Anyone paying attention would probably have noticed that we keep talking about the importance of workspace resolution vs final output resolution. And how in many situations they can be pretty radically different.
When you are designing a visual effect or artistic image, you really want to do it in a workspace resolution that is comfortable to work with. This part of the process is all about designing the effect, designing the overall appearance of the artistic image.
You also want a working resolution that actually corresponds to your visual perception of the overall whole artistic image. The spatial frequency of your visual perception is limited. This is the whole reason why image compression algorithms even work, you can't see the fine detail past a certain point unless you zoom in.
When you are done with this artistic design process, then you are ready to generate and output whatever your final output resolution needs to be. At whatever higher resolution is needed for the artistic task you are pursuing.
But in many situations, doing the design work at your final output resolution is a recipe for misery. Even worse, it can lead you astray.
Let's consider the Vectorizer for example. Running a Vectorizer effect on a 1 or 2K canvas size can lead to radically different results than running the same Vectorizer effect on a 8K canvas size. And in 99% of potential situations, running the Vectorizer effect on the 8K canvas size is going to be the incorrect artistic decision. Compounded by the fact that it is going to run significantly slower.
We have had various approaches to this dual design at a working resolution, output at a final resolution process over the years. From the very beginning, Studio Artist had the concept of recording what you do in a Paint Action Sequence (PASeq), then re-rendering the PASeq into a higher resolution canvas. Studio Artist from the original V1 release having a unique hybrid vector-raster graphics model that supported this way of thinking.
Of course the paint synthesizer quickly eclipsed the original raster paint nibs laid down on vector paths model. For example, some paint presets appear to have a certain size brush to an artist using them, but internally you are working with a cloud of multiple interacting particles (virtual brush). Pure vector paints and Region Fill as Brush pen mode are additional examples of current paint synthesizer techniques outside of the original raster nibs laid down on vector paths conceptual re-rendering model.
And the endless opportunities for organic randomization in all of the various Studio Artist processes can sometimes get in the way of successful re-rendering strategies.
SVG vector file output was another feature introduced to help with this process. Any vector effects in Studio Artist can be output into a resolution independent svg vector file. You can then import the svg vector file into any program that supports it, and generate raster output at whatever final resolution you desire.
You can even import the svg file back into Studio Artist and render it up to a higher raster resolution directly in Studio Artist if you so desire. This was limited prior to V5 due to the 32 bit nature of the older version. V5 had a 64 bit build for mac, and V5.5 is 64 bit on both mac and windows. So this is actually a very useful strategy in V5.5, if the particular effects you are working with support it (vector paints, vectorizer, some IpOp and MSG effects if they are vector rendering in nature).
The way this works currently places a high conceptual burden on the individual artist. You need to think through exactly what you are doing, and then keep track of that as you build your effects, build a PASeq for them, render the PASeq out to a svg file, then re-render the svg file appropriately depending on where you finally want to end up. Because of this conceptual burden, many artists either don't bother, or get confused by the whole process.
Another factor that we dislike about svg is that the svg file format was really not designed for how we are using it. My take is that they were thinking about simple things like the Linux penguin when they were thinking about how people would use the svg file format. Not the kind of complex organic looking vector imagery that can easily be created in Studio Artist. The side product of this is that the svg file sizes can get huge for vector artistic imagery of the kid of high level complexity that Studio Artist can trivially generate with even a single Vectorizer action. So svg is not necessarily an ideal file format for the kinds of things we are trying to do in Studio Artist.
Using this output to svg and then rerender afterwards strategy is also limiting when you start talking about building output imagery that is the end result of an iterative animation process that may play out over multiple frames in an animation.
And of course the svg file format ignores any raster processing or raster painting going on inside of Studio Artist. Many raster effects can actually be scaled to do appropriate things as the canvas size increases. We want to make sure we have an easy to use re-rendering strategy that works for everything you can do in Studio Artist, not just a more limited set of purely vector effects.
So i have briefly laid out one of the design tasks we are trying to address in the next Studio Artist release. In the next post in this thread i'll give a status report on some progress we are making with the whole notion of building visual effects at a working resolution and then re-rendering them to a final output resolution. So it will be a quick glimpse into the exciting future of where Studio Artist is headed.
