinguish from an extremely long loading time. Only upon attempting to cancel the previous function would substance freeze and become completely unusable. This means that I had to wait long periods of time between every single input just encase it hadn’t actually crashed, more often than not it had.

At this point I feared that I had lost up to 3 days work as that is when I last backed up. With today and tomorrow being the last full days this would be disastrous.

I have spent the entirety of today trying to export as many textures as possible. After dozens of crashes and exporting some textures individually to reduce the load on substance I have managed to save some textures.

 I managed to export all the textures for Sub Object 1, except the opacity. I have also exported all the Textures for Sub Object 3. I haven’t managed to export any textures for Sub Object 2. I can’t even view any layers related to this ID they have completely Brocken.

This means that I have lost up to 3 days work on the engine parts of the car. This is what I will be doing tomorrow.

 My plan this morning was to do a dust pass on the chassi and tyres, then I was going to erase streaks where fast moving particles have removed dust. Then I would have made slight adjustment to the whole car in substance and photo shop.

I will be doing this tomorrow as well as doing an old oil pass on the engine, new fresh oil pass, then dust and finally then erasing streaks out of this engine dust

This is where the vehicle was at Tuesday Evening.

1 day to go

Friday, 20 May 2016

FMP: Replying to mentor feedback and Answering questions

Response to feedback on the rear tyres within which changing to a road tyre influenced design was advised. 

Thanks for the feedback

The tyres are based off a motor sport mud tire, as seen on the Batman Tumbler. I simply thought they looked cool, were a real life reference that proportionately fit and might have fit with the gnarly, beat up home built aesthetic I was going for.

A more road sport orientated tyre may have better described speed, fitted the cars stance and further pushed the future aspect of the vehicle. This is very true.

Currently the finale deadline is Friday the 27th, so I have 

exactly a week including today. For now I am going to carry on texturing in substance. Post deadline I might be able to remodel a high poly rear tyre and re-bake.

Would it be possible to ask what the final poly limit became in the end, did you stick to your budget?

The budgets I set in my brief for this vehicle were simply the highest budgets I could get away with as we were told that we would not be marked down for undercutting our budgets but would for the opposite.

The briefs budget for this vehicle was 100k tries and 2x4k texture sets.

The car is currently at 90k tries, so I am well within the budget. I am using 3 x 2048 texture sets. These are Broken up into the following sub ID’s.

I’m using 3x2048 sets as this a manageable amount of work within the time, it is also a much more realistic Game ready budget, also I’m texturing the vehicle in substance painter. When Exporting from substance I have found that there is a huge loss in texture quality, because of this I always export from substance at least double the finale texture size I intend to use. I will then rescale the textures in Photoshop. 4K is the largest texture size that Substance will export, so 2k is the largest usable texture. 

Would there be any areas you’d like to reduce detail so you can add detail in other areas?

The interior is the main area where allot more geometry could be added. The interior was never going to be a focus. In hindsight I modelled the vehicle as if it was going to have side windows therefore concealing the interior. I then made the cab open, but I never really added much more detail. I may go back in and add detail to at least the top of the head rest as its silhouette is clearly visible.

The mesh that I’m using to join the exhaust manifolds to the engine block needs additional geometry. 
Also the roof line could have done with less geometry cut out when I was retopologizing the initial turbo smoothed high poly mesh. Also when retopologizing this mesh I tried not to create many tries, and general not do any drastic alteration from the turbo smooth. This is in an attempt to keep flowing geometry and not break shading/reflection. As a result there are areas where geometry lines meat creating points of relatively high poly density. I don’t know whether this can be avoided when creating complex curved forms.

Thanks for the feedback 

Tuesday, 17 May 2016

FMP: Lots of baking

I've baked down every single part of the car. I made shore to add some thing to everything. A bolt a seam a vent etc. I can still sea some issues here, but I realy once again must move on.

Towards the end of baking I learnt how to bake in substance painter. This meant that I could easily bake multiple complex forms at once. This is much more efficient.

I have spent far to long baking but i'm about to start texturing in substance, so the more work I put in here the greater detail I can get out of the smart materials. This will result in a much high quality finale vehicle.

I still have 10 days to texture which is still a reasonable amount. Thus my FMP brief will still be easily met. I can still polish all the vehicles before degree show. 

I will now finally start adding materials and detail to this vehicle.