So it's been a while since I posted anything in the portfolio, and that's because I've been burning the candle at both ends for two and a half months (mid January through to early April) producing my first animated short!

Inspired by my initial efforts with my previous portfolio animation pieces - a related advert, a stylised video pre-roll, a Dr Who fan film credit sequence and the Kokorogenic Animations pre-roll, I sought to capitalise on the promise of the latter by producing an animation of respectable quantity and quality, to further develop my skills and (depending upon that quality) make or break any prospects for future projects.

The first thing I learned is that it's possible. When you attempt something like this, the first thing which occurs to you (if you're reasonably earthed) is "how realistic is this with the resources I have at my disposal?". My PC is about five years old at this point (born in Dec 2017), which is nowhere near antiquated, but neither is it really cutting edge or seriously high end workstation material, and it's doing all the heavy lifting - it isn't as if I'm sat under a flickering light drawing frame after frame of this - but was it reasonable to commit to something like this?

There was only one way to find out.

First I had to come up with something. The more I pondered, the clearer it became that my bots were going to be the entry point for this format, and at least artistically there was potential for a slew of such stories. It had to be something sufficiently ambitious that it was more than a test piece, and could stretch all I knew to date, but also reasonable enough that it didn't spiral into a bottomless pit demanding intimate knowledge of the inner workings of the universe to get it over the finish line. Keeping all angles in mind the story distilled down to something revolving around the fate of three robots stuck together in a confined space of what's presumed to be an escape pod, and a tale of their attempt to resolve the issue.

All such projects should begin with a storyboard, or so I've been told. Having tried it I can now confirm the value. Reaching for the most handy pad I set about stitching something together, coming up with the pencil scratchings you see above, and I would reflect later on how close the sketches translate to the actual finished product. With this done I could immediately begin setting out all the technical aspects - what I could do now, and what I would have to learn in order to execute the scenes.

Surprisingly the animation received an immediate shot of confidence. Up to this point I'd already played with character creation and animation - Reg and Bert in the pre-roll sequence were a tidy, quick proof of possibility, and I was confident in their creation, how to rig them (stunningly easy compared to human forms) and their operational capacity.

Somewhat counter-intuitively the robots are actually crafted from primitive cube shapes. For the uninitiated, when you craft something in Blender you usually start your model building with a primitive shape - a plane, cube, cone, icosphere, UV sphere, cylinder, torus - they're the basics. I discovered when I began making these characters that better form was attained by not using a sphere.. odd I know. The eyes are another separate project file, using shape keys to create expression. This model is simply appended into each bot project as required, and modified for their colour component. With this approach I can craft an almost endless variety of bots ready for any occasion!

The squeaky toy was a very simple item, though as I learned later when it came to the replication animation process, it needed to be all one piece, body/eyes and hat. The greater challenge was the spider, a character for whom I have to thank You Tube creator POLYFJORD. His tutorial on how to create the basis of this little guy was instrumental, and one of the more intricate learning curves demanded by the ambition of the story.

After the characters came the set designs, which called upon the services of an escape pod, planet, star and station!

The fundamental design only needed to be that of a simple stage prop and I kept function in mind to avoid unnecessary complexity. Basic greeble technology was embedded according to what might appear inspiring, anything functional worked on the basis that the vessel is purely for the use of inorganic life. The aesthetic for the pod came from the idea that it was either printed or grown. The pod and planet share a similar surface texturing principle, both being procedural rather than textural in nature - somewhat analogous to the difference between vector and raster graphics.

The planet is a tri layered sphere model. A base layer for the surface and city component, and two alpha channel (transparent) layers; one for cloud formations, the other for a thin atmosphere.

The star is more simple than the planet, comprising of only one sphere with an emissive procedural texture, cranked up to respectable levels. The image seen here is the Cycles engine, but pre-render it doesn't have the glare node component, and thus appears flat. In addition to the glare node, the final result also employed some post production emphasis through the DaVinci Resolve editing workflow.

The station was an experimental model which I loosely improvised as a daily practice doodle before this animation project came to fruition. It's one of many such projects crafted for its own sake, but it found a home when the plot required a station. It's not the original concept that I had in mind, as evidenced by the scrappy effort in the storyboard, which suggests a structure more akin to the "Close Encounters.." mothership. Bringing it into the project demanded a little something for the interior, so I continued to form something that was acceptably eclectic/eccentric. Maybe further down the line we'll see more from these vessels? You know as well as I do at the time of writing!

