Growth Visuals

Now that the timing and content is close enough to being finalised, I have spent some more time polishing the visual style of the cells.

Feedback has shown that the visuals are a bit 'murky' at the moment, and there needs to be more definition - something to make the cells 'pop' and stand out. The cells move in a really unique and interesting manner, so it is important that this is the focus of the viewer (so that all my hard work can be seen!).

My project supervisor suggested I take a look at a DVD in the library by Jeremy Vickery, titled "Practical Light and Colour" (more information here). This DVD discusses the fundamentals of light and colour, and how a better understanding can improve practical work. If anything, it serves as a fresh source of ideas and inspiration, which will hopefully develop my cell aesthetics and drive towards the best outcome possible.

After watching this DVD, I started experimenting with colours and light in Maya, trying to develop something which was much more interesting, less murky, and had more focus on what I wanted the viewer to see.

Below are two renders; the first showing a still from Growth before this DVD/experimentation, and the second showing a new version, which uses entirely new lighting, colours and Maya shaders to give a different style completely.

I have yet to test this new style on an animated sequence, but I definately prefer the clarity that this updated version offers. The cells have significantly more contrast (against the background) and are less flat - by adding an edge glow, they really draw the viewer's attention.

Growth Progress Update

Since my previous post, I have continued working almost exclusively on 'Growth'.

All of the cell visualisation stages in Maya are completed (e.g. scripting, dynamics and render setup), and all the camera work has been (pretty much) finalised. Most of the rendering has now taken place, and I have spent a large amount of my time compositing the different shots and passes, using After Effects.

This uncovered several problems in the 3D work, such as dust particles floating in front of the camera, or 'through' the cells (which looks like they are being absorbed). These issues have taken longer to fix than expected, as it has meant rotoscoping out certain areas of some image sequences (primarily the depth passes). Although this is a time-consuming fix method, it is certainly much faster than recreating the particles and their movement in Maya, and then re-rendering several layers to be composited again.

One of the main issues I am currently experiencing is the use of depth of field in my shots. I want a really shallow depth of field, giving the correct impression of (macro) scale. After Effects has a feature called Lens Blur, which allows me to composite using Maya's depth pass. Unfortunately, this feature is quite limited in terms of the amount of effect that can be applied, and isn't enough for what I am trying to achieve.

Because of this, I have started setting up one of the shots in Nuke. I tested this by using After Effects to render out a composited sequence, and applied the depth using a ZBlur node in Nuke. I was much happier with results that Nuke gave (and the use of the 'focal-plane setup' feature) so I have decided to rebuild the shots properly, using only Nuke.

Since most of the composition is blocked out and timed in After Effects, I will render each shot entirely out of Nuke, and take the completed shot sequences back into After Effects (which is the most straight forward way of working for me). The previous alternative would have been to create the shots in After Effects, render out composites for Nuke, apply the depth and tweak the shots, then bring them back into After Effects - a much more complicated process. I still want to complete the final project in After Effects, as I am much more comfortable laying this out on a timeline (which is already done anyway).

As part of my testing, I have generated 3 versions of the same frame to show the visual differences. The first is the original composited shot, rendered out of After Effects. The second image shows the outcome of using After Effect's Lens blur to add depth of the field. The third image shows the outcome when using the ZBlur node in Nuke.

One thing that became apparent using Nuke, is that not only did it allow me to apply depth of field in the way I wanted, it created less banding in the rendered TIFF files. Banding is something I have struggled with so far (because of the subtle variances in colour), so this is another advantage of using Nuke!

Moving forwards, I plan to continue compositing the shots initially in Nuke, and then bring everything together in After Effects. This should be the best way of achieving the look I want, and avoiding the banding that was present in After Effects.

The next step beyond this, is starting work on a 'making of' video (for the degree show), which will contextualise the content of Growth, and provide more information on the methods used during it's creation.

Growth

Over the last two weeks, I have been developing the concept for the final outcome of my Masters project. This will take the form of a short video (approximately 3-4 minutes in length) and is intended to showcase some of the cell visualisation work I have completed. In addition to this, I plan on creating a second video which will be a technical showcase of my skills and abilities and also contextualise the work I have completed.

