I'm looking at learning and buying a landscape generation package and am interested in some current opinions. The other option I'm considering is Vue.
I use Max, itoo FPP & GrowFX currently.
I guess the million dollar question is - is it feasible to create animations of trees and landscapes in T3 as a freelancer with a very small render farm? I don't really see it as an option to outsource the rendering at the moment, plus I need to learn first! :D I've seen some great work but without knowing render times it's hard to evaluate.
How have you guys found learning the software to the point of getting realistic results? and the ability to create renders in a reasonable amount of time? One big plus for Vue is that there's tons of training material. I really hate being at a wall without any means of improving. I find forums in general to be hit and miss with information especially when the solutions become a headache. And fair enough - people are busy.
For treescapes - are people pretty much exclusively using Xfrog? And would it be a no-brainer to buy that with T3?
I've heard many reports of Vue's poor stability and customer care - how do people find Terragen's stability and support? I've been reading a few posts over the last few days, and first glance looks pretty good.
Cheers
Hi DPS, welcome to the forums!
TG stability and customer care should probably made as example to the industry.
Together with this forum I think in the past 6 years it almost never happened that somebody was not helped by either forum users here as well as representative of the company behind TG, Planetside.
So that's something you should definitely not worry about.
Learning TG has mixed feelings for many. Some pick it up relatively quickly, some not.
If you're familiar with node-based workflow, like nuke, or how to set up shaders in Max with nodes (Which you probably know) then you will at least already understand that concept and the flexibility it offers.
I think, after observing this community for about 6 years, this is one of the key issues with getting started in TG2 and TG3 (they use the same UI and approach/workflow) for many people and perhaps mostly the ones without any prior experience with node-based workflow.
I was one of those and taught it myself in my spare-time next to study/work in about a year.
At least I knew how to operate it and how certain things worked. From there it expanded further and so did my capabilities to make more photorealistic stuff.
In regard to your animations. What are you looking for? Photorealistic, noise-free fly-overs of forests with detailed GI and at full-HD?
Perhaps with a bit of water or clouds here and there?
Then TG, at this moment, will not be your best choice. Nor would Vue by the way! It's as simple as that.
Such frames take multiple hours each to render at that quality. Vue is a bit quicker perhaps, but has a strong cutoff point where its performance dramatically drops. Often because of high-poly instanced objects or when scene-scale gets too large.
That's where TG shines though, it swallows many many high-poly instances and renders these relatively effortlessly, though relatively slowly, on full planet scale. The ideal situation would be to have the best of both ;)
Unless you have a bigger renderfarm than we think now. By small I suppose you mean anything <10 machines?
About XFrog. The newest collections look pretty good, but the older ones look a bit outdated, although the majority is still useful for mid- to background distant work.
Silva3D also sells great models for low prices. Definitely should have a look there!
Cheers,
Martin
Quote from: DPS on July 24, 2013, 04:13:26 AM
I've heard many reports of Vue's poor stability and customer care - how do people find Terragen's stability and support? I've been reading a few posts over the last few days, and first glance looks pretty good.
Cheers
I'll leave others to comment on the other aspects of your post but thought i'd give you some feedback with this part. Firstly Vue's customer service is notoriously bad, so if you need some support it may be a bit of a gamble.
For me running Terragen 2 on Windows has been very, very stable, pretty impressive considering the kind of application it is. Support is also excellent from the Planetside guys. If you look through the forum you'll see a lot of posts from Planetside, especially Oshyan.
If I were to be critical i'd say that the documentation has been weak in the past and slow to develop. Hopefully this will improve with TG3, just looking over the new TG3 doc pages that Jo posted today it looks promising. Also there doesn't appear any where near as many support\tutorial sites compared to something like vue's, which is a shame. It can take some time to understand the TG concepts and good, detailed tutorials can really help you get up to speed faster.
Thanks guys for your replies - I guess I should probably download a demo and have a crack at it first. Glad to hear the positive responses.
Quote from: Tangled-Universe on July 24, 2013, 04:57:27 AM
In regard to your animations. What are you looking for? Photorealistic, noise-free fly-overs of forests with detailed GI and at full-HD?
Perhaps with a bit of water or clouds here and there?
Do you really need to ask? Avatar2 of course... all done by me! :D
The animations on your webpage look good. The promotional vid here is also very inspiring: http://terragen3.com/terragen-3-in-motion/
(though also a promo for Ranch Computing)
I saw Richard Fraiser's webpage too and some truly awesome work.
So I'd be lying if I said I wasn't aspiring to achieve these types of results in time! And I mean learning time rather than ridiculous rendering times. I'd be open to creating scenes in tarragen and then projecting renders back onto low res geometry. Can anyone comment on how easy that is to do (the Terragen bits, passes etc)? I'm not a compositor but I know that I really need to do more. And I currently have a workstation and two slaves in my setup.
