Term 4 – Cel shaded room tour – FMP

For my final major project i wanted to branch out and try something new, previously in Term 3, i had used Blender to try and create photorealistic rendering in scenes, as well as the previous terms had focused on creating realism in Maya. With my thesis being focused around cel shading and it’s uses in speed compared to photorealistic rendering, i chose to make my final major project an animation in the cel shaded / toon shaded styling, i had no idea how to do this and had done some surface level research on how to do this and how i could go about doing it and realized it would be possible to do in Blender.

Initially my intention with this project was to make a 1-2 minute animated shortfilm in the style of cel shading, where i had planned out a rough draft of what would happen in the animation and some inspirations behind it. With this project i wanted to build off of the modelling and animation i’d learned with Blender over the last two semesters and use that to both further my portfolio of modelling and animation as well as create something outside of the traditional rendering style i had worked on prior.

Below is a short synopsis of the summary of what the short film would be about

The short film will be a 3 Minute Animation focusing on an Artist’s struggle with a lack of motivation to work. The animation will focus between two characters; the artist and a small rat that lives inside of his wall. The primary bulk of the animation will come from the POV of the rat’s perspective. The animation will go through the story of an artist being given an assignment and throughout the first minute of the animation shown over the course of days through passing time shown through the window and the character being moved to different positions, things go well for the first day however in around the second minute of the animation the artist begins to feel a lot more stressed, becoming visibly distraught and feeling burnt out. Much of the animation will also focus on the rat also living in tandem in it’s own hole, and occasionally viewing the artist working. Towards the point of the artist passing out from exhaustion, the rat will then see this and attempt to make some tea to wake up the artist. The rat will finally succeed in this and then wake up the artist with a noise. The artist then gets back up, finishes the work and seems content. The rat then views this and is happy and goes about his hole.

Throughout the summer i had been experimenting with a very experimental motion capture software for use in animation. My initial plan was to use this motion capture software and record mocap, this mocap software would record movements for every frame, and my plan was to do so and then delete every 3 frames and keep one left, resulting in less fluid but more realistic animation styles.

That experimental program can be found here:

https://www.mocapfusion.com/

A lot of my initial inspiration for this came from the Disney film ratatouille, the tv show arcane and the tv show love death and robots.

This project frankly was way too ambitious towards the start, as i was both learning to replicate and build upon an art style i had never done before, as well as to Model and texture an entire scene as well as animate, rig and model two characters. Later on in the semester, with the advice of kristos, this project was very heavily scaled back to something that could be achievable in simply creating a room and modelling the objects in it and focusing heavily on perfecting the art style and cel shading that i wanted to go for.

Initially i started off with some very simple modelling of the room and building it out with some very small things, a window, a desk and a small lantern.

Next came some experimentation, i wanted to see how regular models would look if i attempted to texture them the normal way i would texture things and then apply a lot of the cel shading effects that i would implement onto them.

I also was experimenting with using HDRI’s initially and using the cycles engine in blender to render, as previously i had been using a lot of the Eevee engine. This was early on in the summer and i hadn’t actually used Blender in around a month so i just went about getting back into things by building out some simple things, a chair and a bookshelf complete with books made from using an array modifier.

Here’s where i went about starting to do a lot of the tests in shading and how to replicate this cel shading style that i wanted to do. In a lot of Cel shaded animation, when light hits an object it will diffuse onto a surface in a way that only shows one tone of the light, or in some cases two, being highlights, shadows and the initial texture. An example would be these two images:

Sources on the two images:

https://en.wikipedia.org/wiki/Cel_shading

http://vodacek.zvb.cz/archiv/750.html

I was initially using the eevee renderer and went about creating the toon shading in the most common way i had seen many others do which involved taking the initial color of the object, making it a texture image, then using a shader to rgb node and using a color ramp to clip any lights that would hit the object and make some very abrupt and jagged shading as opposed to a gradual slope of the shading of an object going from full color when exposed to light to slightly darker where no color was hitting it.

I still continued to experiment with realistic textures and experimented some more regarding the previous method of using a color ramp shader.

Next i also went about importing some of the test mocap that i had done using Fusion Mocap VR, the experimental software still in development that uses VR headsets and controllers to track your movements. If you have two infared base stations as well as three other infared trackers for hips and legs, it can be a very cheap and usable mocap software if you want to do mocap from home.

I started also using a simple whiteboard and writing down small tasks to complete on the project along with a trello board to try to get a bit more organized and also to try and work on getting some environment layouts and rough character designs done.

Next i wanted to add some emissions and some glass to my scene for the window, as well as i wanted to create lines around my objects This was where i made the first big switch of going from rendering in EEvee over to cycles.

Initially in Eevee there were two problems, one being that i couldn’t use any emissions or reflections in any way, as at that reflective objects in blender could not be seen through, as well as i believed emissions also could not be used in Eevee.

