Overview and initial decisions
After a long summer period spent working on personal projects and learning new software packages, I was itching to start back at university and get stuck into my first major project.
Initially, I started with the environment brief to make a laboratory type scene. My take on it was to have the lab built into the cabin of an aircraft, however I later decided to change to the vehicle brief as quite a few of my peers had picked the lab environment and I wanted to do something different.
I didn’t want to waste any more time, so I carefully considered my other options and decided on making the hind attack helicopter. I felt this project would push me to make something bigger and more complex than I had ever made before, forcing me to work faster, develop new techniques and improve. I picked the project for the very reason that it was something outside of my comfort zone. I was already thinking of the challenges that different parts of the model would present.
I approached the project with proper planning as this had been my downfall during projects in the 2nd year. I decided what kind of helicopter I wanted to make, and with regards to hinds, there are a lot. From my research it appears that the Hind has been in production since the late Soviet period up until today. All this time means a lot of different versions.
I felt that the best bet would be the most popular model, my theory here was that this model would have a lot of reference image on the internet to work from. I was not disappointed and quickly found many images to work from, most of which in quite good detail. As well as using internet images, I also like to work on from my own reference images. I am currently in the process of building my own large reference library.
To this end I made a trip to Imperial War Museum Duxford on a reference gathering mission. When I arrived however, I was disappointed to learn that the reason for my visit was not on display and under restoration in another inaccessible part of the museum. The trip wasn’t a complete waste however and I spent the entire day gathering reference material of the other exhibits in the form of photographs. At the end of a long day I had literally thousands of photos to add to my reference collection for future projects. Below one of the photos I took, although not stricktly related to the helicopter, perhaps the surface qualities of the metal might be useful reference for when building the materials in substance painter.
I learnt later that the same helicopter was on display at the East Midlands Air Museum, to which I planned a visit for the purposes of gathering high quality reference images to model from. I was not disappointed with this trip and it yielded lots of images of parts of the helicopter that I couldn’t find any good images for on the internet.
With reference images I have discovered that, if you are going to model something from real life, you cannot rely on images sourced from google. as the camera angels and quality are not good enough. For example, when I tried to model a gun previously for a project I did over the summer, I was unable to find decent images from behind. This led me to kind of making up the object from behind leading to a lack of accuracy. Now this might be ok in an object made from a concept image as there is no real-life counterpart and people will generally accept anything from a sci-fi or made up object. However, when it comes to real life objects people are much more critical. I am myself a bit of a perfectionist as well and always strive to make real objects as real as I can.
As the brief was quite clear that this model needed to be an accurate representation of the Hind helicopter, I went one step further than I ever have with any other project and purchased a model kit of the helicopter in question. I built this plastic kit as a 3d reference that I could use to work from with making my helicopter.
After some more research and reference gathering, I felt that I was ready to make a start on the project. I compiled all my reference images into a large Pureref mood board separated into the different parts so I could quickly jump between the different elements I was modelling.
I started the project by setting up orographic images, as the brief was clear that this helicopter should be a replica and not diverge too much from the original. I found working from the orthographics that I had found on the internet hard as the quality was not amazing, they lacked detail in certain key areas and did not appear to line up all that accurately. In “hind” sight it might have been more prudent to start by making my own orthographic images, a note for next time… Very basic set up of the orthographic planes and early version of the helicopter mesh.
I decided to make the helicopter in sections, instead of as one single mesh. This helped me alot as I did not get caught up on trying to perserve edge loops across large sections. This also had the added benefit of making the helicopter look more grounded, as the real thing is not all one continoeus peice of cast metal, but instead a series of bolted together major elements.
After much back and forth I finally got the shape down and felt happy that it was a close match to the original. I did however spend many days striving from 100% accuracy, however I was forced to give up on this when I realised that I was never going to be able to achieve this without having the actual factory blueprints used to make the real helicopter. My decision to give up on trying to achieve a 100% perfect replication was partly influenced by the advice of my peers telling me what I already knew, that I would not be able to make a perfect replica without actual blueprints or for that matter a massive budget.