Although I don't want to give away too much, the main piece will be split into two halves, titled Growth. My title concept can be seen below;

As a sneak preview, the video below shows one of the shots I am currently working on in After Effects. Using the VERL render farm, I have already generated most of the content I plan on using (in 1920x1080 lossless TIFF format) which gives me a great deal of flexibility. The original shot did not feature any depth of field and looked flat and uninteresting, whereas this updated version has much more style and visual interest;

So far, I haven't decided if I will include the dust motes/particles idea which I previously worked on. Although I have considered this option, I don't have any sequence renders which I can use, so I took a still frame and applied the effect, as a test. This frame test can be seen below;

Moving forwards, I will continue developing the look and feel of these shots. I am also currently looking at using music to add impact to these shots, but this is an ongoing project and I plan on trying to finish work on the visuals first, so that I can get a good 'feel' for the experience I want to create.

Stereoscopic Update

Following on from my early tests of creating stereoscopic content out of Maya (here), I have completed rendering and compositing of my full cell visualisation sequence.

The process of working with a sequence instead of a single frame was not hugely different, and I feel confident that I could apply an anaglyph effect to a scene in the future. Currently, as it is anaglyph, I have been working in grayscale, but moving into colour is something I would like to achieve.

I am planning on meeting with my project supervisor and programme leader, to discuss the possibility of showing stereoscopic material during the Masters show in August. I am not sure if this will be a viable solution,  but it is worth looking into...

Falling Out

Matt Cameron, one of my classmates, is currently directing a short film titled "Falling Out" (production diary here). The logline for Falling Out is below;

A pregnant woman struggles to find direction after a nuclear blast decimates Scotland. Locked in a bunker with a jaded survivalist and her dying boyfriend she faces tough choices in this brutally changed world.

I will be assisting Matt in creating some of the visual effects which will be added to this film - in particular, using 3D computer graphics and animation to create a dynamically-driven nuclear explosion which can be composited onto a live plate at the beginning of the film.

So far, I have recreated the camera setup in my 3D environment (so the CG cameras match the real-world cameras), and completed some early explosion tests. Over the next few weeks, I hope to tighten up the explosion (which is not nearly nuclear enough) and generate some nice high-definition renders which will be suitable for the opening shot of Falling Out...

Composited Sequence

As part of my cell visualisation project, I have been working on a second set of data, which also represents the growth and development of cancer cells. This dataset however, focuses on the stage of which the individual cells are in, with the intention of identifying the most effective time for treatment to take place.

With thanks to one of the mathematicians working in this area, I was provided with a large quantity of data (around 33 million lines of numerical information) which had been generated, and could be used to drive my 3D visualisation. I created a MEL script which could 'read' each section of this data and then generate/animate the appropriate objects in 3D space.

After completing this stage, I then setup the scenes for rendering - adding lighting, placing cameras, adjusting render settings and creating render layers. Using the VERL render farm, I generated approximately 33,000 renders which could be composited together to create a completed 3000 frame sequence (depicting 600 hours of cell growth and development).

Using After Effects, I combined all of the image sequences and created several individual compositions which would be layered to create the final output. Because almost all of the work had been completed in Maya, there was very little that had to be done in After Effects, except combine the appropriate layers. Rather than generate a QuickTime MOV, I rendered a master image sequence (after speaking to one of my classmates) which could then be used to create suitable video files as and when needed.

Because the source of my output comes from unpublished mathematical models, I am unable to post any of the video online (it would be unfair to the ongoing research still being carried out). Instead, I have opted to include an image of a single frame, which can be seen below;

Over the next few weeks, I will be meeting with the mathematician that I created this for, as there are several upcoming maths conferences and events where my work could be used to illustrate the research they are carrying out.

RealFlow Character Fill

Over the last couple of weeks I have been working on a shot for one of the other students on my course. Kaye is working on applying an illustration style to an animated commercial, and her blog can be found here.

After creating her animatic, Kaye's opening shot showed a 3D character filling up with liquid, and she had decided to use RealFlow's liquid simulation tools to achieve this. I already had previous experience of working with RealFlow and was able to help her in building this sequence.

To begin with, I imported her character model (built using Cinema 4D) and created emitters in each of the feet. My target was to fill the model in approximately 180 frames, as this is what Kaye had advised. I started by filling the model using these two emitters, and adjusted the emission speeds to suit the fill rate I wanted. Around halfway up the model, it became slightly more difficult, as I had added two additional emitters (one in each hand) and then needed to try and balance the levels of the fluid (this involved manually keyframing each emitter and re-simulating the liquid several times, a time-consuming process). Upon reaching the head, a fifth emitter was added, to speed up filling the volume (as the narrow neck creating a fountain-like bottleneck). Finally, a polygonal mesh was added, and the radius size adjusted so that the mesh was not larger than the original model.

At this point, I passed the source files back to Kaye to implement in her own project. As an additional step I imported the RealFlow files into Maya and created a simple render to show the outcome of this short project;