Lumitone
Introducing Lumitone
0October 23, 2011
*This is a 3D video. Please select your required 3D format and a higher resolution on the video control bar once playing. If you do not have an NVidia 3D display or some classic red/cyan 3D glasses you can watch the 2D version by selecting ‘Turn 3D off’ on the YouTube control bar. If you are using a portable device you will need to watch the 2D version of Lumitone below.
The second chapter of the Music Box Chronicles is an explosion of light and colour. Lumitone is the story of a lonely music box who dances to electric dreams and sings with spectral audacity.
Written, Animated and Directed by Aaron Bradbury
Music: Minuit Jacuzzi (DatA Remix) by TEPR
Lumitone is a follow on from Twinklebox, a short film about the mechanics and music of time. The Music Box Chronicles is a collection of films exploring image and sound.
Visit the project page at http://www.luniere.com/projects/lumitone/
Read more about the ideas and production process behind Lumitone in the production blog here. To view Lumitone in a wide variety of other 3D formats, please visit 3DF33D.
Lumitone in 2 Dimensions
S3D Streaming Services
2October 13, 2011
in Lumitone
A work in progress still. Almost done.
I’ve started looking around on the Internet for different 3D services to decide on the best streaming method of stereoscopic videos. Currently YouTube is the best option, NVidia’s 3D Vision Live looks interesting as well as 3DeeCentral. I might also just make the over/under video downloadable and point people in the direction of Stereoscopic Player, but that seems a bit too big a deal for most people to bother. Streaming is definitely the future, but it’s not currently the best experience.
Post Effects
0October 12, 2011
in Lumitone
This is a 2 frame comparison of the kind of effects that I’m going with.
The amount of effects on this project is far less than Twinklebox for several reasons. Firstly, this project has had more colour considerations during the production process, so most of the colour preparation is already there. I want the colours to remain vibrant, so the less crunching and shifting that is done here, the better. The stereoscopic considerations also feed into this process as some effects will affect the stereoscopic outcome. Depth of field can flatten the stereoscopic effect slightly but also adds to the perceived depth as a different cue, so it can still be used to great effect, allthough due to the dynamic nature of the camera in this project it felt a little jarring after several tests so I’m probably not going to be using any. A little vignette-blur helps with the negative parrallax cropping at the edges of the frame. This is something I would have applied anyway to give it a slight vintage lens look. Chromatic Abberation wouldn’t be a problem if everybody was watching on active displays, but when viewing on YouTube as an optimised anaglyph, this will cause some issues as it’s basically shifting the reds and blues in a similar way to the anaglyph effect. The compounding of these two shifts can be awkward to converge. Any 2D effects are just that, 2D, so they wont appear to be volumetric if that is the usual illusion when applied to 2D footage. Most 2D effects, such as glows or blurs are okay to use, but Trapcode Shine for instance will appear as a flat plane, even if it has been offset to push it into positive or negative space, it will still appear as a flat plane. There are some glows at work here to give a more naturalistic film blooming effect and tie the particles into the scene properly.
Stereoscopic Resolution
0September 26, 2011
in Lumitone
After some resolution tests i’ve been doing for another project, I’ve come to realise the importance of resolution for sterescopic work. I hadn’t realised before just how much stereoscopic information was lost between the pixels. There’s a number of ways to increase the sense of depth in any given shot, but a basic one is to simply increase the resolution. That is, if you have a display that is capable of displaying the additional resolution.
I *was* working to 720p due to the render times per frame and the maps not quite holding up at 1080p. After doing some thinking and having a cup of tea I came to the conclusion that most people who will see this film will be watching it online, or on a TV, and most of those people who have a stereoscopic display will have a full HD display, as the two usually come hand-in-hand. So, if you’re going to do a thing, it’s worth doing that thing properly and thus began the conversion of the project in HD.
This isn’t as simple as it sounds. First of all I had to re-render the final beauty pass, pretty simple, just some increased values to compensate, shadow samples etc. Then rerender an uber low res RPF pass for the new camera positional data… In After Effects, the RPF camera import data changes with the resolution, even if you’re not changing the camera position. Then every comp, solid and adjustment layer needs to be replaced with a higher resolution version. Any pixel based effect needs to be multiplied… blurs, mask feathers etc. All the particle comps need to be converted to a higher resolution to hold up. Every 3D layer needs rescaling to match. The new RPF data needs importing as well as any positional data exported from 3dsMax.
