Repository navigation
Replies: 1 comment
|
There's really no way to do that at the moment. The Canvas (working area) and all character motion, for example, are based on offsets from the current cursor position at the time of execution. Character appearances, such as gradients, are calculated across the grid of the complete text input. Most effects are actually interacting with all characters simultaneously. Rendering the final text appearance occurs within a two-dimensional list whose size is limited to the input text. Etc. Supporting streaming text is a cool idea but would require a bottom-up redesign and the resulting effects would be strictly limited in diversity as characters could not interact with, or base their actions upon, other characters. All that being said, you can get a rudimentary version of this by re-instantiating the effect with each chunk of text. You will need to logically organize your chunks by line. In addition, the next line cannot start until the previously line has completed. So, for something like the Slide effect, you could not have multiple lines sliding in together, they would need to occur sequentially, which is not as visually appealing. Thanks for the question and we can discuss this concept further if you'd like. |
Uh oh!
There was an error while loading. Please reload this page.
How do I use TextEffects (even a restricted set of them) with a Python generator that provides chunks of text at a time? Can I please get a minimal example? Thanks!
All reactions