Jump to content

blueapples

Members
  • Posts

    85
  • Joined

  • Last visited

Everything posted by blueapples

  1. Yeah all I am really asking for is a couple options on each vegetation item that allow us to specify the range. Seems like art direction should be up to whoever is making the game, if the tree size is inconsistent that's the developer's problem. The engine isn't supposed to just be a game maker with all predefined presets right? This would be less painful for me if the limit on the number of vegetation items was higher.
  2. Thanks for all the ideas everyone! Very helpful. Recalculating the normal maps definitely seem like something I need to do, but so far ShaderMap doesn't seem to be compatible with an iClone character (with separate materials for each part) that's been put through UU3D's PolyReduce command. PolyReduce seems to move everything to a single mesh which causes ShaderMap to produce a giant normal map for the whole model rather than one for each separate subset. Any other tools that might work better to do either the poly reduction or normal recalculation? Is there a feature in Leadworks I'm missing that might help with LOD management or do I just need to do it myself in script? As far as the level design aspects go, I've definitely been thinking about breaking things up more. I wish that the terrain counted for occlusion. Overall I'm hoping that as I fill in the map with more zones and buildings that things will sort of be a wash and I can at least stick to 30 FPS when a lot of stuff is occluded and not being rendered. Ideally I would get this up to 60 FPS with these characters looking good as a baseline.
  3. I'll play with colliders a bit more too. Ultimate Unwrap 3D was able to really crunch down these character models down to 8k and 10k each, but they end up looking pretty weird around the faces.
  4. So the models have about ~54,000 and ~39,9000 polys. The Merc from one of the workshop assets has ~13,000 so I don't think these are really *that* far off. Unfortunately I haven't been able to find a way to reduce polys on the models. Adobe's Decimator really messes up parts of them like their feet (they get 180 degree inverted upside down, very strange)... though it otherwise does a pretty good job, it's not usable. Not sure how else to approach that. This entire game is about NPCs so these are custom iClone Character Creator exports. Definitely want to have more than 2 of them and want to be able to render 12+ characters if at all possible. Adjusting billboard distances seems to have helped a little but it's still pretty tough to keep it above 30. Is there a way to FPS lock at 30 so it won't render higher FPS? I'd prefer it to be consistent.
  5. That's 4m polygons not even looking at the characters, just at terrain and vegetation. It's actually a bit higher when looking at them. What is a reasonable target for polygon count?
  6. I have an 1024x1024 outdoor scene with terrain that I can't seem to get above 35 or so FPS: When looking at the NPCs it drops significantly, down to 19-20 FPS or less: I've had a really hard time even getting this to perform this well and retain some of the look I'm going for. Originally I wanted a much denser forest (though I'm unsure if reducing density even helped, there are so many variables it's hard to keep track). So what can I do to speed this up? Josh you should have access to our SVN in a PM from me. This is the start.map file. Here is a shot with debug stats:
  7. Hmm, I guess it does, but the effect is very subtle for me -- so much so I didn't detect it in my map. I would expect the largest version to be at least 3-4x the size of the smallest (along a bell curve) since that would represent a realistic distribution for a real forest.
  8. Maximum outer radius for the paint tool is 256 while the maximum terrain is 2048. Even at a 1024x1024 grid, it is very time consuming to cover the entire terrain with 10+ vegetation layers.
  9. Allow filling entire map with the selected vegetation layer in one button click.
  10. Allow for selection of multiple terrain vegetation layers, then painting or erasing all selected layers at once in the editor.
  11. Is there any way to have random tree/vegetation sizes other than having multiple copies of each model at different sizes and painting over the same areas? Maybe something I can tweak in a shader to randomly scale vegetation models based on their x,y position on the terrain?
  12. Bethesda had a great talk about this at GDC (including stuff about how the editor handles it, particularly Local Snap Parents is neat feature), if you haven't seen it: http://www.gdcvault.com/play/1023202/-Fallout-4-s-Modular
  13. Ah, I see. Sorry about that. It would be indeed challenging to approach the level of content available through the third party sites. Maybe if Dexsoft wanted to do a deal where their entire membership content library was converted, then it'd be more realistic, but that seems like the amount of content that would be needed. That would be awesome.
  14. Well you were talking about controlling the art production which would be viable if we're only talking about Workshop items.
  15. blueapples

    Widget Progress

    Will there be a rich text multi line control of some kind? Thinking of a developer console or chat log with different colored text.
  16. Dexsoft Membership has been worth it so far
  17. A fix that only applied to Workshop items would be painfully incomplete for me.
  18. Cool I was thinking it had something to do with the size as I had noticed the scaling as well, glad the collapse thing is considered a good fix. Maybe it would be useful to ask Workshop authors do that as a last step by default and/or require tested colliders?
  19. I think it makes sense for your flowgraph system to be essentially like a functional scripting language GUI, but not as a complete analog for Lua. Procedural or object oriented programming doesn't lend itself to flow graphs in my opinion, it is very confusing indeed to try to do this. Functional programming does, however. That could be a useful limitation to impart a consistent design to it and make it usefully different from Lua.
  20. Here's an example from the map I'm working on. I might actually be able to play with the scale of the pieces to get them to play well with the grid snap (they look really close to the drawn grid at least), but this is really tricky when using multiple kits together because it dictates what the scale has to be for each kit based on it's original model size. This is a bigger problem though. In this case, the kit is really clean and vertices all seem to match well on each joined piece. From what I have seen I seriously doubt it's even possible to make better matching pieces, and if so, that type of art just isn't available publicly. In spite of that it's still seems to be impossible to manually position them perfectly. It was so frustrating it took me out of my flow last night and I'm pretty discouraged right now to be honest. Hence the idea. If it's not possible, that's fine, I'll have to try to clean up my assets more I guess. Using pre-made assets is a critical part of this project so I have to figure out how to make this kind of thing easier on myself or I'm not going to finish this game.
  21. Snap mode where geometry from a model being moved will snap to other nearby geometry exactly. Context: I am working with a modular set of cave interior geometry from Dexsoft's membership package. Pretty good, if a bit basic models. However it is very hard to get them perfectly lined up since they are not sized correctly to use the grid snap. This feature would help with this immensely.
  22. Did you actually get this fully working? I'd love to play with it. Haven't really tweaked my water much at all past throwing shadmar's newer water + caustics shader on it (which looks really cool, especially in the dark).
  23. Haven't fully fleshed out the concept here but I like the look.
  24. Did seem to help a bit though there is still a defined darker outline around the billboards.
×
×
  • Create New...