Freestyle lines are the lines that can render around the object to give it that cel shading effect, however this is a post processing effect, meaning the image will initially render and then redner the freestyle lines based off of things like the mesh, camera angle and lighting. However this can take a bit longer and at the moment i was using a different way of doing these lines around an object in the way of giving things a solidify modifier and then using some alphas to make the original model display overtop of the solidify modifier.

This initially worked but had many severe limitations. for instance objects would often clip into the ground as the solidify would go through other objects, and every individual object would need to be modified to be an appropriate size on the solidification.

I ended up switching over to cycles but this posed two initial problems. The shader to RGB ramps don’t work in Cycles, meaning the toon shading inner texturing would need to be done another way, the render would take longer, howver i gained the ability to have all lines generate relatively, which looked a lot better, and i also gained the ability to use emissions again.

I also went about building some more small things to the room, a lava lamp and basic bed as well as sculpting some small pillows to test how freestyle lines would affect more diverse heavy objects.

I built some very basic humanoid figures and a rat figure out of blocks to then prepare for more heavy modelling and animation later on.as well as designing and using the sculpting tool to create some clothes that i could use in the scene.

I started doing a lot more heavy duty texturing in this portion, and i found that the toon shader available in cycles could somewhat replicate a lot of what i wanted from the cel shading art style, however it didn’t completely get what i would want and i would need to continue more experimentation later.

I went about building a lot more to the outside to be able to see through the window, as it wouldn’t need to be super detailed but i still wanted it to be visible, i also built some small cacti by using a cylander and making hair particles replicate a spike i made and using vertex groups to only come from certain parts of the cacti. I also experimented with making a color ramp for the trees that i could generate using the sapling tree gen plugin from blender, however it made more sense to make it one single color to fit the simpler art style. I built a small road with gates and posts using some arrays and some buildings int he background that would have some assigned texture emissions on lit up windows.

Things were starting to come along, however the constant technical problems and swapping back and forth between cycles and evee was becoming a problem as i really just couldnt get the art style i wanted by using either one of them.

Around this time i met with Kristos and he reeled my project back hard, which was frankly much needed. i was trying to do way too much in such a short time and the project was already bogged down with a plethora of issues. it was settled that what i would primarily work on was the art style of the scene, while doing modelling and texturing and decorating the room to show off this art style being the main focus as opposed to a large animation on top of experimenting with a new art style.

One of the main issues i started to encounter with this project was the issue of simplicity. Everything i was doing felt too simple, the shading looked like one to two colors so it looked off to me, or the models seemed too simple so it was an issue. This was due to the toon shading art style being simplistic in nature, and everything i’d done prior in the program being very highly detailed in modelling and texturing.

I went about then finally making the bed look a little better, by using some lower poly pillows and then sculpting them, then subdividing i was able to get some decent looking pillows that were smooth enough that they wouldn’t have messy freestyle lines but still giving some definition, as well as creating a blanket by using an assortment of objects under the blanket and giving it cloth / collision physics and then freezing the result. The blanket especially was definitely giving that diffused toon shading i was looking for with light hitting it.

I also managed to update the small laptop i had been working on the desk on and give it a proper keyboard with extrusions and a keyboard texture with some photoshop work, as well as giving it some small emissions.

I decided to make a couple more things to build the room out involving some fake vines to hang across the ceiling similar to some modern bedrooms using a spline curve that i couold then give depth and use a particle emitter on to distribute some leaves that i would make and texture paint on to.

I had started making larger strides with making many other smaller objects like potted plants that i would use a displacement map on for the soil terrain and arrays to build, as well as making items like a tablet, a monitor, hard drives and a very messy cable structure to attempt to be true to every artist’s desk layout in real life.

I also went about adding on some small details to some more clothes that i could have strewn about the room to give some added environment building and realism, including my old college’s hoodie as that is what i had on at the time of modelling.

Afterwards i decided to make some very small things like balls of paper to be strewn around as well as trying to work out some issues regarding objects not appearing to have freestyle lines behind windows. This was due to freestyle rendering whatever was in front of it. If anything obscures the viewport, it wont render anything behind it, including reflective surfaces.

My solution at this point was to fully remove the glass from the window after some debate with others, as really glass doesn’t fit the art style nor is it really toon shaded and is more in line with photorealistic rendering. I also kept truing to get the proper shading detail on objects but was still unable to get it to get the detail i really wanted.

At this point i ended up switching back over to eevee as i wasn’t quite getting the results i wanted, and started experimenting with using eevee again, this time getting the freestyle line generator to work in eevee. I started trying to use things like bloom instead of emission and caching light casting data using a reflection cubemap I also tried to experiment with making highlights directly from light sources and making their angle at 0.1 to get some very hard shadows on lights.

I also wanted to reuse the man and rat model somehow and decided to try and turn them into plushies to put on the bed, however remeshing these models proved to be detremental to blender.

Here are some of the bloom experiments. I realized if i wanted this to look good at all i would need to make the bloom extraordinary light.

