Friday 23 May 2014

Dissertation Summary

The end to the dissertation has arrived and it's definitely been a journey, that of which I'm pleased to say that I've learnt a lot of new things on the way. One of the most enjoyable aspects of the whole production process was in fact the researching of the history of ancient Chinese architecture. I have always had a personal interest in ancient Chinese architecture, however I never had the opportunity in which I could use the research for a specific purpose.

In particular, I enjoyed the fact that the carpenters of the ancient Chinese world were able to construct massively complex structures using only a small range of simple assets, such as 'Dous' and 'Gongs' hence the name, the Dougong system. Furthermore, I also discovered that the system was so successful at withstanding the test of time due to the fact that the downward force of the higher tier frameworks are dissipated equally throughout out the whole construction system.

I liked the challenge of my project, however at the same time I felt as though I had over scoped the amount of time it would take to complete a working system of my own. Because it took so long to complete this system I therefore had to rush other processes such as texturing, unwrapping and testing. In addition to this, I failed to model the amount of objects that I had originally intended to create. From this, I also was unable to create a set of Normal maps for the objects I initially planned this mainly being the roof tiling texture.

While I struggled to complete my model within my self set time period, I however managed to achieve a primary goal that I had set during my proposal of completing my whole kit within 45,000 polygons. After the completion of the modelling process my total polygon count was 25,043. Although I am pleased with the fact that my polygon count was lower than anticipated, I feel as though it would have been better to make use of the full amount of proposed polygons. On the other hand, if this was the case, I believe that I would have been less efficient at completing other goals and therefore completed overall even less tasks.

At the end of the whole process of my dissertation I found that the gathering of the final assets within UDK produced a large set of unforeseen issues. For one, my mesh instances did not work correctly on all of the assets due to the fact that UDK does not support more than one base material per item. Although 3dsMax did produce the result I first desired for my final outcome, My main goal was to produce the assets and place them within UDK. Another error during this process was that the objects origin point was not centered and I completely ran out of time so I was unable to fix this.

On completing my research, I was able to answer my overall question around my dissertation, which was to see if complex structures could be made and constructed from modular assets for use in a game engine. I would respond with that they can indeed be produced using this method however it requires a lot of measuring and preparation beforehand to produce assets with such precise and accurate parameters.

Adding the models to UDK

Here I will be exporting my Chinese village kit of models and placing them within UDK to create a small village. I first then exported each individual object from 3dsMax as an ASE file extension. Overall there were 47 files to export.

Adding all the static meshes wasn't difficult as I put them all within the same grouping of 'Meshes.' Next the materials were to be imported. At this point I like to add one material at a time so I can tell when I have a problem. Because I was using PNG's I instantly ran into a problem with my texture diffuse files. The texture came out in a variety of multicoloured yet some would come out perfect. I consulted our lecturer Chris Janes and he challenged me if I had given the page an alpha channel as within UDK if a PNG is used it needs one. To make sure it does, the texture needs to have the wasted areas either white or transparent. So at this point I went through all the diffuse maps and changed them so that they all had an alpha channel which then resolved the issue.





















The same problem also occurred with the specular maps and so those also needed to be edited. I then had to assign the textures to their model counter parts. At this point I realised my earlier mistake of giving them mostly all similar names, this proved to cause a few organisation problems when assigning textures.





























Here is a problem that I only encountered within UDK, within 3dsMax the above model uses separate texture files and is able to do this without any flaws, however within UDK the program does not allow the the user to use multiple texture maps for a single object, for example 2 diffuse maps per object. In the object above it uses the material 'Courtyardwallmaster2' but it also contains several mesh instances from another model. This seemed fine in 3dsMax as it allowed the texturing of both instances on the model however UDK does not. At this point I did not have the time to improve and fix this issue, however if I did have some extra time at the end of the project, this would be an issue that I would try to have overcome if possible.

