Skip to content

Finished Product

Finished product:

Finished showreel:

Follow me on facebook:

My website:

Work in Progress 8 – Finishing Touches

So I have been tidying up all the various bits that need to be done. Some last little bits of modeling. For example, the wheels on the shoulders were empty and needed some geometry inside.


Some screenshots of the bevelling process:

bevels2 bevelsbevel2

Rigged the car. This is a simple rig, as all it needs to do is roll into the scene, however as it needs to have some body roll as it brakes, it was actually a little more complicated than expected on account of the link constraints used:


A couple of comparison photos of WIP vs final quality below, you can see the difference bevelling the edges has made to the transformer. I decided to make the windows of the car transparent and model an interior. The headlights have also been improved:


So, I am creating a 3D modeling/animation show reel. But as I am working alone, I still need to do some compositing work. I have used Adobe After Effects to assemble my renders. I found it quite annoying that it doesn’t contain a high pass filter so I found a tutorial online to create my own. Added colour correction also at this stage.


Below you can see the difference the filters have made:


Work in Progress 7 – Quality Overhaul

My last ‘high quality render’ looking decidedly draft quality after more materials and lighting refining. I also spent more than 5 hours chamfering edges.

test60comped WIP9931test31WIPR04


Check out the progress blog for my showreel supporting assets:

Time Plan 2



The bulk of the work has been completed at this point. I need to do some character and car animation, but this will take a fraction of the time of the animation completed so far. The rest of my time is allocated towards responding to feedback. Polishing the model, the texture, optimising lighting etc.

Work in Progress 6

High quality render (change quality settings to 720p):

Breakdown (change quality settings to 720p):

Wireframe render (change quality settings to 720p):

Work in Progress 5 – Refining Details

At this stage, I have finished the composition of the character. To progress from here, I spend time analysing test renders, I look at areas that lack detail, then go in and refine them:


I have also been doing some more in depth render testing. I have isolated the settings that I will be using for my final render. With rendering there is always a trade off between quality and render time. My time management on this project has been good and I have a lot of time left for rendering, so my aim is to produce the highest quality render the human eye can easily distinguish; my decision is to use settings, any higher than which would provide diminishing returns. Put simply – I put sliders up a notch, render and repeat until I can’t tell the difference between the higher quality render. Each time I recorded the render time for the frame; I have left a lot of time but don’t have months to spare.

Another important facet of this stage is to identify what is necessary and what is not. For example, global illumination is widely considered the holy grail when it comes to realistic renders. However, I have found that it is not really necessary and increases render times by around 50% even with medium settings:



The first image has no GI and the second does. Not only does it not bring much to the composition, but I actually prefer the look of the first image. I like the graded background, and it helps to give a sense of scale as the camera zooms out. The colour bleeding is a nice touch, though, and I would consider including GI if I received feedback to suggest that people preferred it. Both of these images have an ambient occlusion pass rendered and composited in photoshop. This is definitely a worthwhile addition; render times are low (<2 minutes) and it really adds depth to the image:




So, an ambient occlusion pass is the first component to achieving final render quality. The next is to change lighting samples from default 8 to 32. This removes the noise from the shadows, however the materials still had some noise. This was solved by doubling the material glossiness samples from 8 to 16 on select materials. These values were determined through trial and error, and increase render time from approximately 7 minutes per frame to 18 minutes per frame:





Of the above images: Image 1 took 5:56, image 2 took 11:56, image 3 took 17:29, image 4 is image 3 with an ambient occlusion pass composited, the ambient occlusion took 2:04. To properly compare these images you would need to download them and open them in an image viewer which would allow you to see them in full resolution and flip between them.

The rest of my test renders up to this point (change quality settings to 720p):

Work in Progress 4 – More Modeling

So I have now linked all the car parts to the transformer and have fully animated a lot of them. It is now time for the next phase of the modeling, although, as always, I will be continually improving the animation, texturing and rendering as new ideas occur to me.

I need to fill in any empty space between parts with geometry. At the moment, the front of the car has some good motion when transforming, but lacks solidity; it is just a load of floating parts at the moment.

This is the stage of the project which should help to really sell the car transforming into a robot, rather than just having pieces of it stuck on to a robot which was always lying prone underneath. The reason I have left such a large chunk of the modeling until this stage is because now I can model around the car parts, I can model on top of the car parts. This will make it look like they really are part of the robot and I can animate the new geometry pulling parts of the car into position, and that position will be moulded around the parts; it should help to make them look like they ‘belong’.

WIP10     WIP19 WIP20 WIP14

Again, I have created videos from some of my screenshot sqeuences (change quality settings to 720p):

At this stage, I have started to put my earlier texturing plans into practice. I had hoped that I would be able to create good realistic textures simply using procedural maps and not requiring any photos or UV coordinates to be unwrapped. However, although the model looked a lot better after the application of metallic properties to the materials, it just wasn’t good enough and I have started unwrapping the whole thing. My first strategy was to pelt map everything, giving visible faces as much texture space as possible, while reducing the texture space of faces that will never be seen. This caused distortion of the texture and so I opted to use flatten mapping for the majority of the mechanical parts. This was less time consuming and gave a better result.


For objects composed of multiple elements, I placed all the UVs of each element on top of one another; I will not be using anything other than the raw material photos (I will not be hand painting each piece with scratches etc), so this will not be a problem.


I have used a single multi/sub object material to texture the new geometry. This way, I can texture by simply selecting parts at the element sub-object level and changing the material ID number. This is useful as it saves space in my full material editor, and it means making changes are a simple affair; just select and change the number rather than select, open material editor, find material, apply to selection etc.

I have sourced and tweaked a number of photos. From each of these I create a material layer. A few materials use no photo reference and just use car paint material with different colours. There is a self illuminating standard material and a glass material for the eyes.


These are google-sourced royalty-free images, all of which have been tidied up and coloured to my needs in Photoshop.