Thursday, 26 May 2011

Props For the Level

These are some of the props i made/textured in the level. Any models that have wireframes are the models that i made, anything that is just a presentation sheet and textures, i made the textures for and the models were made by our environmental artist Ceri.


Coal Pile - Modelled and Textured by Artist.



Locker - Modelled by Ceri, Texture by Artist

Coalhopper - Modelled by Ceri, Textured by Artist.


 Modular Hand Railings - Modelled and Textured by Artist



Rock Stair Case - Modelled by Ceri - Retopology and Normal Map creation by Artist. 

Final Renders of Kids



Again this model is a work in progress due to time constraints and A assets needing to be made for the level work. In the time i spent on the texture (6 hours) and the time i spent modelling it (10 hours) i am very pleased with the outcome.

This piece of art actually fit really well to the style guide because the texture isnt over detailed but is still painterly and interesting.

The main issues i have problems with is the face, because the cheek bones are too high for the child making his profile strange.Also without the opacity map in the textures the hair looks very off putting and incorrect, but that is only due to time that it doesn't have one.

My unwrap is definitly the tidiest one i have made as everything is straight and easy to add overlays. Also it is packed fairly well even though there is some space left over. Even though this model is incomplete, it does show a few of the key processes i have learned i have improved in them through out the project.

C Priority Character - Child Civilians

The initial concepts for the kids were to make them fun and interesting, but this idea was scrapped as they were just background NPC's. spending so much time on creating something that will hardly be noticed isn't worth as much time compared to a major NPC.



The modelling for this piece was completed within 10 hours, which is the fastest i have modelled anything in this project so far. After recieving crit on the work, i was informed that 10 hours on a model from a box is a good speed to work at.

Final Renders



Due to spending too long on the model and having assets for the level to build, i had to leave the model in the state it was in. The texture is mostly just base colour, apart from the face and hair. My main mistake here was spending most of my time texturing on the face and hair, the 2 parts on the model are hardly seen due to the hood and face mask.

Compared to the style guide, its in a similar position to my A priority, the textures incorporate details in the diffuse that aren't neccesary or aren't stylised enough to get away with that detail in, e.g. the scarf has needless stitching in, as it doesn't add to the overall model. The normal and specular maps were created using software called CrazyBump. This saved alot of time just so that the model had variation in lighting when put into UDK.

Rendered with the black outline, it does make it fit more into the world created. Going back to the model, in time i created this model to this stand in a much faster time than my A priority so it is clear that ive have been learning and gaining speed from the process i completed before. 

B Priority Character - Concrete Ape Member

After pre production was completed, i picked the most relevant design that would reflect the style of clothing similar to my A Priority character, because this character is a gang member of his gang. 





The base mesh for this character took 2 and a half weeks to create. This time took me past the schedule time allowance, so i had to make up the time in evenings to unwrap and start the texture.

Overall the shape is defined well from the concept. The belt and the trousers being different break up the symmetrical details like the arms which is definitely minimal change compared to the changes i made to my A priority. Some of the details from the concept i didn't put into the final mesh because it wouldn't have added anything, and would have just made the model more detailed and complicated than it needs to be especially for a minor NPC.

The only issues i can see are around the shoulders as they are very tight and have alpha planes around an area that will move quite alot, but due to time i have to just move on and get texturing.

Comparing to the style Guide


Comparing the model to the style guide, its clear that the model isnt far off and actually has retained alot of the styles of a good few of the games in the style guide, mostly from Borderlands and Prince of Persia. This is mostly due to the bold lines that are rendered on the character as well as the painterly textures.

Overall the piece is less stylised than i hoped and that is mostly due to in experience. In the future with my work it would be a good idea to create models to other company styles so that i can get used to working in different ways and also to be more flexible.

Looking back on the model, there are many areas that i would change, and that would mostly be taking out details to aspecially to the boots as the time spent on a piece that is hardly seen isnt worth the time. Either that or make them more prominent in the design.

In the end, i am pleased with how the model turned out because its my first fully complete model that is built to specific specifications and has forfilled them. All the processes and programs learned on this journey have been invaluable and will only help to improve my art in the future.

Industry Comparison



  • Authors work - 10k
  • Nintendo's Ganondorf - 5k
When looking at my work back, lot of the style has been retained but has been lost mostly through over detailing the character and not giving sections to the character that have detail, giving the audience something to focus on.

Nintendo's Ganondorf model shows how it isnt neccesary to have a detailed normal map and alot of geometry to achieve what i achieved with my model. In personality and style both the models are similar which is why im comparing them to each other. 

I didnt need to go through so many problems just to make the model asymmetrical as it shows that if the details are refined and not everywhere on the model that a symmetrical model can be interesting by just adding a few details that breaks up how the eye flows over the model. 

Final Renders

With the model complete, i decided to render the models in the Marmoset Engine, as it renders the models in real time, the same as putting a character in a game engine and running it.