In addition to this, whilst manipulating the models within UDK, I immediately found that the pivot points of each individual object did not retain the settings I originally hoped for. This was a problem that I would have also hoped to change however, I had 47 individual models to alter and thus I didn't have time to make the changes.



















Subsequently, this caused the above problem to occur. I believe that this was the reason why the kit within UDK did not function properly, in that I struggled to fit some of the pieces together and had to keep adjusting the scale, this was an issue with all the pieces that within the original design were not set on a factor of 16.










Thursday 22 May 2014

A look at the kit...

Here's a quick look at the kit in all it's glory in the perspective mode of 3dsMax.


As you can see, all of the objects are organised and textured appropriately in order to aid quick level production, each of the individual building s re arranged by size so that a level designer can select and grab multiple objects at once.

Diffuse Texturing Process Completed

I have now completed the diffuse texturing process and have still to do a set of specular maps for the models. As of now here is a quick screenshot of a UVW unwrap and it textured fully with a diffuse.


Here you can see I've put emphasis on the grain of the wood, the majority of the objects are placed along similar grains to save time during this process of texturing.


Here you can also see the texture of the tiles that I've begun to complete. The detailed dragon was taken off of a picture of an already existing Chinese tile, which can be seen here...


http://www.123rf.com/photo_5525979_chinese-temple-in-the-summer-palace.html

I changed the colour values to match the tone of roof I desired for my models, which was a swampy green. This was done with the colourise function within Adobe Photoshop.




Unfortunately I don't enough time to complete a set of normal maps for the models as I've run out of time for this process which I really wanted to complete. It was a goal I set myself If I had more time but unfortunately I can't.

Some of the grass pieces I used seem less textures which I also found helpful online, they were also free to use and can be found here...

http://seamless-pixels.blogspot.co.uk/2012/09/free-seamless-ground-textures.html


Wednesday 21 May 2014

UVW unwrapped and texturing process begun

I'm pleased to announce that I UVW unwrapped all of my models earlier utilising instances where possible. Some examples of the unwraps can be seen below. I often struggle with unwrapping but I seem to have coped okay with the overall amount of objects I had to unwrap. I have given certain UV islands more space if they were used frequently within the model or were at the players head height. This was to reduce stretching of the pixels that is obvious to the player and could be seen from the players point of view.


This UV map is the whole unwrap of the main building excluding the roof. This is also both double building and triple buildings. It's tried my best to pack the UV islands so that there is the as least possible texture space wasted. The unwrap is a 1024 x 1024 render as this was what was stated on my proposal for my maximum unwrap dimension size. 


Here you can see the full scale of the set of models I have made. The blue indicates the model pieces being fully unwrapped. The black area is the part that was being currently unwrapped when this image was taken.

From this point I thin started to create the diffuse texture for the models. I began on the part of the structure that was most important to the dissertation, so I began with texturing the building, a piece in particular was a Dou piece that had been instanced nearly over 100 times throughout the model kit.


To get this wood texture I used a hardwood texture online and then multiplied it on a separate layer to give the textured layer on top of a base red colour. The texture I used can be found here and was posted and has a public domain license.



http://www.publicdomainpictures.net/view-image.php?image=25761&picture=wood-texture&large=1


From this point I will just keep working on my texturing process, until completion.


Kratochvil, Petr . (N/A). Wood Texture. Available: http://www.publicdomainpictures.net/view-image.php?image=25761&picture=wood-texture. Last accessed 20th May 2014.



Thursday 15 May 2014

UVW unwrapping started + Preparation techniques

Just a quick update, I have prepared the models for unwrapping, as a solution to the many pieces I have to unwrap that are similar I have developed a simple solution. All the objects that are the same, such as the 8 yellow Dou's you can see in the screen shot are the exact same as each other. My plan is to now delete all of those that are copied and create a 'Master' clone. From this I will then copy the master clone creating instances and replacing the deleted clones with the replacement instances. This will then mean that I would only have to unwrap the single 'Master' Dou and it shall unwrap all of the copied instances. Although the textures will come out the same, the materials of what they are made of (red stained wood) will be the same and that's what I'm going for.