I stopped focusing on some of the issues i was having between eevee and cycles and decided to work on making some small things and texturing instead. I Purchased a cel shading photoshop action pack to save some time when creating certain textures like the computer screens and making them a little more in line with the overall art theme. Around this time i made the switch to finally go back to Cycles and make it work no matter what, and to stop switching between render engines.

I started making a lot more smaller objects and UVing them if they were meant to have a proper non single cel shaded color on them like a mousepad and mouse.

I built out my scene a bit more focusing on filling out small things like adding more cups, fixing small issues and adding some plants and more cables. At the behest of kristos i decided to do my camera angles and then build around those camera angles to give some more focus to my scene. I wanted to explore using depth of field in these scenes and seeing how that could help my scene, as well as experimenting with some other things like atmosphere and dust. I also started making freestlye lines baked into some meshes like the cup, meaning that no matter where the camera was looking some lines would be hard baked in and always try to generate.

I also started adding in some small things like plants i had made last semester to fill out my scene and even some things like the coin and ring i had done term 3. However one thing i needed to be cautious of was also making sure that these objects were decimated as if they were cel shaded they didn’t need to be super high poly, and all being high poly would do would be drive up the render time and the generation of the freestyle lines.

I experimented with using a particle system in a small cube box with transmissions to create dust, however this really didn’t add a whole lot to the scene and even after making freestyle lines excluded from the dust it really didn’t look right or add anything.

I also experimented with using volumetrics and while i did like the look, it both didn’t fit the art style and had the same issues with going outside lines that bloom had.

More experimenting with the volumetrics and the dust particles.

Even with making the dust very subtle it seemed to just not fit the style and look off as well as the volumetrics.

I decided to go back to the way i had it before and simply tweak my emissions harder.

At this point i ran into my first major issue. In the above photo everything is in focus and there’s no depth of field, however using depth of field would only cause the render to blur. Freestyle lines are generated in a post process after the render, meaning they would all still be in perfect focus and ruin the way it looks. However i came up with a fix to this that i will discuss later on in this post.

Now i started working hard at building out the areas where my camera angles would be as well as adding in things like posters and other objects that were all artwork i had done in the past and then used the photoshop plugin to create in a cel shaded style to fit the scene.

I had a wall of lights that i wanted to be one of my main focused shots, i used a bezier curve to build out the cords and made the string lights and emitted the lights the same way i had done with the vines, however first i went about photoshopping a load of my polaroid photos into the art style that i wanted to fit the cel shading art style, then going in and putting these onto each of the picture s i was hanging from the string lights.

The uv’s needed to be remapped on some of these which definitely created some funny looking stretched textures.

Next i made my light instances real instead of a particle emittion as i wanted to tweak the light colors and rotations along the string lights, and made sure there was no overlapping emittion geometry, deleting where need be.

A consistent issue i was having with a lot of my lighting was that it was emission lighting, not direct lighting. This usually would be fine however, emission lighting has an odd way of not working super well with toon shading and can cause a lot more gradual diffusion on a surface instead of the sharp diffusion i was looking for. I tried to edit the emission values and see if there was any way of making it so that i could tweak the angle or radius of emissions but this wasn’t doable. i started to have the idea of taking direct lighting and putting it beside or fully inside the emission instances, removing the object’s shadow casting properties and then either curbing the angle of the light if it was an area light, or the radius if it was a point light to get some really sharp curves and get that simplistic cel shaded lighting style. For now i was simply experimenting with it but later this would become my main way of lighting.

I tried changing the texture of certain light emissions but these still had relational data as they were grouped as particles, i removed all instance data from them and went about tweaking the lights of each of them and rotating them to give them some differences.

I also had some odd geometry slip ups when moving cables occasionally but fixed these by simply smoothing with the sculpting tools.

Once i had made a lot of the lights different colors and started to make them a little different from one another, i tried out making individual lighting for each to see if i could get the sharp lighting i was looking for instead of diffusion

Around this time i also experimented with tweaking the scale of diffsion on the walls but this didn’t seem to do a whole lot for me.

To get the lighting to be a lot wider, i moved the actual physical lights away from the object lights to give a kind of circle of light around each one and removed their shadow casting capabilities, which made it look a lot more appealing and in the art style i wanted.

I then went about using this new direct lighting system instead of the emission based lighting system i was using previously around my scene.

Next i had to fix a lot of the little pictures as the UV’s were still off, however it made some pretty interesting little photo errors.

For the posters around the room i used a combo of some of my old art for the tapastries and made them cloths that i could deform and freeze, and used the photoshop filtering i had done on both them and the posters.

I wanted to add some more detail onto my lava lamp and decided to make some small animations in after effects. Initially i wanted to hand animate it but then decided to use the AE particle effect cc mercury.

Using mercury i could both make the color but also remove the color using the shading and only have the outline if i toned it down heavily

Initially i just doubled this up and tried to render it, however after i decided to render out the color and shadows as seperate compositions and then recombine them. Using the multiply function in after effects, and even double up the outlines if i wanted it to be darker.