This is a render of the texture and wireframe in the viewport of 3DSMax.

Overall i am pleased with the outcome of the model as it turned out alot better than i thought it would. As much as i spent alot more time than expected on it, all the processes and programs i have learnt in the process are invaluable and will be one of the biggest things i will take away from the time i have spent creating the character.



Texturing

The texturing process was pretty simple for me as alot of my texture work was forfilled through creating an ambient occlusion map of the high poly so that all the details were brought over. With the normal map it made a huge difference.
This was my original diffuse but the colours seemed very unnatural and not even right for stylised work. The colours are also very dull and hardly made the details of the occlusion map stand out.


 With the texture started again, i created base colours that were more natural but had depth to them due to overlaying various colours. (The red skin will be overlayed with a brown colour to make the blood of the skin show through and feel more natural and less pale.




This was the final texture sheet i used for the model.


In total the texture took me, 30 hours to create. This is long for the texturing process, but the problem i had was tackling a texture like a 2D illustration which is where my error was. because it isnt the same at all.

Projection Mapping High to Low


The first normal i created was made in a projection mapping program called xNormal. The problem with this was because my model had complex geometry where some overlapped, i had to split my model into various pieces and back them all separately. The normal map above is a failed attempt because there was alot of artifacting where the projection mapping cage passed through another part of the cage.




This is a combination of 5 different normal maps. They were edited together in Photoshop to create one single detailed normal map for the character.


The bake turned out better than i expected, but as expected i spent longer than the schedule allowed. an extra 10 hours was spent figuring out all the issues and splitting up the model. Again all i can really do is take what ive learned and apply it to future work and refine it.

ZBrush Work


Zbrush has taken me alot longer than i anticipated to get to grips with. This is mostly because of the interface and it being a completely new way for me to create 3D art. Once i got the hang of the basics, it made the process a bit easier but still took alot of time and effort to get this outcome. Initially i was using an optical mouse to sculpt which really wasnt the best option. I moved onto a digital tablet and it made the work flow much faster.

The main issue that made Zbrush slow for me was because i didn't know all the tools that well that are in the program. ZBrush has so many different options and tools to optimise the speed at which you sculpt but i didnt really know any of them and for the time it was taking me already, to take even more time out to learn the program even more was unjustifiable.

Overall the ZBrush sculpt took me 50 hours over 6 days which is far too long to conform to professional work, but has helped improve my skill base set as i now have a much better understanding of what can and cant be sculpted with the geometry in the base mesh. The next step is baking the high poly to the low poly and again this is a new process for me so im expecting it to take me a day or 2 to complete this.

UV Unwrap


Wrapping is usually a process that takes me a week from previous models ive made, but this one took me a day and a half which is still going over time in schedule but is a new personal record.
From reading tutorials on-line, i have head that anything that isn't on symmetry should have its own space in the texture map which using ZBrush as different information will be layed over each other otherwise. This is why certain parts like the hands arent overlayed because they will be different.

THE NEW MODEL


Compared to the old model, the new model has a much more of an interesting sillouette, and has captured the characters bulky structure better. Since i am happy with the form i can now unwrap and export the model into ZBrush (the software i will be using the 3D sculpt the character and add details)

I received some critique from an industry professional that visited during the production.

            "The concept is nice and has been captured in the model well. Improvements would be to make the jacket collar less like a symmetry line down the model. Also bring out the shoulders to accentuate the muscles as when animated they will look smaller." - Elliot Upton (Lionhead Studios Concept Artist) 

Head Changes


The original head, after seeing it with the updated body build made it so thin that it didn't fit the character any more. Edit 1 makes the character look like one of those dense characters that don't actually have any intelligence which didn't portray the character at all.  For the final edit, Using soft selection, i changed the width of the face and neck as well as rotated to the profile to make it feel more neutral. As well as using the extra polygons in the face to add subtle wrinkles to make the character seem older.

Learning about Alphas


Development of the New Model

Using my place holder base mesh, took away and added polygons to define the shape first. I had issues with the jacket because it wasn't in a natural shape and i had no idea how to model a draping jacket. 

Initially i used 3DSMax's cloth modifier to try and create a believable drape, but it failed as the models i were making didn't have enough geometry in them for it too fall correctly and also it was too complicated a shape.


Both my teachers told me to use more reference in my work as nothing can be achieved correctly purely from your head, especially if you want something to look correct.


Initially, the designed changed to the image titled Old from the place holder. I came back to the boots though because after some crit from another member of my teaching group, Aeryn he said:

                 "The boots don't capture enough of the detail in the concept for me to say that they are doing the best job they are. Add some more hard surface modelled geometry, and then put it into ZBrush and add the fine details there, your concept has too much detail for just diffuse to pull it together." - Aeryn James Davies. 


From the crit, it clearly necessary that the detail i want to achieve can only be achieved well if i take the model into ZBrush, but this i am worried about because i hardly know the program at all. This will help forfill professional practice as well which is definitely useful.