Aston 7 character generator
Had a look at the new Aston 7 machine. It has created a useful interface and is being used for some real work now. The Aston team have wrapped the Brainstorm 3D engine with an interface that reflects the older Aston interface which was quick and simple for operators. It doesn’t expose all the functionality in Brainstorm, but tries to give a regular interface that previous Aston operators will find comfortable. These are just a few comments that I took down during the demo and it doesn’t claim to be comprehensive.
The interface is a click and drag interface with a toolbox of elements that can be brought into a graphic. Materials, colours and resolution can then be adjusted in the interface with a context sensitive properties window. Single objects or groups can then be dragged back into the pallete area to save a standard item or a template for use in other pages. The view is a locked off camera and the animation is done by moving the models in the scene. Lighting looked good with point and spotlights, with and without attenuation. The lights are all global at the moment but selective lights are being added. There are no shadows yet and no physics which are in Brainstorm. Primitive objects such as spheres, rounded cubes have adjustable complexity so the operator may balance performance with quality. It is easy to texture cubes on a per side basis, a global selection may be over ridden by selections for particular sides. There are some blending optiopns but I’m not sure that all those allowed for in Brainstorm come through. Not as obvious as the Photoshop blends that we are used to. This may well be a later addition.
All their pages and data are XML files. There is no database involved. Assets are held in folder structures and there are tools to make sure that the on-air boxes are refreshed with any new images as they are needed. The fonts on a page are logical fonts; they can be defined as part of a project along with colours etc. Shared assets in a set of pages can be updated and they will update the next time the page is loaded. The pages are reference as a four digit number adn can be recalled quickly using their interface. This will suit the Aston operators of old and people used to vizRT Trio systems. The keyboard interface is very reminiscent of the older Astons and has some quick text handling that operators like, character kerning, movement of selected bits of text; the scale to fit option can have a combination of X scaling or XY scaling as the operator chooses.
Being Brainstorm driven, the graphics are state based, meaning that the system can move from one state to any other at any time. There is no fixed timeline. This can be a benefit for flexibility but it can be difficult to produce some types of graphics where the path of the animation is fixed and can be previewed. The designers are used to timelines, so they may have difficulty here. There was talk about some type of timeline being brought in, but it would be based around a series of events rather than a fixed animation timeline. The cool facility in Brainstorm, to create image buttons by dragging and dropping states to a window is not available in Aston, although they thought that it might be possible to open up. They don’t really have the content pilot type functionality without this. The file formats for 3D objects didn’t seem to be as full as Brainstorm, but this is just a mater of letting it through the Aston interface.
The operator had a few problems when trying to use masks and opacity with textured items. This might be a lack of functionality from the Brainstorm interface, or a bug; it is available in Brainstorm underneath. There is no scripting at the moment. There are some actions that might be triggered when data changes but these are relatively simple value comparisons. Any more complicated logic has to go into an external control application. It would do the logic and then provide the Aston with data in a simple table. Strangely, the Aston needs external apps to act as a data server. (??) When the graphic changes it polls the app for the latest data. I’m not sure how this would work in an app that drives the machine completely as in the sport controller case. We would need to tell the machine to load a page and then the page would ask for the data. We shall look at the API docs in more detail.
Tickers seemed to be easy to make, with multiple fonts and could have 3D objects placed in a font for use between normal text items. Looked quite smooth although we only saw 2D onjects. The data form that fed the ticker is read as it takes each line on screen, so a change to an item in the list will be updated on screen as early as possible.
A number of pages may be on screen at the same time, allowing graphics to be mixed anywhere on the screen or even overlapping. The system has a series of events that can be generated that will fire a state change in other graphics. This allows a strap coming on to fire an event to squeeze up a full page graphic to make room for it. I’ll be interested to see how effective this is and how complicated it gets as the graphics interact. It compares to transition logic in the vizRT machines. Video clips are usually Quicktime or AVi but the system uses ffmpeg internally so any codec to suit that should be possible. The performance of the system looked OK, but Aston are still working on a numeric display to show render time per frame which will help to judge the complexity of the page being built.
The screen layout can be configured for different operators; the windows are just moved around and resized even over two monitors which is welcome. The playout operator is able to get thumbnails of the pages in this wider layout with large page numbers and text. The assets folders can then also be made larger with search facilities that might not be used in designer mode.
We haven’t seen the automation interfaces, but Aston and Softel the parent company are well used to these. They would probably be using the Brainstorm versions which we’ve seen previously, but it would be good to see it in action. Hardware wise, the system is based around a PC with nVidia quaddro cards and an AJA card for video in/out. This card gives a couple of SD inputs; maybe only one in HD as there was a question of what the performance hit would be. It would be a box (4U) per channel, but design stations could sit on a normal machine and link to the main boxes remotely. This could give a cheaper overall feel to the system. The system is definitely getting useful, but has room for improvement. I like the engine within it, the ideas in the interface look useful but I need to look more at the detail to see the quality and speed of use.