I doubled it up and tweaked the outlines and then re-exported it

I attempted to put this on the lamp, fixing some odd geometry with the lamp and re-uving it, however even when being properly uv’d it just looked wrong, and needed to be simple to fit the art style as this was complicating a simplistic art style.

I also went about filling many picture frames around the room with more photos and overall trying to build out the scene a lot more. Even creating some small things in illustrator for some small photos.

At a certain point i also intended to build a small lucky cat as well, however this ended up not being built later on due to time constraints and technical issues.

I tried once more to also incorporate some of the blocky models into my scene with the hopes to subdivide them, however blender’s remesh just simply isn’t built for it and it would have been better to build out the model as one object from the start.

I also wanted to build a small roman dodecahedron as a small piece of the scene, however i wasn’t entirely sure where to start with building one as there were no shapes even remotely like this in blender.

Image taken from wikipedia

I managed to find a plugin that allowed for different shapes not traditionally in blender vanilla and used this along with the math function to make a dodecahedron.

It had some odd topology even to start, however to do what i wanted, i combined the faces for each side, inset them all by face and then deleted the inset face.

It took me a couple tries to get what i wanted with the right thickness and i used vertex groups again to make spheres come from every outward verticies to match what i had seen from a lot of the roman dodecahedrons.

At this point i was needing to make some final touches and updates, and made some more clothing and sculpted it to fit on the bed, as well as fixing some small cacti and making some cacti variations.

This was a point at which i had major issues. I had optimized my scene as much as possible, yet a full frustrating day of rendering later, things refused to render properly and would constantly freeze on render. I worked on optimizing to the max hopinjg this would fix it, decimating things and even completely grabbing everything and putting it in a new scene and tweaking the backgrounds, nothing would work.

I then learned that the version of blender i was using had a known bug, and after updating would render properly if set to GPU rendering.

One good thing about this change is that it had removed any lighting glitches i was having in the old scene, such as world enviornment interfering with colours, and i was able to rest my colours and get a bit of a nicer darker blue tone to the scene overall.

I made some last minute adjustments to my scene and started rendering.

At this stage, i need to talk about an issue i discussed much earlier with freestyle lines not blurring properly.

I had never touched the compositing side of blender until this project, so going into it was completely foreign to me and took some time to getting used to. One thing i found was that i could render both the z depth from my camera and then render my freestyle lines both as passes, then make the export an open EXR and combine the two after the render to get the focus depth on the freestyle lines that i wanted. This was an extreme headache to initially get but once i figured out how to work the compositing workspace of blender it became repetition.

All that was left to do at this point was fix some minor issues and make some transitions for between the shots in after effects.

After all this comping the project was finally done and can be found here.

Collaborative Project – Liminal Spaces short film

To do this personal project i originally started researching doing a project like this all the way back in January. to not tread over the same ground that project was here: https://bradrobertsonvfxblog.myblog.arts.ac.uk/2022/01/20/term-2-maya-experimentation-with-liminal-space-builds/

When proposed with the idea of another collaborative project, i decided to take a different route and collaborate with people outside of the university. The four people i chose from my group were three friends who i could count on to help with the project, Eric Vonunnius, a Canadian based editor at Publicis in Toronto for the audio, editing and overlays we would need, Andrea Pinilla, a graphic Designer based in London Ontario, and Barbaros Saritas, a Voice Actor based in Hamburg. I originally approached them all in January with the idea of making this project, however it didn’t go a far ways as i was needing to lead the group and ended up getting covid. This ended up being a blessing in disguise as later on i learned we needed to do a collabrative project, and decided to make this project we has wanted to make a reality.

For a small recap of what was discussed in an earlier post, many animations have been appearing over the last couple years of “Analog Horror” on youtube. These videos usually follow a similar format of using video distortion effects and old vhs style formats as both an aesthetic choice and a way to conceal non studio quality animations. Some of these animations are of a very high quality and some are of a bit lower. owing to some animators not having as much experience.

We started with doing a bit of digging into some older liminal space paintings and portraits for information and inspiration. I originally made all of this information into a powerpoint to meet with my professor regarding it for clarity and as a guide for my groupmates and myself when coming back to it.

After this we took a look at a lot of different channels on youtube and in games and saw what they were doing, seeing both what looked realistic and what didn’t as some has poor animation quality due to lack of knowledge with modelling and UVing.

Many of these projects that others have done surrounding liminal spaces usually come from very small errors that result in something not being realistic and makes it obvious that the animated video is CG. A lot of these problems come from unrealistic Camera Zooms or movements looking more static than dynamic, colour grading that looks too unrealistic, walls or surfaces being either too reflective or completely unreflective, the lighting being off, not having any bump maps or having UV’s repeat too much. One of the core problems in making something like liminal spaces is that you’re intentionally making something look surreal or abstract but must look realistic enough to appear to be a real place at the same time. One of the things to focus on was to focus on making geometry and UV’s as realistic as possible, and one of the main things that would need to be done is get realistic lighting along with colour filtering.

