I've just downloaded the 2023.1 preview version and I have to congratulate the Twinmotion team for a new version of their application.
However, I was disappointed when I realised it still does not support the one thing I and many others have been waiting for for the past year: Lumen support. Marketing and Support know this as well, for it is listed almost at the top of the listed things to say about the new version in the Release Notes.
Now, I do sympathize with developers and their hard work in knitting core Unreal technologies into Twinmotion and make them all work flawlessly with all the assets, all the different imports, old version files, as well as with a fair number of new materials and capabilities, let alone a full upgrade to Unreal 5.0 from the previous version 4.27.
My question is, how much will we have to wait until we finally get access to core features of the Unreal 5.0 / 5.1 / 5.2 revolution, namely Lumen and Nanite? Are we talking about the next inter-version 2023.2? Are we talking about version 2024, which means a full year? Are we talking about 2025?
It's an incredible shame to watch such useful technology be solely used in games like Fortnite and not in here, where 120 fps optimizations aren't actually a big issue, and real time presentations to clients could almost fake feel path-traced at some 20 fps, or make movie clips, and so on.
Apart from the upgrading the Unreal version, some updated materials and some minor improvements, how can anyone call this a "massive update"?
I had already acknowledged these and referred to the Release Notes. I don't like being snarky but you replied precisely as someone who couldn't read to what he was replying to.
And I understand the difficulty of upgrading the Unreal version and with it a completely new file format. I get all of this. This is why I congratulated the team before expressing my disappointment, and the most important part of my question is precisely when is Lumen (and Nanite) going to be incorporated into Twinmotion.
I know exactly that if someone in Twinmotion's team were to read my comment, they'd be annoyed by it because they also want to be able to release such a thing. But here's the catch, they have a sense of what is possible and when it is possible. We do not.
Will it be the next trimester or the next year?
Is it still a pipe dream?
We in the office kind of need answers, because these two things, and especially Lumen, are probably the things that will finally upend all other alternative forms of ArcViz for us, and as such are by far the biggest promises of this application.
Was dissapointed that virtual shadow maps didn't make it over too :(
Hopefully the groundwork is now there with the port to 5.0 to be able to add these modern features like lumen, nanite and vsm, decals in pt. Just opening up old scenes at the moment and I'm hard pressed to find a difference visually!
Edit: Just did a video render and there is no difference visually between 4.0 and 5.0, except glow has changed and also the stock tree models are slightly different color, gi is a little different too.
I wouldn't understimate the effort it took to move over to 5.0 - so hopefully the visuals come soon.
I was super excited too when i heard that twinmotion was going to be updated to ue 5.0. not so excited when i saw it didnt have lumen or nanite though, super easy now to import the scene into ue, still its one extra step i know.
Just want to hope that maybe it might change once the release version comes. A clarification from the staff would be perfect, sure.
Twinmotion is incredibly fast to set up and iterate on architectural projects, and that's why we use it. Unreal is far too much of a complex tool for all of us to set up and generate a process that quickly goes from BIM to a rendered virtualization of the project with all the vegetation and lighting, camera position sets and movie clips. Twinmotion does this process in an incredibly fast paced way, intuitive and cleaned. The material substitution file is just wonderful, for example. Unreal is too heavy an app to set all of this up, nevermind the learning curve to do the basic of the basics, nevermind the idea to set up several movie clips for a single housing project.
It's not worth it, especially considering that Twinmotion has the same technology underlying it, the only thing missing is UI support and guaranteeing that the whole process "just works", which I appreciate is the most difficult part of it (devs, I do appreciate your work). It is a matter of time and that's the relevance of my question, because I have to ask myself, "do we have alternatives here"? Should we invest in other applications? Should we invest in learning Unreal? It would be ridiculous to spend entire weeks learning the bolts of Unreal just to watch Twinmotion incorporate Lumen and Nanite a few weeks later!
To , I haven't seen any changes at all in the GI, I even opened the same file in both versions to check that up, since the Release Notes indicate that the new Unreal version "greatly improves the Viewport display compared to previous versions". I'd love to know exactly how, because so far it completely eludes me. Virtual Shadow Maps necessitates Nanite doesn't it? Meanwhile what I always do when editing videos on Twinmotion is be constantly monitoring the shadow distance in each shot. If it's a very precise small shot, like a small space, it's possible to give it like a 30-50m total distance, and improve the shadow quality a lot. If my shots are much bigger I have to watch out that the far-out geometry is at least having a coherent shadow. Most of my outside shadows in housing projects have a 60 to 80m distance.
Well, it's no coincidence that this topic came up.
As I remember UE 5 integration is WIP, but there's no specific time when we'll see Lumen and Nanite working in TM.
I totally agree with LuisDias on this topic.
How good is this update anyway? We are happy with every update and congratulate the developers - the TM team.
Why are we writing such comments here?
Because we see how much work is a new update for the software developers and we want these energies to be directed towards the most useful things.
Priorities are the most important in developments. Have to work on those things that are important to the most users.
That's why we write on forums, that's why we write suggestions for the roadmap, that's why we ask.
I'm very happy with the "LED Wall" and "Precodural Cycloromas", although as an architect I haven't needed it in 25 years .
In contrast, "grass" was needed in all outdoor work. In all of them. That's why I'd be much happier if the "Vegetation scatter" tool finally kept the surface boundaries....