Replies
We've been working on all kinds of different cool new features for the next version of Studio Artist. Let's take a look at one particular example, which is all about coming up with better solutions for the workspace resolution vs final output resolution dilemma i previously discussed in this thread.
What we are doing involves directly expanding on the original Studio Artist design strategy, which revolved around using the rescalable nature of vector bezier paths to build rescalable raster paint effects. But our new efforts work to expand that concept out to as many different Studio Artist effects as it can possibly be applied to. In new and different ways.
And doing it in a way that allows all of the work to be done directly in Studio Artist. While at the same time making the conceptual burden on the individual artist much lighter. And doing everything just using the existing Studio Artist Paint Action Sequence file format.
So the goal is to just make re-rendering a PASeq work transparently for everything you can do in the program. As well as allowing effects to be 'bezier encapsulated' at any working resolution (if that is appropriate for the kind of effect you are working with). Once a particular effect is bezier encapsulated, it then subsequently plays back directly derived off of the internal encapsulated bezier paths that define it. And the effect can be rescaled to any output resolution you desire at that point.
So if you bezier encapsulate a Vectorizer effect, then afterwards the first several vectorizer control panels are effectively ignored when you run the effect (the ones associated with generating the vector shapes), and only the later control panels directly associated with rendering the vectorized vector components are run (the encapsulated bezier paths in the effect have captured what the first several control panels generate).
This is very different than what happens when you render a vector effect into a svg file. For example, a single vectorizer rendered vector shape might consist of a fairly concise bezier path that defines the shape, but when the vectorizer renders it into the canvas, capturing that complete vector rendering process in all of it's internal complexity in a svg file leads to a ton of vector information stored in the svg file (stored in a very inefficient way in the svg file).
With the new approach we are pursuing, that additional complexity is generated by Studio Artist at whatever canvas resolution you are rendering to, but it is all based on an active generative process derived off of the initial encapsulated bezier path. Based on the vectorizer's editable parameters for the generative rendering process that is derived off of that initial encapsulated vector shape. So we only need to store the vectorizer parameters and the single encapsulated bezier path in our file format, but we generate the full rendered vector complexity whenever we render it out (into whatever resolution output canvas we want).
Here's a few quick examples to show off what i'm talking about.
This example shows the output from a 100 frame PASeq animation. I generated the PASeq for the visual effect using a 1280 pixel working canvas, and then bezier encapsulate it.
There are 3 different version of Studio Artist output for the final frame of the animation in the image above, (shown in the Preview screen capture at 1280, 4K, and 8K resolution).
Now what about PASeqs that combine vector effects with additional fully raster dispersive effects. As i previously stated, many raster effects actually can be re-rendered. Some parameter(s) associated with the raster effect might have to be adjusted. The re-rendering process for the PASeq does that automatically.
Again, we're showing the output for the PASeq created in a 1280 pixel working canvas at 1280, 4K, and 8K output resolutions. So the vector components are scaled appropriately, but alos note that the raster processing effects were also scaled. So the appearance stays essentially the same, but textural detail associated with these raster processes is being generated at the full resolution of the output canvas. This detail can actually be fractal in nature depending on what you are doing with the raster dispersive effects.
This next example showcases something truly unique to Studio Artist, a hybrid vector-raster geodesic effect based on bezier path embedding and anchoring. Everything associated with the effect is raster in nature. But because we anchor the raster processing to an encapsulated bezier path, the raster effect scales to different resolutions while still staying crisp at the final resolution it is rendered to.
I can run a 100 frame animation on any of the 3 effects shown above in about a minute in a 1280 pixel canvas. So i can quickly try out different artistic design decisions, and then see the results of those design decisions interactively.
However, if i naively started with a 8K canvas size to do this design work, i would quickly get very frustrated. For 2 different reasons.
One because everything runs much slower for many of the kinds of operations i'm working with.
The other reason why is that running the effects using a full 8k source into a 8k canvas. will give me something that does not look nearly as good. My 1280 working canvas is actually much better suited for designing the effects. With the new approach we are taking, those visual effects designed at 1280 resolution are then fully rescalable into 4K, 8K and even larger output resolutions.
We're still putting all of the internal pieces of this puzzle together for you, but i wanted to give this little status report to give you an indication of where we are heading.
The end goal is to make the whole process as transparent to the individual artist using Studio Artist as possible.
this sounds very impressive