Chaos Corona Forum
General Category => General CG Discussion => Topic started by: Digital Esthetics on 2024-10-02, 19:27:28
-
I can't help but getting a bit depressed when I see the extremly beautiful and photorealistic realtime videos I see in UE. while I'm sitting here with my RYZEN ThreadRipper waiting for my frames to finish (each 10-15 minutes) to get the same result.
I can't help but wonder when 3DsMax and Vray/Corona gonna be obsolete? Should I go over to UE already?
Some people say that Corona is Physically correct. but how does that matter when a Client want his project done faster. and want VR and all of the above.
Why cant we have Corona GPU based? why can't we have a render engine that is faster?
I would love an option that says. MUCH FASTER but slightly less realistic result.
Where is Corona when it comes to AI? why did it feel like Corona slowed down development after it got bought up by Chaos?
-
well, jump over and check it for yourself
IMHO, the thing is simply over-hyped, as almost every tech demo is
maybe visit Threat Interactive YT channel (https://www.youtube.com/@ThreatInteractive) for a bit deeper graphics reality-check
also, what you may gain in rendering time you loose during setting (which in case w/ offline rendering, is your profit ~ be aware that you're creating a different product).
-
Why cant we have Corona GPU based? why can't we have a render engine that is faster?
I would love an option that says. MUCH FASTER but slightly less realistic result.
Sounds like Chaos Vantage. For sure that's one more software (and one more sub) to add to your current workflow (while UE is free) but you don't have to learn a completely new way of doing things. Even if UE delivers much more accurate results than before, sometimes it's off and requires some severe tinkering with the settings to get something similar to offline rendering, where you never have to question the result.
You can also check Omniverse (https://docs.omniverse.nvidia.com/composer/latest/index.html), it's free and based on the same tech than Vantage. You'll see that even if the realtime RTX renderer is more accurate than a rasterizer, it's still behind an offline renderer. But having the two in the same software ? Really nice.
On a more general level, I don't know if there's much progress to be expected on the performance side of thing, at least not in the same way than what was done during the previous decade. Most research papers revolve around real-time rendering and GPU based raytracing, which is what Vantage is for.
Check V-Ray 7 Beta (https://www.chaos.com/vray/3ds-max/beta-version-7)for example, while some of the new features seem handy there's no big jump in performance, even though it's a major release.
Still, GPU support would be nice for a big interactivity boost without giving up on the CPU, a bit like Renderman's XPU.
-
Well, certain tasks aren't well-suited for GPUs due to their technical limitations, such as a lack of precision. Also, there's inverse relationship between quantity and quality, though not in strictly proportional way. Similar phenomena we're experiencing with Corona engine development, where improvements in speed and maneuverability (more features) have come at cost of reduced rendering quality and less efficient workflow.
-
Well, certain tasks aren't well-suited for GPUs due to their technical limitations, such as a lack of precision.
Could you elaborate ? I'm curious because if some GPU render engines take more shortcuts by default, others are well able to deliver quite precise results, like Octane or Iray ( both capable of spectral rendering btw ).
-
Technical limit is meant in regard to computation (lower floating point precision) making sim. (algorithms) either more prone to errors or simpler, introducing more bias, less quality - physical incorrectness. Simple fact. Even regulatory standard for industrial simulations demands double precision.
It's good for games, imagery and illustrations, but lacks for predictive rendering, visual prototyping, engineering, ...
-
Why cant we have Corona GPU based? why can't we have a render engine that is faster?
I would love an option that says. MUCH FASTER but slightly less realistic result.
Sounds like Chaos Vantage. For sure that's one more software (and one more sub) to add to your current workflow (while UE is free) but you don't have to learn a completely new way of doing things. Even if UE delivers much more accurate results than before, sometimes it's off and requires some severe tinkering with the settings to get something similar to offline rendering, where you never have to question the result.
You can also check Omniverse (https://docs.omniverse.nvidia.com/composer/latest/index.html), it's free and based on the same tech than Vantage. You'll see that even if the realtime RTX renderer is more accurate than a rasterizer, it's still behind an offline renderer. But having the two in the same software ? Really nice.
On a more general level, I don't know if there's much progress to be expected on the performance side of thing, at least not in the same way than what was done during the previous decade. Most research papers revolve around real-time rendering and GPU based raytracing, which is what Vantage is for.