From the image above, you can clearly see the 'master' Dou for this model, unwrapping this piece would unwrap all of the yellow ones. So this method should hopefully save me some extra time whilst unwrapping. I've never done this method or thought of doing this method previously to this but I'm hoping it can shave some production time off of my creation process.

To show clearly what I'm talking about I have a quick demonstration below, I'm going to refer to this as the 'master and slave' technique. Here you can see the whole main model with the master and slave technique. The master pieces are in black and the slave pieces are in bright green.



From this you can see how much less unwrapping I have to do, this will also be the same fore texturing also. Let's have a closer look at a whole set of brackets.

Black pieces = Master | Green Pieces = Slave

This side shows a lot of master pieces in use, however I tried to keep the master pieces to all one side of the model so I can keep them all grouped to ease the unwrap process. Here below you can now also see the other side of the model, this can really show the amount of reduction to the unwrapping stage this could potentially do, green being the parts I wont have to unwrap again, only once with the master piece.



For a quick demonstration on a single master piece of how it will work. Here is a small 'Dou' with it's master being affected in some way, as you can see it affects all instances within the scene.



Honestly I wish I though of this much, much sooner into the production process as it would of saved a lot of time, but oh well. We learn and must soldier on...

Thursday 8 May 2014

Resetting my Xforms and finishing the modelling process.

Over the last few weeks I have noticed these white bounding boxes around my models becoming slanted and I was concerned if they would affect the final outcome of my project. I did some searching online to find out what the issue was and according to a forum post on the CGArchitect.com website, the issue is with the bounding box. The bounding box coordinates have shifted and have turned on their side, and seem to have warped completely out of shape, this can be seen below.


I did some further investigation on the Autodesk 3dsMax help document on their website and it states that.

"The bounding box is the smallest box that encloses the maximum dimensions or extents of an object. A bounding box appears around selected objects in non-Wireframe viewports."

I also then found out on the CGArchitect forum that to reset the bounding box, all I had to do was reset the objects Xform.


And now you can see the bounding box fits parallel to the ground now.



In the picture above you can see a cross section of a small roof piece, here I am applying a symmetry modifier as I was advised by a friend of mine that I could cut down polygons by using this modifier and then merging it at the seem on the roof tiles. This would also remove any unseen anomalies within the structures dimensions.

A technique I've been using myself is that when a model is ready for UVW unwrapping, I will change their whole model to a certain colour, in this instance for the walls I'm using green.



Here you can see this and how close I am to unwrapping, the yellow model is the current model I'm preparing and So after this post I will begin the unwrapping procedure. Which is something I can't wait to begin.

AutoDesk. (N/A). Bounding Box. Available: http://docs.autodesk.com/3DSMAX/15/ENU/3ds-Max-Help/index.html?url=files/GUID-4F9C1D31-E5EB-4641-94C9-5C1988178AC4.htm,topicNumber=d30e600845. Last accessed 8th May 2014.

Wildman, Peter. (2010). How do u reset the bounding box to align with the object?. Available: http://forums.cgarchitect.com/39227-how-do-u-reset-bounding-box-align-object.html. Last accessed 7th May 2014.

Thursday 24 April 2014

Last minute issues

I'm so close to finishing the modelling process now and I have been checking over the models for their scales and I have found a small issue that will need addressing now before I go onto the unwrapping and testing stage. I have tried to keep all objects within a 256x256 pixel block so they are functional modular pieces. However I have just noticed that the wall pieces are not flat on the factor of 16 grid plane. A small print screen here shows this.


























Here I have highlighted the issue with the walls each grid space is 16 units and is the scale I've used for the whole model space. The red line shows the centre point of the current wall piece, and the blue line is where the centre point should be for this to work.