After this we discussed a timeframe and how many scenes we would want, as well as the length of the project.

This amount of work was originally much too ambitious, and i would learn both from my professor’s urging to scale the project down and later efforts, that the project needed to be scaled down tremendously.

We then decided on some of the timeframe we wanted as well as formats and overlay ideas, these were all rough concepts that may be changed later but we wanted to get a general idea of a timeframe to base the story off of.

Originally we had intended on seven different scenes. This was later cut down to 4, and then later still cut down to 2, as we realized it would both be too long, take too much time to model, and we wanted too make sure what we did model was of better quality.

After all of this and meeting with my professor to discuss details, i decided to get to work, the main issue was that this semester i wanted to challenge myself to use blender to complete this project, however i had no knowledge of blender, it’s shortcuts or really anything about it.

I started by trying to build the hallways scene, as it was the first scene, and would be the longest and largest scene.

I started with fidgeting around with the controls in blender, learning just simple things like face, edge and vertices selections, as well as placing around shapes.

I didn’t get very far, and started up the next day with trying to build some simple light fixtures and learn extrusions, i struggled a lot with the hotkeys of doing simple things like moving objects around, learning about the different rendering methods, and simply using face selection tools. The blender hotkeys website became an absolutely invaluable source of information at this point along with tutorials.

https://docs.blender.org/manual/en/latest/interface/keymap/introduction.html

After this i began working on learning extrusions to block out the scene more properly and learning how to do UV’s in blender as it wasn’t as simple as in Maya. This was one of the most difficult things as well as learning lighting, as using emissions wouldn’t work for The renderer that i was using as i was using blender’s Eevee render as opposed to the cycles renderer that was in blender.

The way that i built these hallways was by blocking it out using cubes and then subdiving them, however this was acually making it much harder on myself, i did this as i worried about the UV’s being messed up, however it would have been better to hollow out a cube and then extrude it from there, then use the cube projection UV Fixer to fix these issues.

Once i had an okay mesh to work with and had built it out a bit, i began using the camera shake plug in add on and rendering out some test animations for our editor to put some overlay tests on. There was an issue with this render as i would later find out that things that are hidden are only hidden in viewport and not the renderer.

After this i began working on some small work for both the MCEsher and Mall scenes, trying to create some small flowerpotted areas with some displacement and bump maps to create a realistic look. However i realized that displacement maps were not supported in blender’s eevee render and i would need to change my whole render engine. Usually i would just do this, however as this was made to appear as if it was a lower quality VHS, we decoded to just use higher bump maps.

I started then working with our editor in adjusting the bump values, ambient occlusion levels and focal distances as when our editor would apply VHS Filtering it would usually crush the blacks down on our video, we went back and forth then adjusting the bump values and the filtering.

For now i decided to attempt to start blocking out the final end scene as it would include water and potentially take a long time to render.

Originally doing this i tried the same method i had used with the first hallways section using cubes, however i found this was a nightmare with UV’s, and found later on that using a hollow cube and extruding was much much easier in the long run.

I also started to do things like affecting focal distance and making sure it was using filmic colour management as well as motion blur. for what we referred to as the “sewers” scene, we modelled it after a flooded subway tunnel. For now i blocked it out and got some textures for it, and made sure to sort the UV’s, this proved very difficult at first, but later was solved by making a lot of seams in places and using a lot of seams across the object.

It was at this point i started going a lot more into shading and making sure displacements were proper on both the original hallway scene and the final scene.

I also went about creating bump and height maps in photoshop, though while this would work it would absolutely break photoshop every time i would go about doing it.

I also went about fixing a lot of the geometry that was overlapping, as well as making some doors using booleans, creating some better looking lights using emissions, and adjusting some of the colours for the textures.

This was one of the later tests when we began adjusting some of the motion blur and defocusing.

After this i went about creating doors and doorhandles, however trying to make a keyhole proved difficult, and we decided it would be better to simply just keep it as is, as it was something so miniscule.

after i finished up using texture blending and adding noise to the doors to give it a bit more texture, i also added numbers to each door as we liked the idea of multiple doors with no clear example of which one would lead a proper way out.

While talking about the story, we had Barabaros write our story. The story was in laymans terms that a man goes to a locations his friends called odd, only to become trapped in an unending cycle or corridors with a creature pursuing him. It’s meant to be slightly vague and open to interpretation as it was more important to have our environments then a fully fleshed story.

At this point i had our graphic designer go about creating some maps for each location to try and map around where each camera would go throughout each scene to get a feel for the timing.

I then spent a large amount of time going about creating the hallways map to our designer’s specifications and came about a small issue., eevee would only support a certain amount of of lights per scene, this would mean that i would need to create multiple renders of the scene and delete portions that didnt have lights being shown.