Once I had gone about making sure the overall shape was as accurate as I could make within the confines of the original budget, I then got to making all the other elements, working as closely from the reference material of the original as possible until I had built up a complete low poly version of the helicopter.
I had made extensive use of the object painting and place along path functions inside 3DSmax to place all the rivets, screws and polts.
Reflecting on the above images, I feel that I probably took the brief too literally when it asked for accuracy, as I was never likely to ever get this level of detail into the texture maps that I had at my disposal. I think it’s an important lesson for me to learn that I can’t always go crazy with a brief and have to work within the confines of what is asked of me. As time is after all money, that being said, I feel university Is a good environment to test out these kinds of ideas.
Overall, I was satisfied with the overall look of the high poly and feel that it would produce a good quality normal map, despite some of the detail being lost due to the relatively small texture size. I feel that the overall finished model is a very close match to the real article in respect to what budget I had.
The process of unwrapping was made complicated by the sheer number of objects that I had in my 3dsmax project. It was a long and laborious process that I felt could be improved on with better planning from the outset, like for example modelling with the unwrapping of elements in mind working with tools like symmetry. I feel for my next project this is something I will certainly do.
I made heavy use of symmetry in the uv map, stacking objects where I could to squeeze out every last drop of textile density. The process of packing the islands frustrated me as I felt that there must be a good automatic solution to this quandary in this day and age.
The baking process was frustrated by baking errors from the very outset, it quickly became apparent to me that I had not made good enough use of splitting my sub meshes down into enough separate objects. My main problem was objects in close proximity baking onto each other. The solution I found for this was simple but time consuming. I decided that anything that was even slightly overlapping should be baked as a sperate object in substance painter, baking by mesh name. This solution worked well and before long I had a completed perfect bake of the helicopter ready for the next step of texturing. I will however look at better solutions for this stage when it comes to my next project to avoid this error in the future.
When it came to texturing, my principle concern was to ensure that I was recreating the real helicopter as accurately as possible. To this end I started the texturing process by sourcing adequate reference of the paint schemes of the real life counterpart. As this helicopter has been in service for such a long time, I had quite a few different options to pick from, but ultimately I decided to for the most responsible colour, the woodland camo.
I started texturing the helicopter with substance painter and quickly ran into a problem, when I applied the colouring as seen on the real life example the helicopter didn’t look quite right. I struggled using variations of the colours with similar results. In the end I came to the realisation that the colours didn’t look right because my model was firstly not lit correctly, and secondly it had no weathering so looked too new and almost “toy” like.
After the correct environment HDR and weathering had been applied the model started to look the part. The colour that gave me the most problems in regards to this was the vivid blue on the underbelly.
I approached the texturing by firstly making the pristine helicopter and then applying all the weathering effects at the end. I had not done this previously in the second year when I was new to substance painter, which had the disastrous result of weathering layers stacking with objects, creating an non uniform effect with some elements receiving more or less. One final weathering touch that I was quite pleased with was the heat damage effect on the engine exhuasts, I created this by using a graident map set to soft light.
Another step I took with this project, making a first for me was to create my own decals to use as alpha decals inside substance painter, namely the black shark on the side of the helicopter and the other markings. I created these by looking at the original helicopter and then drawing the designs out in photoshop, saving them as a PNG and using the alpha channel in the PNG as the alpha for the brush inside substance painter. This simple and effective additon to my workflow will no doubt be used on all my future projects.
I have enjoyed making this project but I feel that I made life harder for myself by wanting perfection. I have definitely learnt a lot about unwrapping and baking efficiently and plan to use this to make my next projects for efficient. I have enjoyed creating a model that needed to be accurate to a real world reference rather than one imagined. This has definitely pushed me out of my comfort zone and that is a good thing. For my next project I plan to collect references to the same extent as this project as it was helpful to have pictures of all the details. I do however, plan to tone down my expectations for what can be achieved with the time and budgets that I have.