Regarding control - how easy is it to accurately recreate an existing landscape - say a few km radius? this would be important. So I might receive a DEM and a few photos. The photos would show areas that are vegetated and cleared - so I'd need to be able to recreate that to a reasonable degree.
I suppose you then just better get started with the software and think later about the other things you mentioned. First things first :)
Mind commenting on the last point that I made regarding control? I just edited my post. :)
The questions are very understandable and the answers to those will always be the same:
Yes you can do that, if you know your way around.
If you can get a DEM of that place then you can even try reproducing a reference photo.
You could paint masks over the reference photo and project these through camera to mask out trees/bushes/surfaces etc.
It's all possible, in different ways.
The best advice I can give is to get started and to slowly dip your feet into the software and especially to not try to do too much at a time.
Cheers,
Martin
That makes sense. And also I'm not expecting to get fantastic results soon, neither am I super hot even in Max but I know with time and effort that great results are achievable.
Thanks for your reply - just finishing off a project then I can start looking at this more.
Just jumping in here ! Thanks for the praise! Woo me etc!
As regards how long a frame takes to render I would usually aim for 20 min a frame at 1280x720 (720p) but I rarely get that. More normally I end up going for something more like an hour, sometimes more. After a while I loose patience and don't end up rendering those scenes as animations. :(
There are plenty of tricks that can reduce render time considerably, a lot of noise can be removed in post (I say a lot, more like "some")
Download the demo, and start playing about with it, have fun and ask questions.
Richard
Hehe will do, thanks. :)
I have another question - and I'm sorry there are so many! I'm creating some rocks to scatter inside of Max; Out of interest would there be faster/more efficient ways to create these in terragen?
In max I've used noises and cellular maps to displace a high poly spherified cube. I then had to collapse the stack so I could use a vertex painting script to paint a black and white mask, with my sharp edges being black.
The vertex channel is used as a mix mask for the Blue/White diffuse colour. So the bashed edges are white.
Then for the dirt, I have used a noise map as my mask inside a blend material to blend between my mud and rock shaders. The vertex map again is used to mask the dirt from the edges. Then a AO shader is used to add dirt on top where they meet other geometry.
This has some disadvantages in that it needs to be high poly to create a detailed rock and to use the vertex paint script So I'd most likely need to bake the vertex map and a normal map if I want to reduce render times.
Quote from: DPS on July 24, 2013, 08:02:42 PM
I have another question - and I'm sorry there are so many! I'm creating some rocks to scatter inside of Max; Out of interest would there be faster/more efficient ways to create these in terragen?
http://www.planetside.co.uk/forums/index.php/board,14.0.html - Here's a good place to find some good clip files (.tgc extension) to try as samples. Some people have made some pretty good rocks from pebbles to massive boulders.
It's a shame NWDA is still down, they had some nice rock packs.
In the meantime you might look at the new mesh displacement, fracturing, and other options in the Rock object of TG3. Using a voronoi displacement on a rock with a low number of faces could give you a similar look as what you've got here. Some of the shading techniques may also be similarly applicable, though importing the maps made in another app might be easiest (for the edges). You could also just import, say, 20 of these rocks and populate them in TG with some random rotation and scaling and it would probably be fairly convincing. TG will handle large poly counts well.
- Oshyan
Quote from: Oshyan on July 25, 2013, 09:12:17 PM
You could also just import, say, 20 of these rocks and populate them in TG with some random rotation and scaling and it would probably be fairly convincing. TG will handle large poly counts well.
- Oshyan
Oshyan does the populator in TG3 allow you to import multiple objects into one populator? If you had 20 rock objects that you wanted to use would you still need to use 20 separate populations?
Nope, not yet Reck. Something like that is planned for the future though.
- Oshyan
Hi
I appreciate the replies.
Rcallicotte - I will try it soon enough but not for my current project.
Oshyan - thanks for the suggestion - I was more trying to get a handle on what procedural control you'd have over something like this. Is there any way in Terragen that you could just create some kind of mask from by face angle? Say if rocks might be more polished where people have trodden on them. Or someone's kicked the moss off the top of a rock.
We don't have a way of keying off of face angle or anything yet, unfortunately. You can do shading based on normals, but it's not relative to neighboring surfaces or anything, so it's not really going to get you "cracks" or "ridges" or anything like a "dirt map". I'm not sure the best way to handle this honestly, but I think if you used the same noise shader to displace/make the rock shape as you did for surface shading/making surface shading, you could potentially get something similar.
- Oshyan
Thanks for the reply, I have a few more question regarding Terragen.
What sort of procedural goodness would you have at your finger tips if using terragen to texture/embellish a height field that has already been generated externally from Terragen? Specifically with regards to creating texture masks for valleys and where water would flow - but without running any additional filters that will alter the heightfield?