The solution for this could be that the wall piece's pivot point be put on a factor of 16 which can be seen below. The green cross marks a suitable location for the pivot, this should hopefully also work when the wall is then rotated back around 180 degrees.








Saturday 19 April 2014

Nearing completion of modelling process

The modelling process is nearing completion and i'm nearing the unwrapping section. I am feeling that the process is going well, although I don't have much time left to complete the whole dissertation, the models and their structures are as modular as I can make them.

The rule that I have had to keep by is that each objects parameters must be a factor of 16 or at least the centre of the vertices must end on a multiple of 16. This has come in handy as I set the scale factor in UDK to 16 so that the pieces literally snap together. From this I have also found issues with certain elements.

Here you can see an error with the building connector element.
This is a small screenshot from UDK. Here you can see that there was an error in the model calculations. However when I checked the model dimensions the maths was correct and also the models vertices were each on the perfect spot. This took a few hours to figure out, but in fact it was just a 3dsMax glitch and after re importing the model from a separate file, the error disappeared.

Here is the corner piece in UDK still showing the error
Now, the piece is corrected and all the walls join seamlessly. The corner piece of the wall is also finished after an extensive amount of maths. I must admit that during this whole dissertation, I have found that the extensive and handy use of post-it notes during modelling something so complex in maths a big help. Here is a small note on the maths for the corner piece on it's own.


On the left is a few notes on the construction of the corner piece and to the right a list of factors of 16

As an outcome, here is a print-screen of the final corner piece in a position it could be in during the level editing phase.


In this picture you can also see the scale of the 3 piece building compared to the original mock up model (in red). If anyone could remember from the first mock up, the size of the building was so small that it felt too cramped to walk around as a game character.

Here a print screen of the whole models in the kit so far, there are some models which are there for reference and will be remove before the end. At the top it says there over 41k polygons in the scene, although this total is getting close to the  maximum of 45k polygons, the mock model itself is 12k polys and will be removed soon thus would bring my total down a lot more.






Tuesday 8 April 2014

Modelling update

Hello there! Just a quick modelling update on what I've been doing. I know I should have completed the modelling stage by now, but I simply haven't yet due to the complexity of how each individual model must sit right and be accurate in terms of scale., but I'll update everyone on my current progress.

Right now I'm working on making modular pieces of wall. These pieces are to fit with every opportunity that the level designer needs and so needed a lot of work, but the whole modelling process is nearing completion.


Here is the workings out of the production of a corner wall piece as well as the variety of walls I've already created. They do look very similar, but each individual one has slight variations to counter for different scenarios.

I've also been working on a system of small blocks that will raise up the whole building in a very similar fashion to how the residential are constructed.


Here you can also see a set of steps that I've produced to help the player up the incline. This will have a sloped gradient custom collision box to give the player a smooth transition when moving between gradients and heights.



Wednesday 19 March 2014

Roofing and general modelling update






Hey everyone, upon designing and making the roof for my models I have discovered something incredible, some genius on the internet has designed and made custom Lego pieces of ancient Chinese roof tiles and are selling them online. It uses very simple pieces which allows a Lego master builder to make any Chinese building roof style they want. The blog post can be found here:

http://thebrickblogger.com/2012/06/custom-lego-roof-tiles-by-bricktw/



This just goes to show how a complex roof system can be split down into small significant chunks, to become even modular of course.


Although this proves that the roof system can be made simply modular, the amount of pieces it would create for the people using my model kit could be quite large in the number of items needed, but it's definitely an option. I also have to point out that the roof style is that of a religious building or even a building of importance, not that of a residential dwelling.

There is also pieces that can be used to make large pagoda like rooftops, these are all well and good and effectively can create a modular system for the Lego building play set, but they don't seem to take into account the complex building structure beneath it, the Dou Gong that I am interested in replicating.


And to finish off, I want to show you my progress towards my own models. Here I have completed the segments of roofing that are required for the large building.