After this i started building the MC Esher scene, at this point in the project, time was running thin and we started to realize we would need to cut things, i went about using booleans and extrusions to attempt to create an mc esher scene, though after having a meeting we decided to cut the scene out entirely and rework the script as it would simply be too much time and we wanted to perfect the scenes we had. I also went about editing some of the textures of the final scene.

During this we also decided to take a look at the mall scene, and decided wether or not this would work in the time we had or if it was also too ambitious, i went about building some small plants as i had prviously done based off of a variaty of different tutorials i had found throughout youtube, using both splines, vertices smoothing and other techniques to make some proper plants that we could tweak and make variations on.

At this point we all made the decision that this would take too long to model, and if we did it may look a lot more fake then the rest of what we had, knowing this we decided on reworking the script. We would keep the two scenes, our hallway scene and our end scene as the two main areas, and we would go with all we had planned for the hallways, as at the moment the hallways scene was very long and we could do a lot more with it.

I then went back to the final scene and started adding things to the floor as well as making a single plane and using that as our water texture. Frankly it would both take too long for a fluid simulation and as it would be fairly compressed and filtered at the end anyways we decided to make an animated water plane instead. I roughly used a subway tutorial for this to figure out how to use the twist function to twist a mesh towards my 3D cursor, however other then that it came down to me simply modelling it based off of what i’d learned through tutorials and researching the blender website.

Now i went about tweaking the lighting and water to interact well together, as well as editing the world’s shading to render in shadows better down the main hallway and make it darker. I also tried using an SVG our artist gave us for the wall portraits, i attempted to make a layered portrait, however as the SVG was many duplicated squares they all came in as seperate splines, i’ve found using SVG’s is very unreliable with blender, and decided to instead use the image i’d been given, create height and normal maps and then use those to create the illusion of 3D Depth as these wouldn’t be looked at in depth.

After this i then went about doing a lot of things, fixing the railings, adding holders to the railings, lining things up properly, and adding in a big cast iron ship hull style door with a wheel well, i tried to make the UV’s on these proper, however at a certain point this would be a dark, vhs filtered red lit hallway and it wouldn’t matter too much. At this point i finished doing the camera animations after a couple hours of work, i then sent this off to our editor and we had our final scene. The run animation was a bit too quick, however we decided we would fix this if we had time as we needed to mainly work on the hallways.

After this i went about adding in the final pieces of the hallways, i added in our scratchy wall drawings making sure the alpha stayed in as an image plane, made a small hole using a boolean so the walls would connect after bridging them together, and then went about doing some lighting animations and camera animations. Originally the hole was meant to have a glass wall, however at the urging of our group i kept it open.

After this it became deleting portions of the map to render in sequences for the lights, and then simply camera animating and rendering. I then sent it off to our editor and he applied our VHS Effects. Below is the final cut

Overall i’m happy with the project and learned a lot regarding blender and animation in general. however i wish i’d had more time to animate and tweak a lot of the camera movements as some can feel a bit stiff or too quick.

I’ve also chosen to show the version without VHS effects as the vhs effects crush a lot of the detail in the final version and i’d like to show some of the detail in the texturing in a separate version just for my blog.

Personal Project Progress

For this project, i had been seeing a lot of different ads on social media for jewelery across things like Instagram and snapchat. Throughout the course i hadn’t had much if any time to work on things like advertisements, and having worked on ads before, i wanted to create a short ad to build this skill for both my portfolio, as well as both learning blender and practicing with illustrator and after effects again. I originally made a presentation to show my professor regarding pitching this idea for this ad.

I wanted to follow the way that most ads are done in this by making two 30s versions at 16×9 and 9×16 with 15s cutdowns as well as a pairing 4×5 in instagram formats.

I wanted to make this a fully CG Ad similar to some ads i had seen in the past like a lot of Apple’s ads with high quality renders and motion graphics.

In terms of programs i wanted to use ZBrush, Blender and Houdini. I ended up using just blender, as i wanted to try high quality modelling without ZBrush as i had used it before, and chose to do my fluid dynamics in Blender instead of Houdini to learn it in Blender instead.

•With this project I wanted to challenge myself to make something shorter and work more on refining something small rather then to create something large as I wanted quality over quantity with this project. Some of the challenges that would be presented with this were creating logos properly that would look professional, creating these animated supers and logos in After Effects as it’s something I haven’t done in a while. I also had no knowledge of blender going into this semester, so i would have to learn it from the ground up.

I detailed and outlined each of the scenes i wanted to do, these were all approximates of what i wanted to create and acted as rough storyboards that would guide what i was doing. Towards the end i decided to cut out the comp shots, as i wanted to focus on the 3D only and i had more then enough shots by the end without them.

