The realtime engines (Unreal, Unity, Crytek, and more) are doing amazing things, and have been for years actually. The thing to keep in mind of course is that the main reason most of it looks so good comes down largely to
good assets (i.e. 3D models, textures, etc.). This was especially true in years past, but even now as the engines get more capable of handling larger textures and higher geometry counts, having good assets matters a
lot. Unreal and others are adding cool realtime effects (that have been standard in rendering engines for years), like Depth of Field, and approximations of GI (even TG's GI is still more accurate outdoors though), but in general the number of capabilities and quality still pale in comparison to a full, "offline" render engine (including TG).
Unreal (and others) come with lots of great, high quality assets that these companies spend thousands of dollars (probably much more in some cases) developing. Take the Kite demo for example. The terrain was created in World Machine by a dedicated artist, and then they sent a
team of people
on location with a bevvy of expensive photo equipment to take 1000s of reference and
texture photos, as well as
photogrammetry of all the objects they planned to use. Then they made 3D models of all those objects
from these real world samples. Sure, then they populated the rocks, trees, etc. around, but consider the actual in-engine modeling process - it was probably some of the *least* of the actual work done here (animation of the main character surely took more time and effort, for example). Have a look at what it took to build it:
https://www.youtube.com/watch?v=clakekAHQx0Oh yeah, and the main character in the demo? They had a *dedicated team* just for him:
http://www.cgmeetup.net/home/making-of-unreal-engine-4-the-kite-demo/
The same is true of amazing demos like this one (not realtime):
https://vimeo.com/71148018Everyone *loved* it when it came out, but look at the Making Of video here. So much comes down to the quality of the assets. What would happen if you took those same assets and used them in TG? How long would it take you to "build" the scene (mainly placing objects) by comparison, and how realistic would it look vs. what Whiskeytree did there? Here's a quote on how that project was created:
Quote
Athens was an internal development project to improve our layout tools and see how much geometry we could push through our lighting pipeline (XSI / Arnold). The project had a 6 week schedule though we finished the project in 5 weeks. At the peak of asset creation we had 14 artists dedicated to the project during the first 3 weeks. Once assets and layouts where complete the crew scaled back to just a few people to finish the shot.
14 artists! Even after it was scaled back, there were still "just a few people" finishing it. The vast majority of scenes here are created by *one person*. That makes a huge difference, as you can hopefully imagine. So OK, you can start up Unreal and get an amazing looking scene in a few minutes. But that's primarily because there is a big library of assets there. Those didn't come for free either, of course, they have an asset creation budget, a
team of people creating things *just* for people to be able to use them with their products. Epic is one of the most successful, profitable companies in the game industry, so it's nice they're using some of that money to create shared assets for people to use.
Anyway, it all just makes me wonder what would happen if you took the same assets that are being used in these demos and use them TG. I think it's only then you'd have really valid comparison. I don't think it's about Terragen "not doing a good job" now, like I said it's more about the assets people have available, and the resources to
create those assets (e.g. the Unreal project has a whole team of asset generators who have budget to go on-location). The animations (camera moves) in the lighthouse demo are all simple, easily done in TG. Even the grass animation doesn't look any better (to my eye) than doing Mesh Deformation in TG, a technique that few have really experimented with but is readily available since TG 3's release.
Personally (well, and professionally) I would
love to see people do more with TG's capabilities, and you see folks like DocCharly65 coming in here and doing some frankly amazing things with animation that no one has tried before. So there's clearly a lot of untapped potential, and it is partly just driven by interest/desire, ingenuity, and persistence. But ultimately for scenes of the quality we're talking about, it really comes down to resources and time (and budget), unfortunately. And few in the TG community seem to have enough of any of them to not only push the envelope, but to come out the other end with new and innovative *finished* products. For example DandelO has stunned the community multiple times with totally unexpected and impressive animations of various things, but seldom if ever has he been able to produce a finished product with it. That's no knock on what he *has* done, I think it just comes down to what I was saying above: lack of time, resources, and budget. Hire DandelO, Hannes, or any number of other talented, creative people here (i.e. give them money), and pair them with a team of asset builders, animators, etc. and you could have some pretty amazing things coming out of TG. Think about what Ulco creates *by himself*. Planetside animated both the Garden of Eternity and Ponte Salario, but all the asset building was done by him or with e.g. Xfrog models. Now imagine what could be created with the kind of resources behind Unreal, but on a TG project...
All that being said, we're definitely interested in TG being a more capable asset generator for other systems, as well as improving tools for world building in TG itself of course.
- Oshyan