It’s not the quickest change to make, but the outcome is revealing, especially when looking at the stereoscopic depth. It’s interesting how much extra information is percieved in depth just by increasing the crispness of an edge. Therefore it’s worth pointing out that the anti-aliasing settings also have an impact on the stereoscopic outcome.
Below is an animated gif showing the difference between the two resolutions, when overlayed on top of each other. It’s not a perfect example as there are subtle changes in random particle positions due to the change of resolution in the comp as well as a slight change in convergence and lower anti-aliasing settings. Nevertheless, it is still possible to see more levels of depth throughout the image. Just seeing one or two extra pixels of accuracy in the parrallax is enough to place an object slightly infront of or behind another.
This amount of subtlty seems pretty much lost when considering the pace of movement in these shots, you don’t get any oportunities in the film to browse the depth at your leisure, but it definately increases the feeling of space when comparing the alternative resolutions as video. As your eye is thrown around the frame in sharp jolts, the brain still manages to build complex pockets of depth around the focus area. This is my experience anyway.
Another resolution comparison:
A selection of anaglyph stills at 1920×1080 resolution (Click to see full size):
Stereoscopic Perception
1July 16, 2011
in Lumitone, Stereoscopy
After a few experiments with altering the logic behind how close or far something is in stereo depth, it seems as though the stereoscopic information isn’t given as much weight as other depth devices such as scale, and position, or perhaps my current understanding of depth perception is lacking. There seems to be a lot of give on how much your brain ‘listens’ to stereoscopic information, when it conflicts with our understanding of scale.
In order to achieve a specific stereoscopic result in my animation, I realised I needed to create a scene that did something a bit odd, something that didn’t make much sense to me logically. I had to make an object appear further away stereoscopically when the camera is close to it, and then make it appear much closer stereoscopically when the camera is further away from it. This doesn’t seem to make any logical sense, and my first thoughts were that it would feel wrong but might give an interesting ‘dolly zoom’ look. What actually happens is very little, the scene doesn’t look wrong, or even odd. It actually looks very similar to how it did before I changed it, I had just removed the awkward negative edge cropping that was previously apparent in the close part of the shot.
When analysing this, I’m not really sure why this is the case. Depth perception is a very complicated process in the human brain, far beyond my current understanding, so I think I will have to do some more research into this to get my head round it a bit better. For now though, this is another technique that can aid ‘brain shearing’ issues, whilst retaining depth in a scene.
French Electro Box
0May 8, 2011
Making stereoscopic French electrophonic Twinklebox.
Using a bespoke version Anaglyph Previewer made by Martin Dufour. It’s been modified by Martin to work with a specific stereoscopic camera rig I made a couple of years ago before one was implemented into the 2011 version of 3ds Max. I’m still running on the 2010 version so my old camera rig is coming in useful again. Unlike Maya, 3ds Max doesn’t have stereoscopic viewports yet, so being able to preview using Anaglyph previewer is incredibly useful. Due to the restrictions within Max it can’t play back in realtime so you have to update it manually per frame or render a preview to disk. Nevertheless, it saves a huge amount of guesswork and time.
I’m mainly experimenting with camera movements in this one, and trying to push the boundaries on what’s possible stereoscopically. I’m almost certain that the way the camera is currently animated will make any humans eyes bleed, so the challenge is to soften the stereo effect without losing depth. I’m going to try smoothing the focal plane position so it stays in a similar position in 3d space, just in front of the music box, and allow the particles that will be emitting from Twinkle Box to break the usual convention of passing through the cropping frame whilst in positive space. After some tests this seems to work relatively well. I will have to do a lot of heavy animation of the convergence in post due to all the particles being rendered in After Effects. I’m currently exporting camera positional data in After Effects using RPF’s and then importing positional data for music box using another script called AE_Position_Exporter_v1.0. This means that if I don’t want to use the toe-in technique for rendering the cameras I will have to animate the convergence in After Effects as the only way I can converge the cameras in 3ds Max would be to add a skew modifier to each camera, which wouldn’t carry through in the RPF render data. I’m pretty sure the way I’m doing this will work.
More soon.