I started off by simply learning some basic things in blender like simple objects and making rings by making toruses and using proportional editing and selecting edge loops, after this i learned about using lights, adding bloom and effecting render settings, and pluging in textures in the shader node graph. Throughout the next couple of weeks, i learned about key shortcuts from the blender website, as well as using the cloth dynamics in blender, and freezing those cloth transformations to create a realistic cloth to sit on a podium and hold the jewellery. Next i went about creating the first ring being a black marble tear stone ring. i did this by separating the textures and then adjusting the edge loops. This took a while but i was successful in the end. For the second ring i wanted to make an okami face on a ring, but decided i would make an SVG Of this in illustrator and come back to it. Next i decided to make the crow skull, this took a long time as i wasn’t familiar with Blender’s sculpting functions. i used my tablet for a lot of this and went through some variations with cutting holes in the mesh of the necklace and using a reference image plane, ultimately i didn’t use the dynotop feature in blender for the necklace to model, as it does create more polys and give a more detailed mesh, but i wanted to try one necklace with this and one without to compare results. Next i went about creating a watch that was a lot more difficult then anticipated. This is due to the fact that some of the modifiers in Blender can be extremely frustrating. To use an array in blender, if you want it circular you need to mess around with adding an empty to that object and then moving the empty to try and align it into a circle, this took a long time to get right but i did eventually get it. I originally wanted to add an LCD screen to the watch and even started making some animations in after effects, but this was scrapped as i didn’t feel it met the aesthetic i was going for in the end. After this i started using some arrays to make a chain for the necklaces and made a simple wrist strap for the watch, i then went about learning using the dynotop feature for a pirate coin similar to that in pirates of the Caribbean for the last coin. This was a blessing and a curse, as it gave very good detail but also made so many faces that it would constantly lag out my scene. I made one half of it and used a mirror modifier to make the second half as it was only the top half of a cylinder. I Also went about learning texture painting and using texture masking to make the small divots in the coins that would be the darker parts.I liked this coin but the dynotop made it look too detailed inf act and i wasn’t super comfortable with blender’s sculpting yet. i smoothed it out as much as i could to make it less jagged, however i prefered the crow skull in the end and would most likely go about the way of modelling i did with that over dynotop, that being subdividing and then sculpting. I then used the same texture painting techniques i had used on the coin for the crow skull.

It was at this point i took a short break from the project to work on the collab project, then came back and had a lot more experience with blender, i decided to try and get the SVG Ring working and designed a simple okami wolf face in illustrator. i originally had it with a styalized brush, but this caused issues with blender’s SVG creation and decided to just make a simple outline.

Something i learned from this is that SVG’s need an extremely specific way or the mesh will be extremely unorthodox and not interact well with lighting or anything else. I went about using the clean up mesh tools to do what i could and a solidify modifier to properly make the geometry. I then texture painted the colours on and was happy enough with it to consider it done.

Next i started doing small things like modifying the wrist straps for the watch and starting to decimate things in my scene as it was becoming too high poly. One thing i shouldn’t have done was decimate collapse my ring instead of unsubdivivde it, as i was trying to conserve memory i made a permeant edit to it which would become a problem as the ring was now very triangular for the okami face.

After this i then repurposed the plants that i had made in my collab project and texture painted them, as i would use them as background elements near the pedestals i would keep the jewellery on.

Next i did a lot of cleanup, as the scene was extremely heavy and things needed to be decimated to render properly. i went about decimation through unsubdividing and also trying to get a small fluid sim working for the ring with water on it shot.

After this i both animated the clock hands and finally got the fluid sim working, making sure to make a highly transmissive material to use as the water texture. I also started making some of the mograph ring shots by using arrays like i had done for the watch.

I now went through making different cameras and then rendering out my shots and putting them together in premiere. I also started doing some rough logo designs for a logo and finally settled on one i made in illustrator and would animate in After Effects.

Lastly i simply went through going in and editing my advertisement into its cutdowns and aligning shots, animating the logo and making sure things fit together.

I originally wanted to have some text animations, though after looking at a lot of other instagram ads, most didn’t have any, and i felt they wouldn’t work with my ad. I then exported my cutdown versions and uploaded them.

This project presented many challenges, though i’m overall happy with the way it turned out.

Below are the two 16×9 versions

Then the 4×5 versions, however these cannot embed properly as they aren’t 16×9.

30s – https://youtube.com/shorts/8IF8PqIMAP4?feature=share

15s – https://youtube.com/shorts/he6pyp8Xf_8?feature=share

And these are the 9×16 versions

30s – https://youtube.com/shorts/eWvW3zuzm54?feature=share

15s – https://youtube.com/shorts/pss1mabFyDg?feature=share

Week 3 FMP

My topic i chose to research was Cel Shading In Animation, Two questions i wanted to ask regarding this research was If Cel shading could be used to save time as opposed to traditional shaders when rendering and if so, how much time would this save when doing rendering. When doing some traditional realistic 3D Renders, things need to be as photorealistic as possible and this can up render times to the point at which rendering simply 90 frames can take days or even weeks depending on how complex a scene is. I wanted to research whether or not using stylized shading could circumvent some of this problem, and if so to how much of a degree.