Therefore, users would greatly appreciate it if the important tools that are indispensable in daily work are included in the program as soon as possible.
At the top of this list is the render engine. The question is legitimate: we are all interested in when we can count on these updates?
Thank you for posting in the community and asking about our plans with UE5.1 and Lumen/Nanite. We are looking forward to being able to deliver you with new features but I cannot provide much details on timeline. However we are actively working on integrating UE 5.1 to be able to take advantage of newer features from UE.
Please provide feedback on the road-map with the specific features from UE if any that are the most important for you to see in Twinmotion. If you have any other ideas or suggestions please let us know as well. https://portal.productboard.com/7pu88c9kpmqtzt8hwg6arujh/tabs/4-under-consideration
We have gone crazy because of the time to render the video, so that we have no choice but to consider whether to give up tw without seeing tw's hope to realize lumen. This is a very painful decision.
Having been a user since the first version I truly appreciate the efforts and upgrades that have happened. I work with custom homes and am able to create 3-5 min videos in a matter of minutes with the new path tracer. Being a user of 3dsmax and Vray these time frames are crazy fast. I am curious what types of projects are being worked on that requires "an hour per second". I recently upgrade my machine for this usage and have found huge improvements in what I can do. Short of making the jump to Unreal I will take the real-time visuals and a little render time and let the TM team do what they need to do.
I rendered a video and was alt + tabbing between the two. Honestly the difference in GI seems to be so minor that it's not possible to tell if 4.0 or 5.0 is better but it seems to be there :)
The biggest real difference is the new dynamic sky and backscattering on plants looks better. The glass is different but I'm not sure if it's better or worse than 4.0 - just differenent.
We too are weighing up to stick with TM or not. After seeing the competition recently get raytracing (not talking about PT it is too slowww, flickering and not possible to use custom trees due to flickering) it is difficult to decide...because as you say the carrot that is lumen and nanite is very hard to ignore if it does come!
Thank you for the feedback, it is appreciated. I am also actively trying to learn new methods and new softwares, including Unreal Engine itself, to be able to do what TM is still uncapable of doing, namely, a good quality few seconds / minutes of animation.
@Luis Dias Thank you for the follow-up and wanted to share a specific feature improvement for Path Tracer on our road-map that we are working on: https://portal.productboard.com/7pu88c9kpmqtzt8hwg6arujh/c/958-path-tracer-improvements
A video denoiser?!? Oh my, you're really making my life difficult! (In a good sense). That's of course really great. I do wonder if it will still take an hour to render per second in the path tracer (or something in that order of magnitude), or if we can kind of get reasonable results in less time if we just use it in lower settings and denoise it!
Regardless, I appreciate that feature, and have high hopes for it. Thank you.
Thank you for the follow-up message and we are actively working on Lumen integration, you can reference it on our road-map and add feedback as applicable. https://portal.productboard.com/7pu88c9kpmqtzt8hwg6arujh/c/1028-lumen
I am curious as well to the kind of upgrades you have made, since invariably in my own projects, if we want to make a path tracer video we will have to schedule around 30 minutes per second of a video in one of the PCs (w/ RTX 2060) and maybe 15 minutes per second in another one (w/ RTX 3070), given the upgrades made in TM with the denoiser.
Maybe you bought a simple computer with two RTX 4090 and are able to make magic with it, or maybe your 3D projects are simple decorated living rooms with white backgrounds behind curtain windows. I have no clue. We build entire houses fully decorated, filled with bushes, grass and trees, and with panorama backgrounds. They are not la créme de la créme, but I like them. Here's an example of a work-in-progress (still too dark, furniture to be corrected, etc.):
Looks like we are doing some similar work. This one has about 70+ desert trees. It averages 15 FPS in TM. My prelim renders and animations render fast. For animations it is about 1-2 minutes per frame with depth of field and 1920 x 1080. I was so used to 45+ minutes per frame in 3DSMAX so I am good with 1-2 minutes. I can see things taking longer with higher resolution and settings.
I like the idea of stacking video cards. It is cheaper than a sever rack.
That's a fine render, I just want to point out that if you take 1-2 minutes per frame with a 1080p movie (which is a great stat, no doubt due to your card), you will spend 30 to 60 minutes per second of a video, for a 30 fps video (48 if you prefer 24 fps, cinema style). This is exactly the statistic we were talking about and you were astonished by.
In our office we like to provide a small 3 minute presentation to our clients. That's a whole 180 seconds, times 30 frames, a whopping 5400 frame count, which doesn't take less than 2 minutes each render with a lot of "hacking" on our part. 10k minutes is about 7 days of continuous rendering. I'm not even adding the ton of corrections we always make in the clips before final product is done. It's just not feasible. I can get away with one or two 10 second clips in a night for adding to the editing the next day and that's about it.
So given what you said, I'm just going to assume you just did not understand what we were talking about.
Thank you, yours is as well. The time frames started to make more sense after I responded. You are correct in the math when the frames are taking that long. I am so used to the longer time frames I get in 3dsmax that TM has been amazing to me. I do prelim models with minimal landscaping and no materials to get the approval (those render fast). Then more polished stills to get approval before doing the final animations. They will still take a bit to process but I can invoice for any revisions after the approval stage.