I based the roof structure off of my latest blog post and quick mock up I made in photoshop, he is a small comparison so you can see it quickly: 

Model structure


Model design

Another object I've been working on but forgot to show in my previous blog posts was the walls I've designed for the structures of this size. Here you can see 2 sets of door frames which can go between the middle of two extender sections or in the middle of an extension and an end cap section. There are also basic wall sections for the same sizes and for one other instance of two small sections. I would like to hopefully work on these wall pieces with normal maps as their basic flat texture will look very odd as the player would be running past them all the time.


 Here is a small image of the sort of building I want to be able to achieve at the end of the dissertation with all the different pieces, I hope.


Thursday 13 March 2014

My timeline and why it needs rearrangement

My timeline has become so back dated that it's desperately in need of rearrangement so I plan to now do this as it will help me over the next coming 2 months of this dissertation. It has become so out of date and I've gotten so far behind due to other modules that it is slowing my progress with irrelevant dates and progress milestones. Here below is my current timeline: 
Current Timeline

                                                                                                                                  
17th February – Continue modelling if needed and start UVW unwrapping on each asset ready for texturing. (May take longer as not as experienced in UVW unwrapping, although research into quicker uvw mapping will be undertaken over the course of the project.)

10th March – Experiment with Mudbox/Zbrush and the NvidiaNormal filter in Photoshop in texturing the assets with diffuse specular and Normals.
                                                                                                                                  
24th March – Import and construct the Village within UDK with textures and assets.

7th April – touch up and add more details to models where needed or where some models issue are found. If no issues are found or are ironed out and total polygon count allows (45k polygons). Create a couple more assets to add a more realistic atmosphere to the scene such as trees, pots and tiles.

 7th April - 15th May – Work on finalising dissertation

 17th May – 16th May to be the absolute deadline of handing in dissertation.


I'm currently just over 4 weeks behind on my personal milestones, which isn't good on my half. The yellow line is where I should be at this current point and the green line is where I currently am. I'm still in the process of creating my 3D assets. One one set of buildings are done, i'm hoping I should easily be able to replicate these models and just adjust them to create the rest of the other models on the asset list. But because the timeline is so out of date i'm going to propose a new timeline for myself to keep to as due to the other modules and workload of the 3rd year, I just simply couldn't cope with it all and so it needs to be changed to help manage my time left effectively. This new timeline will be the date in which the object must be completed before.

New Timeline

20th March - Finish all of the models needed for the construction of the buildings.

3rd April - UVW unwrap all models that have been created and get them prepared for diffuse mapping.

17th April - Create and supply all textures with a diffuse texture, Normal maps will need to be made for the roof and if time is spare start creation of Normal maps on the woodwork and walls.

1st May - Import and construct the Village within UDK with textures and assets. Test and arrange houses similar to some of the previous designs.

1st-15th May - I will be keeping this time for any overhanging production for models and objectives. I will also leave this time for myself to create more blog posts as to describe how the dissertation is going and also to organise my models and the final environment, adding in some environment effects and even organise my level pack in a organise manner ready to be shown to people.

Although I am many weeks behind, I still have made a lot of progress overall. Other modules have been getting on top of me especially my Design masterclass, Our group has even volunteered to help my other colleague with Dare to be Digital to help him with his grades although we wont be getting marked on it. I know this is only small amounts of work, but it all seems to add up pretty quickly as I've recently seen.

But as a general catch up to my 3D modelling work state, the first main building is nearing completion, all that is needed to be constructed is the roof and the corner strut to create the corner. When this is complete I can take this structure and simply copy how it's created and apply the same principles to the other models.


Friday 7 March 2014

Making progress

It's been a while since i've been blog posting, but I have made some progress on a personal scale. I've been confused with the way in which I can model the walls of my structures whilst still keeping the structure architecture fully visible and not covered up by the external walls. I've also been trying to figure out how to keep the models modular whilst minimising or getting rid of all mesh clipping. This is when two planes of a model are on or are very close to the same position and the shader has a hard time choosing which face to render and instead it renders both polys and shows an odd jaggered texture. This isn't good when modelling, especially on a part of the model that can be seen by the player, I needed to avoid this at all costs to have a decent looking final product. Here is an example of mesh clipping.


