Last post by nih - September 17, 2019, 05:11:26 am
My 16k world build was taking around about 7 and a half hours. Kind of painful to sit and watch it, even with it running fairly unobtrusively in the background. It also means some I'm memory limited I have to output the world at various stages to file since I don't want to regenerate the whole thing for a small tweak down the line and I'd disabled retaining interim steps to save memory.
A friend suggested I set up an on-demand AWS instance, so here I am with a 96 core 180gb memory instance that costs $8 an hour to run and takes 30 minutes to build my world. Neat. All I have to do is remember to turn it off afterwards....
I can really recommend this. Setup only took about 40 minutes and it's all hooked up to dropbox.
Last post by mattnava - September 15, 2019, 02:51:43 pm
World Machine is a brilliant program. I think that it's core structure and approach is the most forward thinking of all the terrain apps out there. But the UI is outdated, and is holding it back.
I think it's time to modernize the program with a complete overhaul of World Machine's UI.
It's clear that the interface was started a time long ago and was continuously modified. The result is a UI with elements from different eras of development all slammed together. Just from looking at it, you can see that as new features came online their UIs were done in slightly different ways. For example, there are many kinds fonts that don't visually go together-- they are different sizes, some are anti-aliased, some are bold, etc, without a seeming hierarchy. Most importantly, the way that you interact with the same kind of UI in different parts of the program is inconsistent. There are some sliders that allow you type in a number and some that don't; some that allow higher levels of granularity when sliding than others. Some things require a double-click on to open their property windows; some things require that you select them and then hit a button on the sidebar to open their property windows. There's a whole lot of similar issues across the board.
The UI lacks a clear, consistent direction for its visual hierarchy and user interactions.
The UI is functional but it is unfocused. It's may have started with a clear design but after accumulating so many elements it's direction has become unclear. This makes it both non-intuitive for new users and frustrating for pros. The lack of an intentionally designed visual hierarchy, the breadth of organizational ideas, and the non-unified UI interactions make it hard to predict where to go to find what you want and how to interact with it. A large part of mastering the program is simply building a mental model of all the different ways to use similar elements of the interface itself.
It probably has never been possible to prioritize the interface over adding new terrain features. Understandably, UI is not as "cool" of a feature, and on a small team it can be hard to allocate resources to focus on it. But it is important. Improving it has exponential gains for users. Using the program requires hundreds of interactions with the UI an hour -- if you can make the basic acts of working with the UI faster, smoother, and more predictable, the simple time-saving benefits are enormous.
To really do this right is a big project. You need a really good graphic designer to prototype new visuals for the interface, icons, etc (I suggest looking into Figma, a nice UI prototyping tool). You have to re-architect all the UI to inherit from a robust, future-proof UI API to ensure consistent interactions across the board and low resistance to adding new UI elements that match the scheme. And you need to focus test the UI experience with new users and seasoned pros in order to identify what should take priority in the design. That sort of testing is critical to forming the right overarching direction for the UI so that it can be re-organized in the best possible way.
The goal should be to make the interface so intuitive that it practically melts away. Interacting with it should be so simple and consistent that the user doesn't even think about it-- they can instead focus completely on the creative task at hand.
Thanks for considering taking on this large and important task. The powerful methodologies that underly World Machine deserve a modern user experience, and that starts with a rich, clear, and robust UI.
Last post by nih - September 13, 2019, 04:15:52 am
With the arrival of Virtual Texturing in Unreal Engine we can have some truly colossal textures on our landscapes.
What's missing from this is a way to get the bitmap output from World Machine into UDIM format so it can be imported into UE as a single giant texture. This means it can be sampled by all of the different landscape tiles and get the correct detail textures from the VT, all the while not losing resolution to the 8K texture limit by having that imposed on the entire world.
Any chance that this could be added? I don't know the specifics of UDIM and I haven't played with it yet, but it would fix a massive gaping hole in the WM to UE tiled world pipeline.
Second feature request: please make it so we can downsample the heightmap on tile export. Then we can generate both the high detail bitmaps and the lower resolution heightmaps in a single job. For instance I'd like to be exporting my bitmaps at 8192px per tile, but the heightmap at 2017px per tile. As I'm exporting in .r16 format resizing them isn't a convenient option.
Cheers! Feel free to contact me if I need to explain any of this better.
Last post by geekatplay - September 11, 2019, 12:34:47 pm
I posted new tutorials on Youtube channel (newly released weekly) and to own it https://gumroad.com/l/gzXBD My step in creating Highfield map, distribution maps, wet maps ...etc to use in Unity and Unreal engines. I am going to add more to this series, mostly how to use those maps inside Unity and Unreal. Feel free to post your suggestion.
Last post by Demostenes - September 02, 2019, 06:29:24 am
As far as I know, there no simple way, how to make realistic looking dunes. Currently I am generating dunes via simple filter in other tool and importing as height map into WM workflow, but I would really like to have terrain 100% made in WM, current approach has some limits. Would be possible to add some filter, or macro? Thanks.
Last post by Polynaught - August 31, 2019, 02:39:28 am
That is soooo depressing to read. Seems like Discord kind of dissipates all the special knowledge into small locked groups where it never leaves again - and once these groups "die", the knowledge also gets lost because it cannot be "googled". Some backyard Discord server might have these conversations stored, but nobody ever will be able to find it. Is this the way the internet will evolve? From open (open-minded, open source) to closed and lost? Hopefully not.