Well, our company is in the process of migrating from Unity to Unreal (due to Datasmith) and since we don’t write any custom advanced graphics tweaks, let me tell you, the quality in Unreal out of the box was a revelation. You might be able to hack Unity to look amazing with a dedicated staff of experienced developers, but for us who just want to visualize our models, just adding a point or spotlight to a fresh project in Unreal/Unity and comparing the quality was a huge difference:
The above image is with the bias tweaked as far as it can realistically go in Unity, and with no tweaks in Unreal (because we don’t know how yet). And yes, we have purchased ambient occlusion assets and even tried a custom contact shadow asset in Unity, but it didn’t help us much (the contact shadow asset fell appart in VR with really complex geometry anyway).
Oh, and the CAD to FBX conversion methods we tried before mangled lightmap UVs, and Unity failed to generate good ones automatically as well, so we were forced to use 100% realtime lights (which we might do in Unreal too, because in some scenes we have a lot of changing geometry).
Now, there’s a lot of things when it comes to the editor itself in Unreal, however, which appears far, far behind Unity in terms of usability (despite blueprints), but we’ll see if that kills the engine for us or not this time around. We first chose Unity because of the much larger community and larger availablity of cheap assets, but we’re hitting the limit of how far we (personally) can push Unity in its current state.
Unity 2018 promises a lot with the custom rendering pipeline, and the forward renderer in Unreal is quite new and doesn’t support many features yet, so it appears to be a race that’s really “alive”…
Wait did they fix the performance in Unity & add easy replication or something? Did I miss this? All unity games suffer greatly on performance and feel very choppy. Just doesn’t feel as smooth as unreal or other engines, its odd. And multiplayer is way harder to do in unity. Are these 2 things improved in 2018 ?
No.
But most people using Unity doesn’t care anyway, so ppl won’t even talk about the awful performance anymore, they think that C# “job system” is going to fix everything… * it will not *
Hmm… this is perhaps anecdotal, but ever since the HTC Vive was released, the worst performing games (or architectural visualisations) I tested were always Unreal, and there were way more Unity based games (both in and out of VR) and none of them seemed to have any issues. I always thought that was down to the higher fidelity graphics output of Unreal, though.
Well VR is a different story, don’t have vive or rift myself. Unity seems to suffer more in multiplayer games with larger player counts, or large outdoor environments with alot of mesh’s/lights especially with alot of foliage. Since this is an arch vis thread I guess that doesn’t matter. I think for games, especially large ones, UE4 is king. Saying it is dying just because of architectural rendering alternatives doesn’t really make sense because thats not the main use for it
Actually Unreal 4 also isn’t good for “large world games” as well, because Carmack’s Mega Textures technology have never been implemented while nowadays for in-house AAA engines this is a given when the game is open world.
I’d say that on the rendering side, UE4 needs an overhaul of heightfield rendering. It seems to be the weakest link. Virtual texturing is good, but by all means not mandatory.
Speaking about Unity, I’ve gave a spin to their scriptable rendering pipeline. While I had high expectations and a ton of hype, it turned out to be just a way of patching up a giant downside of Unity, that is it being closed source. UE4 offers all that out of the box, just not as beautifully served.
But if they’d make source public then they would have to do what Epic did, removing a ton of licensed libraries, so for Unity users it’s fine having the source closed I guess. For most mobile projects source code isn’t a necessity and Unity is mostly used for mobile games anyway.
For example I’d rather stick to Scaleform instead of UMG, although UMG is easier to use, but Scaleform had a lot of potential and last versions were very very fast rendering with automatic drawcall batching UI elements; it’s a shame Scaleform is dead.
CGarchitecture has released their annual rendering survey They send it out to all the architectural visualization people that frequent CGarchitecture and related social channels. Just like last year, they had roughly 2K respondents.
The last survey was released at the end of November 2016.
Summary of 2018 results:
UE4 is the 4th most used production renderer (including all offline renderers)
*1 in 5 ArchVis people are using UE4 in production *
21% are using UE4, increase of 100% in one year
#1 is V-RAy (63%)
#2 is Corona (30%)
#3 is V-Ray RT (30%)
#7 is Unity (8%) - UE4 is 3 times more popular
UE4 is #1 renderer being experimented with (including all offline renderers)
41% are using UE4, increase of 50% in one year
#2 is Corona (27%)
#3 is Arnold (12%)
#8 is Unity (10%), - UE4 is 4 times more popular
UE4 is #1 by those using real-time renderers in production:
52% using UE4
#2 is Lumion (34%)
#3 is 3ds Max Interactive (21%)
#4 is Unity (19%)
UE4 is #1 for those researching real-time solutions:
77% using UE4
#2 is Unity (20%) - UE4 is 4 times more popular
#3 is 3ds Max Interactive (16%)
Those researching UE4 are 69% likely to use in production in the future
Those using real-time want to use it for everything (presentations, design, fast rendering, XR) equally
Overall, 73% think real-time is important to very important, this increases to 88% in five years
Students: Unity (10%), but Unreal is stronger (17%)