Check V-Ray 7 Beta (https://www.chaos.com/vray/3ds-max/beta-version-7)for example, while some of the new features seem handy there's no big jump in performance, even though it's a major release.
Still, GPU support would be nice for a big interactivity boost without giving up on the CPU, a bit like Renderman's XPU.
Yeah, I've tried Vantage, and so far it doesn't seem to be perfectly in sync with Corona. And like you said. it's another subscription I have to have which is very annoying. racking up so many subs for Chaos I should be an investor at this point.
When it comes to precision. Sometimes its needed. and its good that it's there. but many times a customer wouldn't see a diffirence. I would Love an option where I can at least use the Power of my GPU to make some parts faster.
Dont get me wrong. Love corona. But it seem like a lot of softwares out there is outrunning us soon. and would hate that.
-
I still use Max/Corona mainly because of workflow features (Slicer mat, all the Include/Exclude tricks, Overrides, etc..), things that help Design&Architecture projects that have revisions being big part of the process.
And also because I need 100perc. precision, not even 99perc. would suffice. So the extra-level of detail/sampling/etc... that pure dedicated path-tracer engine outputs is still beneficial.
In faster, more purely result oriented work, I am not sure I wouldn't be using Unreal already. For something like Real-estate marketing, I think the switch would absolutely make sense. Faster visual development, super fast rendering, easy VR/Animation/etc.. outputs, etc..
Technical limit is meant in regard to computation (lower floating point precision) making sim. (algorithms) either more prone to errors or simpler, introducing more bias, less quality - physical incorrectness. Simple fact. Even regulatory standard for industrial simulations demands double precision.
It's good for games, imagery and illustrations, but lacks for predictive rendering, visual prototyping, engineering, ...
Can really agree with it. The more you are on the design side of the industry, the more is classic off-line renderer still valuable. The more is your work veering into pure marketing imagery, the more attractive Unreal and other real-time solutions look.
And then there are all the hybrids, like D5 renderer, etc.. I don't have time to test them, but they can be possible bridging solution.
-
Definitely still a need for offline engines, for that precision as noted. Also for the ease of workflow - Unreal is a game engine after all, and the whole toolset is designed for creating triple A games, which means it is not very considerate of what an archviz artist needs and you'd have to wade through a workflow that doesn't take you and your needs into account (Unreal long since backed out of their original plan of moving into the archviz market, and refocused on their core of making games).
As for why no Corona GPU, the answer is because all of Corona development would stop for 2 to 3 years for that to happen, and that would not meet the needs of our core userbase. Instead, there is the option to move to Vantage, which means you can build your scene with tools you are familiar with, and then get a very similar result all done by GPU, giving a "best of both worlds" without a freeze on all other Corona developments. Also keep in mind that some of the visual differences between Corona and Vantage would happen with the hypothetical Corona GPU too (another reason we never planned on doing a GPU version, as people would expect it to be "identical to Corona, just somehow magically faster thanks to the power of GPU!!" Which would never be the case).
On development slowing down, the merger happened in 2017. Corona released in 2015 (next February is our 10 year anniversary since commercial release!). Are you sure that development slowed down 7 years ago? :) Any impression of "slowing down" is simply due to product maturity - in any engine's early days, things seem to be going fast as core features get implemented (now there is displacement, etc etc) but of course as all these major things appear and a product matures, it feels like less is happening as all the main things are already in place. What is left are refinements, or problems that are trickier to solve and so take more dev time to get a working solution, and so on.
On AI, that is coming at Chaos, the AI Enhancer is already out for Enscape and is planned to come to Corona and V-Ray, so that your 3D people can finally look like people ;) Of course we are considering what our users actually need - the fact they need "specificity" in their images means you can't use AI to pull things out of thin air, but instead what is useful is refining things such as people which generally look "rendered" while the rest of a scene looks photoreal. So we're looking at where AI actually makes sense, and looking at how that can be nicely integrated with existing workflows so that it all fits together naturally. Keep watching this space :)
Thanks!
-
On development slowing down, the merger happened in 2017. Corona released in 2015 (next February is our 10 year anniversary since commercial release!). Are you sure that development slowed down 7 years ago? :) Any impression of "slowing down" is simply due to product maturity - in any engine's early days, things seem to be going fast as core features get implemented (now there is displacement, etc etc) but of course as all these major things appear and a product matures, it feels like less is happening as all the main things are already in place. What is left are refinements, or problems that are trickier to solve and so take more dev time to get a working solution, and so on.
