This is the artbook for the hero piece project.
occlusion and normal bakes
high (right) and low poly (left) renders in blender.
Hero Piece Evaluation
At the start of the project, I began to ponder ideas of creating a sci-fi sword for the project. This was followed up by research around sci-fi themed swords. I was influenced to research sci-fi swords as I liked the aesthetics of swords from games such as Cyberpunk 2077 and Ark Survival Evolved. I went into these games and looked at the different swords in cinematic modes to gain an idea of what type of sword I should produce in the final development of the hero piece, since I was set on making a sword for the final development. Furthermore, I went onto Artstation and Sketchfab to look at a variety of sci-fi sword models that I could adapt features of into my hero piece, for example the neon lights on the blade and handle. The swords I found aesthetically pleasing were screenshotted and saved onto my art book to improve my initial ideas and to provide examples of what the hero piece could look like.
Later, I looked at the possibility of a frame for the sword as it would provide a place for the sword to sit for the final rendering and to make the final render look more complete. I used cinematic cutscenes from video games as a primary source of research which allowed me to interact with and see the textures of the swords, but I could have extended my research to looking at the creation of these swords and the topology that they followed. I did use few research methods which created a variety of research that I collected but this could have been more open, and I should have used a questionnaire for additional primary research that would have allowed me to gain insights for the hero piece. When looking at the swords on Artstation, I could have also looked at the wireframe models and examined the topology that they used within the design. This would have allowed me to adapt their topology into my hero piece when I started to develop the model.
My primary research, whilst providing a close look at the models, was not as effective as the secondary research; the secondary research provided more details that went into the model. The Artstation models had the ability to look at all the polygons of the model and give an idea of how the topology should be laid out, the seams should be placed and the layout that should be followed whereas the cinematics in various videogames did not provide any in-depth insights and only provided a visual of the sword that could not be looked at in a 360 degrees viewport, restricting the viewer’s ability to examine the full visuals of the sword.
However, the research taken was sufficient to create the hero piece but lacked in some areas that would have been beneficial for the development of the sword. An example of this would be the UV map layouts as the model’s UVs were not precise and caused the normal and ambient occlusion maps’ bakes to look atypically. If the research were conducted beforehand, the UVs would have been seamed differently and could have avoided the odd bakes. As for the primary research, I focused too much on video game references and could have made questionnaires the prime method of collecting research, since this would have gathered insights from players of videogames rather than videogames themselves. Along with this, I could have put secondary research as a prime focus since the research this gathered was more informative.
At the start of the creation of the hero piece in Maya, I used a 2D reference which allowed me to make the proportions of the model accurately. The basic model of the handle was first completed then the sword’s blade was worked on as a separate object that was later attached and welded to the handle, allowing me to change the topology of the blade and handle independently. This meant that I could later separate the models and edit them easier for the high poly model if I needed to. Once the low poly was ready, I used cleanup tools and auto seams to create the UV maps, which came with various consequences that I would have changed if I had time. The UV maps created consisted of lots of UV that required more accuracy when creating the texture map in Photoshop. Along with this, the bakes came out strange and, even after adjusting the layout of the map, looked extremely abnormal. Since this mistake was from the poor auto seams, I would have and will avoid using this tool to ensure that bakes on future projects are up to standard.
I liked how I used the extrude tool to create the model from the first initial triangle I used for the pommel of the sword as it was less likely to create non manifolds and was quicker then creating new objects and joining them together. However, I did not use this for the sword’s blade and wasted valuable time trying to make a new object for the blade so I would change this by extruding the blade out and adding a different material to it as this method would have been less time consuming and more effective. I also liked how I used a small rectangle to make details into the sword by using the A – B (Boolean) method and I would recommend this method for future use however the original extrusion of the triangle was not rectangular enough and created an arc in the model which competed with the Boolean method so I would refine the extrusions by confirming they’re rectangular extrusions if they needed to be.
As for the details, I made a crucial error of adding the details to the low poly instead of cloning the low poly and adding details there and could have caused the bakes to become abnormal. I should have made the high poly before the details to make the bakes finer as well as having the models look noticeably different for visual purposes. For next time, I could consider extruding details into the design rather than using the Boolean method.
The bakes were unexpected to say the least, and it may have been a result of the UV map layout. For next time, I will make the seams without the auto seams tool to reduce the number of UVs which may have caused the bakes to turn out so odd. The bakes started initially as expected but turned out abnormal near the end and may have been due to the difference in models as the low poly was 1500 faces and the high poly was exponentially higher with a smoothing effect applied, adding to the tremendous face count.
For the texture maps, I screenshotted the UVs and pasted them onto a 2048 x 2048 quality page in Photoshop. This was a high quality with the 72 pixels per inch resolution and allowed the textures to be a high quality for the blender render. I used high quality images online for the textures to ensure that the quality of the texture in the final render was satisfactory. I also used only two different textures, so making the UV maps required less precision but was made slightly difficult by the auto-seamed UVs that I made and resulted in a misplacement of a texture which can be seen on the rendered sword’s blade. Other than this slight error, the render turned out great and was able to be uploaded onto Artstation and Sketchfab with ease.
I would agree that the final renders turned out great except for the noticeable texture differences which could have been avoided with a more accurate texture map. Undoubtedly, the visuals of the swords are aesthetically pleasing for the viewers, the quality of the sword was excellent and the came close to my grand expectations since the sword’s blade could not be made with the original texture, due to quality issues, but still provided a sight with the crystal-like texture presented. Overall, I would only improve the texture map to compensate for the missing texture on the blade for the rendered image.
The requirements of the brief were hit but with limited skills since there was cleanup and auto seams used which maximised the result of the project. Undoubtedly, the fundamental areas I can improve is the seams and UVs for the renders and bakes with a focus on secondary research to see more informative data to benefit the creation of the models.