>>23949
>Wouldn't Unity at the baseline drive up the graphics requirements? Like GFX acceleration always on?
Since the application is targeted at users who use programs like Photoshop the graphic requirements can be relatively high. I'm more concerned about ram usage. Photoshop has a feature called scrap disk which uses you hard drive if it runs out of usable memory.
>As this whole project would be web player or webgl based, I think users would have to download the whole app each time they use it.
The app either exports an html, image file or a project file which you can import with the program. Maybe I can even export a 3D model, but the application itself will be locally installed.
>Prepare whatever font you want to use. If you want to use different/variety fonts at runtime, I'm sure there's a way to do it right, but I think most quick implementations will cause pixelated fonts with strange spacing if they're not prepared beforehand.
I really need to experiment in unity here, there's nothing on the internet about font rendering at all. There seems to be something called Dynamic Fonts which sounds like what I want to have. At least I now know that Unity supports both .ttf and .otf files.
>>23950
>Also it'd be prudent to implement some way to generate and mix together imported textures into an atlas at runtime and find ways to lower memory usage and draw calls if you're letting people compose scenes with all these different objects. Might want to limit the size of textures or dynamically resize them on import somehow.
Most objects are going to be simple 2D shapes with rgb colors without any textures which change their form at runtime. There will be a handful of prefabs with basic functionalities like text forms, buttons, lists and so on. Are images a big choke point? They don't need to be big, but depending on the project there might be a hundred images (which are not all visible at the same time though) There will also be a limit on how close you can zoom in.
>If you're letting them import 3D models too, you could create a classification for different objects (human/prop/etc...) and maybe use low footprint placeholders when setting up scenes... only showing the actual model when they hit a preview or render button.
It's something like this. Most of the time the user will use a '2D' mode that works like a canvas in most graphic programs. The only time it's going to be 3D is a preview function like pic related. There's going to be a predefined selection of 3D models so importing new models is not really important.
I've been working through some tutorials, the manual and some other resources now and I'm still clueless as to how I'm going to manage memory usage efficiently. I guess I have to find out the hard way. I'm still pretty gung ho about this, it's going to be a fun project.