I just loaded up and played around with the Landscape Mountains project you guys released with 4.4.
I want to preface this that I love all the examples you guys have been releasing, some of this feedback might come off as harsh on this particular project.
I have no idea why you guys made a hang glider demo when it’s pretty clear the majority of forum users are making games where you run on the ground. From my standpoint, construction of terrain an assets for a flight simulator is drastically different than from a game that takes place on foot. It’s evidenced in this demo too, the terrain looks great from a distance but might as well be a skybox. The on the ground detail is very low, even the normal maps look odd close up. This demo kind of suffers from the same thing the vehicle one did- by restricting how you experience the project you are limiting the number of methods you are showing off. Maybe that’s on purpose?
Doesn’t show off foliage well. No grass and a little over 2000 trees in the scene? Again, I think a lot of people are trying to create areas with heavy amounts of foliage, grass, trees and run at a respectable rate. This doesn’t show off any of that. I was expecting you guys to have a bang up example of grass, a cool looking forest- maybe some water elements. Something like this:
://youtu.be/1uQtCOWNn5M
Material complexity. Almost the entire scene is red when viewing material complexity. The material setup itself is so bizzarre and it doesn’t seem to be using the Landscape Layer Blend node or texture height maps. Which I found odd seeing as those features were made specifically to create landscapes.
I know there is some duality here, on one hand I’m asking for new tricks and on the other I don’t know why you aren’t demoing existing methods. I was surprised this was the area you decided to create complexity with instead of showing us a fast running 10,000 tree forest with fields of grass. Foliage wasn’t push at all in my opinion. And by choosing a mostly barren landscape, you avoided that issue all together.
Anyone else feel this way? Or were my expectations off?
I’m glad it was a glider instead of an fps project or just a matinee sequence really. There are tons of materials out there for walking characters. Also they said they are preparing a larger landscape project(probably an example for World Composition.)
No, I get what you are saying, but its a base. Why not take 10x the number of tree instances and populate the base scene and throw up a youtube video?
I am not trying to be a pedantic ***, but they do provide nothing but great material and examples, rather quickly.
I don’t think we should be a sad panda when its not the “perfect for my use case” demo, or might require some work on our part to really make a good tech demo of the engine per se.
Probably because it’s pretty easy to right click one of the other example Character (third/first person) blueprints, click Migrate, select the Landscape Demo’s Content folder, open the project up, click World Settings and select the non-glider Character blueprint.
Personally I think it’s that they gave us another, completely unique, Character setup to use, rather than just including one of the ones that we can move into that project ourselves with ease.
Yes it looks very “retro”, I found using the UDK overlay approach with a color map helps then you lay grayscale textures and normal detail under it. Simple to do and cuts material complexity, another thing to do is add plenty of decal to cover the base and add LMIV’s to try a focal points to try and draw the eye. Not on the same terms of performance or pretty when comparing to CryEngine though…
Even in Unity, whilst the base shaders need work you can easily fill out a terrain full of foliage and detail meshes with little to no impact. For grass it’s a 2D tex applied to a plane then it’s batched, even the terrain can be quickly offset by modifying the pixel error amount and detail resolution / per patch. A terrain brimmed with foliage and detail you can easily get away with 500 - 600 draw calls / 13 ms latency 60FPS + on some pretty rubbish hardware.
There’s plenty of options to reduce overhead and impact, culling and LOD system works pretty well. From others complaining of speed issues with Foliage + Speedtree, I’m beginning to think it’s nothing to do with how UE4 applies it. I’ll update the speedtree thread with the reasons…
The only issue I have and this is a prime example of what I see on terrains is the lighting, I’m actually thinking about ripping out there current LPV solution and replacing it with something completely different like I did with Unity although it was more of a hack job with Unity, nice to have the source code :)…
I was talking about UE4 with Culling and LOD :), I understand how Unity does it and we both know it’s limitations as it was one of the reasons for both you and me went to Eval UE4. Also I agree the CE way of doing things is pretty good…
This is my own experience as well. Been trying to put out a lot of grass meshes and I get quite bad performance after a while with nothing special in the scene other than grass which are planes, and trees which are rather low poly with LODs.
This really needs to be addressed, as well as the landscape blending in general. So far I’ve seen almost no outdoor landscapes that look halfway decent and I’m pretty sure it’s due to how complicated landscape layering is.
Previously I had downloaded the (Landscape Mountains Project) which was compatible with UE 4.4.3…
Now; I switch to run it with UE 4.8.3 … I discovered the birds blueprint is not running properly as it was running before… and I have no idea how to fix…
Need an advice… Edit
Same issue with UE 4.5.0