The animation itself became a very fluid, spontaneous and improvised process. Without any overall preconception of a full short movie production I opted for the same approach as the pre-roll animation - take bite-sized chunks and figure it out in a head first dive, so to speak. I anticipated that there would be a lot of very focussed attention to details with a fair amount of  back and forth (not as traumatic as I imagined), and guessed right that a workflow using Cycles for the whole create/analyse routine would be the very definition of hell. Throughout production the Cycles engine was spitting out frames at the rate of one per 3-4 mins on average. With scenes using between one and six hundred frames in length this would make a "trial and error" approach wholly impractical. Thankfully Eevee, though not as realistic looking as Cycles, could do the same between one and ten seconds per frame and look respectable enough to make sense. With a slightly different lighting model between the two styles it became a very efficient way to do back and forth - getting the "dailies" back on those special effects shots, as Lucas would talk about on the Star Wars production. I could try a character/acting animation scene, render it out between a couple and fifteen minutes at a time depending upon length and complexity, and when I was happy with the final choice I could then set it aside ready for proper Cycles rendering. By the time I'd completed "principal photography" the process came in at a repeating total of thirty five times. Thankfully the interpolation process that Blender employs for "tweening" (the nitty-gritty frame animations between main animation poses) took care of a lot of the minutiae, such that the process became almost muscle memory in terms of what I would get in the render vs what I felt was about right with the character expression, timing, etc.. It became an interesting thing to become a director/god, deciding what I wanted of each character and how they would interact in a scene, given their individual choices and motivations. I learned a lot technically, organically and characteristically about what works as a process for this kind of project and certainly feel that I know these little characters of mine inside out.

With all the footage in the can the next step was to jump across to DaVinci Resolve - a true marvel of video production software.

The scenes and sounds were brought in, labelled and organised accordingly, edited as necessary, along with opening and closing titles with a suitable flavour. The sounds were collated from the Soundly app and Freesound.org website, with care taken to ensure that they were used in accordance with their licences, and all creator credit given. Sadly there was one false flag copyright situation, which unfairly undermined the production on first pass - I hope that I can work with the creator in question again using said material on a future project. That little hiccup aside all went swimmingly. I employed some minor post production effects work to enhance the scenes involving the star and replicator.

As with the original pre-roll animation, the project was exported in two step fashion - firstly in Quicktime DNxHD 10-bit format (resulting in a horrifyingly large, and thankfully temporary, 15Gb file!). This was rendered through secondary software (Wondershare Uniconverter) for a You Tube preset 1080p output (with a custom and very generous 20kbps video bit rate), to a significant but altogether more reasonable 1.3Gb file. The reason for the two step process? I found even with the unavoidable You Tube compression, banding issues were significantly improved by going through the 10 bit/Uniconverter process rather than outputting straight from DaVinci Resolve. Why I've yet to fathom, and I shall persist in exploring this, albeit on a fairly casual basis. If anyone can shine a light on this, or have a better way of dealing with banding which doesn't involve production hoop headaches, then I'd be very grateful and open to your insights!

In all the flick came in at a running time of just over nine minutes, which was around what I'd hoped for. I didn't imagine I'd reach as quickly as I did. I honestly felt that the production would take around six months or so, though admittedly all based on arbitrary calculations and presumptions.

The style of the animation obviously rests with Pixar, who I'm happy to applaud as the originators of the style I embraced for this. I adopted this not so much out of any aesthetic laziness as much as it was to have a quality yardstick, to see if it was possible to produce something of sufficient quality that wouldn't be embarrassing. It was created using nothing more than a singular PC with admittedly a decent, but modest six core CPU (AMD 1600) and GPU (EVGA GTX-1070SC) setup. No render farm, no secondary production assistance. Without counting the actual power usage and hours spent, which are a fundamental given anyway, this project went from start to finish for £0.00 overhead. As stated it's taken around two and a half months of work, more often than not leaving the PC to render the final scenes overnight (and it has to be said keeping the bedroom nice and warm through the winter - I consider that a bonus efficiency).

The only real downside I've encountered to this project is the realisation that I could've worked in an even more efficient workflow than I did. For technical and quality purposes the animation scenes exist as rendered still PNG images collated into scene folders, which are then strung tidily together in sequence by DaVinci Resolve. It was only too significantly into the project that I discovered my "spider saviour" POLYFJORD had created another video talking about EXR workflow, a more proprietary format which has caused me to explore re-rendering the project, for storage purposes (thankfully the quality seems equal). Working with the EXR format rather than PNG requires very minor extra work in the post production stage, but the savings in disk space appear very significant. To illustrate, scene I'm currently having a go at re-rendering exists as a folder containing 270 PNG files, each around 9.3Mb in size for a total of 2.43Gb. The EXR equivalents come in between 843K and 1.08Mb per file for a total folder size of 253Mb. Spread over a thirty five scene project coming in at 100Gb total, I anticipate the savings from EXR conversion to be very significant, and consider this production path well lit for future animation projects.