I've been experimenting with the world machine demo - I'm quite impressed at the moment (apart from the low res world on the demo which isn't great for a proper evaluation). I've imported both a heightfield generated from lidar data and also one from a mudbox displacement map. I'm liking the software - I think the erosion filter is great, and the varying mask outputs. I'm currently trying to work out whether you can gain any of these flowmaps without running the filter and altering the heightfield.
Why do people tend to use both WM and Terragen in conjunction? Is world machine better for some aspects?
I will try Terragen soon - but I've also just bought a house so need to be sure that pick the correct tool(s), hence the caution. :)
Cheers
Dave
World Machine is undoubtedly better at heightfield modeling. So people use it for that purpose. Terragen is better at "infinite" procedural detail, clouds, atmosphere, and rendering in general ("better" in that World Machine simply doesn't do any of those things). They are essentially complementary products, if your goal is final rendered output. But of course both are used separately. World Machine would be the better choice if your goal with the output is a heightfield that you can use in an application you already have, or for some other purpose than rendering (e.g. game environment). Terragen would be the better choice if you are aiming, as I think you said at the beginning of this thread, to render out fully realized photoreal, animated scenes. All World Machine is going to get you is the terrain (it's great terrain, but it's just one piece of the puzzle). If you could afford it, having both together would be best. ;)
- Oshyan
Hi
Thanks for the reply. And yeah I'm leaning towards that idea - just need to purchase in phases :) The Terragen atmospherics look sick and I definitely will be trying the demo soon.
One thing though - I mentioned in the previous post - is there a way to generate a flow map, or deposition map in terragen (or WM!) for an existing heightfield, or texture procedurally based on similar criteria?
Attached is a test render in 3DS Max - so in this test I've used a deposition map from the WM erosion node as a map to distribute the boxes and a flow map for a mask between land and water. But ideally I'm looking for a solution where I could do this without applying any additional erosion filter or the like. For artistic purposes it would be fine but if I'm working on a construction project, then I may be using a site survey and I probably would want to avoid creating my own erosion so as to not confuse anyone.
Quote from: Tangled-Universe on July 24, 2013, 04:57:27 AM
Hi DPS, welcome to the forums!
TG stability and customer care should probably made as example to the industry.
Together with this forum I think in the past 6 years it almost never happened that somebody was not helped by either forum users here as well as representative of the company behind TG, Planetside.
So that's something you should definitely not worry about.
Learning TG has mixed feelings for many. Some pick it up relatively quickly, some not.
If you're familiar with node-based workflow, like nuke, or how to set up shaders in Max with nodes (Which you probably know) then you will at least already understand that concept and the flexibility it offers.
I think, after observing this community for about 6 years, this is one of the key issues with getting started in TG2 and TG3 (they use the same UI and approach/workflow) for many people and perhaps mostly the ones without any prior experience with node-based workflow.
I was one of those and taught it myself in my spare-time next to study/work in about a year.
At least I knew how to operate it and how certain things worked. From there it expanded further and so did my capabilities to make more photorealistic stuff.
In regard to your animations. What are you looking for? Photorealistic, noise-free fly-overs of forests with detailed GI and at full-HD?
Perhaps with a bit of water or clouds here and there?
Then TG, at this moment, will not be your best choice. Nor would Vue by the way! It's as simple as that.
Such frames take multiple hours each to render at that quality. Vue is a bit quicker perhaps, but has a strong cutoff point where its performance dramatically drops. Often because of high-poly instanced objects or when scene-scale gets too large.
That's where TG shines though, it swallows many many high-poly instances and renders these relatively effortlessly, though relatively slowly, on full planet scale. The ideal situation would be to have the best of both ;)
Unless you have a bigger renderfarm than we think now. By small I suppose you mean anything <10 machines?
About XFrog. The newest collections look pretty good, but the older ones look a bit outdated, although the majority is still useful for mid- to background distant work.
Silva3D also sells great models for low prices. Definitely should have a look there!
Cheers,
Martin
Support is 1a++ and i agree to all other terms! :)
Hmm, very late reply here. One thing I've played with in the past - with some success - is running the erosion sim in say WM and then outputting the masks, but applying them to the original terrain. The idea being that this is how erosion would have occurred, thus implying it's how water might flow anyway. Maybe not strictly realistic, but still useful perhaps. Also I think there's a way using blur and subtraction to get a sort of "crevice map", contrast detection of a kind I think (or perhaps there's a more direct way in WM now) that could be similarly useful...
The options for such things in TG are currently limited as those are mostly raster-oriented operations, and TG's heightfield tools are basic at this point. But tools like WM already do a great job and TG can make use of all their various outputs.
- Oshyan