So for the last few days i've been working on wall that will fit in with the columns, the walls need to be of a variety of size length ways to counter for nearly every situation that the level designer needs to offer. Here you can see different size of wall being used to make up the modular components. It also shows off the structural composition which is a big goal of this dissertation. 


I have also begun work on the end caps for the structure to aid me with my wall construction, they are not at their final stage yet but i'm debating whether to cut them up into separate segments to aid more so with the modularity of the kit.

So this is what i'm currently working on and should be finished with it soon, once these are complete the hard task of UV unwrapping each piece begins, a hard an grueling task, but a task that I look forward to starting so that I can be one step closer to completion of this modular kit and experimenting can then begin.

Monday 24 February 2014

Breaking the habit

I'm back again to continue the working on my dissertation, but I've had reason as to why I've had so much time off. Over the course of the last 2 months a couple of other projects did arise on my part and have been a burden of work and it has held me back massively. One of the projects is of dare to be digital, which doesn't require that much time however there is another project that I've become a part of that has taken a lot of my time up and that is Design Masterclass. I've been asked to work on an up and coming iOS/Android game and this position is taking up a lot of my spare time to get to my producers requirements. However I do plan to now hit the dissertation hard by the fact my first wave of milestones on the design masterclass are complete. I need to quickly get up to speed with my modelling that I already started in January and finish the modelling process. The reason being for my delay is that I discovered that the blog is actually our dissertation as a whole and I do not have to write a separate one to go along side my assets/artifact, this gives me much more freedom as I did in fact leave time at the end of the dissertation milestone to write my dissertation out which I now realise I no longer have to do.

I will be starting where I left off, at creating the large buildings walls and test if these work, if these do, then the rest of the modelling process will be quick and simple as they are in theory just smaller versions of the larger models.

So thats just my catch up plan, you will see more from me soon.

Thursday 23 January 2014

Presentation - self evaluation

Hey there again guys, time for another blog post, this one however will be short and sweet and judging by the title you can easily guess what its about. Well I had my solo dissertation progress presentation not 3 days ago and I thought it went pretty well, It was quite nerve racking leading up to it. I feel that i presented well, better than my average performance when it comes down to pitches and presentations. I feel that i was nervous as i had a lot of content to share with them, over the Christmas holiday i had worked hard on my design aspects and thus had a fair few slides with my digital drawings. All in all i feel it was a successful presentation, one of the lecturers did have a question about why i split the modular assets where i did. And i explained to him that the bet design wouldn't actually work the way he was suggesting. From this i plan to maybe look at the way Dave was wondering about splitting it up.

Sunday 19 January 2014

Current progress leading up to January 20th presentation

I've been working hard on trying to get some assets modelled that I can show on Monday, and I'm making huge progress. I've begun progress on a large building structure that has 3 sections. A end piece with struts, a middle piece with struts and an end piece to finalise the building. In each of the screen shots the small pink square is the size proportion to the player. All of which are simply the structure and does not include the walls as of yet. But hopefully will do soon.


Large building middle section with strut 3,087 polys


Large building end section with strut - 3,087 polys
Large Building End cap piece without struts - 2,859 polys
These are the 3 pieces I've constructed that fit together to create the one building. I plan on giving the player a variety of walls you use to kit out the building. They would include: a door way, a window, a window with blind or just a plain wall.
 
Here is a small screen shot of them together within UDK.
 
Comparison of old mock model to new pieced together model
Pieces individually
It also shows the comparison and changes made since the mock model, the roof has been raise to get a steeper as now I have put in more research I have found that the buildings had steeper roof structures.
 
I have also constructed another asset list to aid myself in creating the small villages, this can be seen below.