After looking into some of the research regarding my hypothesis, it seems that while photorealistic rendering is heavier on machines overall, there are some things to keep in mind when researching this topic, for example the software that is used in edge detection to create these outlined and cel shaded models needs to usually be tweaked to fit the correct style. However a lot of the techniques described like Bump mapping in the Towards real time photorealistic rendering: challenges and solutions paper, are not present in Cel shaded animation and thus cut down heavily on render time. The time that is saved by not rendering in Photo Realistic Renderings may be allocated to getting the correct look when doing NPR Renderings.

Keywords used: “Shaders in rendering”, “Photorealistic rendering times”

Sources:

Cel shading with OSL in Renderman 21 for Maya: https://nccastaff.bournemouth.ac.uk/jmacey/MastersProject/MSc17/02/FGalluzzi_Master_Thesis.pdf

Edge-Enhancement – An Algorithm for Real-Time Non-Photorealistic Rendering: http://wscg.zcu.cz/WSCG2003/Papers_2003/D73.pdf

Towards Real-Time Photorealistic Rendering: Challenges and Solutions: https://dl.acm.org/doi/pdf/10.1145/258694.258701

Nuke – GreenScreen Motion Graphic

During one of our sessions doing greenscreen, we went and filmed on our green screen studio a set of two videos and then were taught how to green screen these out by using the keyer nodes and creating a garbage matte. We also learned about then using these to build out a 3D Scene and creating enviro nodes based off of videos. I originally wanted to do a different area to greenscreen in, but had some issues regarding the environment nodes and getting it into a 3D space, so i chose to go about coming back to that task later with advice from my professor.

This shot had already been keyed and had a lightwrap applied, so i went for the motion graphic of adding something to the screen when our filmed student would point. I went about building out a small board using constant nodes and ramps as well as transform nodes, and using a switch between both using constant and linear keyframing in the dope sheet.

A lot of the keyframing of this came down to going between copying the keyframes and using a constant 2 frame animation style, unfortunately this isn’t done through simple ctrl copy and paste but after a lot of headaches and google searching i found ways to do both this and opacity keyframing.

I used some opacity keyframing on both the underlinings and text on the small weatherboard

I then set keyframes on the actual merge using the mix option as an opacity.

Personally i really do not like using Nuke for almost anything motion graphics related. While i did like doing motion graphics again, doing things in Nuke takes an absurdly long amount of time to do the simplest tasks even if you know exactly what you’re doing compared to something like after effects. This coupled with the fact that after effects still has illustrator compatibility between creative cloud leads to this having more control but at the cost of time. Doing something this simple motion graphics style ended up having me looking up a lot of different ways in nuke to essentially get around simple things like using an opacity gizmo or not being able to copy keyframes as easily that bog down this process overall. While Nuke could be used for motion graphics i would most likely end up using something like after effects overall.

Below is the animation, though the animation has been converted to web res at 1280×720 and the bitrate has been dropped due to converting this in media encoder for web.

Week 2 – FMP

I’m currently looking into researching cel shading in 3D Modelling. I’m currently researching this topic as it has become more prevalent in the last two decades of media, with primarily games using this form of shading of 3D modelling to create an art style reminiscent of early comic books. Some cell shading can also be similar to cell shading but not actually be truly cell shaded in technicality, as cell shading refers to stripping shaders in the place of flat colours and shadows as defined by TVTropes. Source: https://tvtropes.org/pmwiki/pmwiki.php/Main/CelShading

Some of the examples of cell shading in media have been in both video games and in other media like shows and movies, some of the earliest examples were Jet set radio (2000) The Iron Giant (199) and as Recent as something like Arcane (2021) and Legend Of Zelda: Breath Of the Wild (2017).

I feel it’s important to research this subject as it’s both something i’m interested in and would like to use in future projects, but also highlights achieving different art styles within the media of 3D Modelling and experimenting with things like shaders to get more stylized views of 3D Objects. Usually 3D Models in media and animation sometimes adhere to a very similar artstyle as they are rendered in similar ways with similar programs, and using something like cel shading can be used to combat this feeling of sameness in media and allow for different feelings and looks through new projects.

Some of the keywords that i used in researching this topic were “Cel Shading”, “Cel shading in media”, “Cel shading Animation”, “Cel shading 3D Modelling” and “Cel Shading in history”.

Four sources i found for this were a Dissertation on Cel Shading: http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.691.5831&rep=rep1&type=pdf

2D Shading for cel animation: https://dl.acm.org/doi/abs/10.1145/3229147.3229148

Enhanced cartoon and comic rendering: https://diglib.eg.org/xmlui/bitstream/handle/10.2312/egs.20061047.141-144/141-144.pdf?sequence=1

And advanced real time cel shading techniques in opengl: https://www.cs.rpi.edu/~cutler/classes/advancedgraphics/S12/final_projects/hutchins_kim.pdf

Researching this is important to my project proposal as i’d like to do a 3D Animation for my project and have been researching different shaders for this purpose to go with a stylized 3D Animation. Cel shading is one of the styles i have been looking at and may want to pursue further.