Hello Tom,
It is interesting how Chaos called this in their presentation videos. For example, in the V-Ray for SketchUp webinar video, there is a Chaos growth timeline presented, where they clearly divide what was acquired and what merged:
youtu.be/J4eYOpjkEYM?list=PLm2Irfq-UvZ3MOnvOjnmhL0KiCOv0B0SE&t=88 (http://youtu.be/J4eYOpjkEYM?list=PLm2Irfq-UvZ3MOnvOjnmhL0KiCOv0B0SE&t=88)
2017: Chaos acquires the Czech software developer Render Legion, known for Corona Renderer.
2022: Chaos merges with Enscape and acquires Cylindo.
I remain optimistic, but sometimes I get the feeling that within the Chaos family, V-Ray and "native" Chaos tools are pushed more to the forefront, rather than being on the same level as Corona.
Even the updated Corona Scatter—Chaos Scatter in V-Ray 6—was presented without any mention of Render Legion, as if it was built from scratch (please correct me if I am wrong).
Corona had own website, which is now just a subpage of Chaos. Old website had a lot more freedom; now, there are unified rules- the header and footer are all about native Chaos products, and even the copyright text on the Corona page refers to V-Ray and Phoenix.
Could you please tell me where I can find these two pages on the new website?
https://corona-renderer.com/doc
https://corona-renderer.com/resources
Just mixed feelings today...
-
I can't comment on the exact legal details of the joining of the two companies, as I didn't see those (not being an owner of Render Legion :) ), but I can comment on the way it was done in regards to autonomy and in that regard it was very much a merger, which is how we've always referred to it (including internally).
Scatter has its own development team now, some from "originally the Chaos side" and some from "originally the Render Legion side", and I am not surprised there is no mention of it originating with the Corona / Render Legion team as its history really doesn't have an impact on the tool as regards the users - what everyone cares about (Corona and V-Ray users) is what the tool lets them do now, today, and not how it got started :) The same can be said for LightMix, or the PRG Sky; going into their history isn't of any benefit to users. I am happy to point out how we contribute in innovation though during presentations, as when talking about Corona it is good for users to know what the team is capable of imagining and creating :) So our history is not gone nor forgotten! But yep I wouldn't expect to read about that if you are looking from the V-Ray side as to what is new in Scatter.
As for the website, the fusion of the two into the Chaos site is about Corona taking its rightful place alongside the other products, and not about it becoming secondary or "lost in the mix". It's an ongoing process, we're aware internally that Corona is not as visible as it should be on the website, and discussions are always ongoing about the various ways the website can be updated to improve on that - all in reference to archviz of course, which is where Corona shines (we're not going to be tackling Hollywood VFX today any more than we were at the founding of Render Legion!) Like most things, it's an iterative process, as it takes a lot of work, and now has to cover all products.
These days, we're part of more things than we were before - if we take partners for example ("resellers", if you like); as Render Legion, we had virtually none, everything was done by direct sales. Now though we have Corona available through Chaos' extensive network of partners, who are getting Corona out to more people than ever before, and indeed I just came back from Partner Days in Sofia where I was presenting Corona to some of the largest partners, all to make sure that Corona is something they know about and understand, and that we equip them to be able to spread the word to more folks than ever before. I'd say we were "slightly more autonomous than we should have been" up until a year or two ago, where this opportunity went unexplored due to our history of never working with partners before. These are some of the unseen things (well, unseen unless you are a partner ;) ) that go on, part of the ongoing process of making sure Corona is fully a part of all things Chaos.
On those two pages - the docs page is exactly where you linked to, under the old corona-renderer.com URL (and no, it was never a page that was linked in the site map when that was our home page, it was always pretty much a "if you know where it is, you know where it is" page and that hasn't changed yet, til we find a good place to put it on the chaos.com site - maybe then it can even get a link to it from someplace to make it more discoverable :) ). The resources page is now divided in two, with the links to Max and C4D on the page it redirects to. The new pages are not identical as there was a lot of old, outdated stuff on the previous resources pages (links to material libraries that were from Corona 1.3, links to model collections that were 404, etc.) so we cleaned this up to provide links to things users would find particularly useful, such as tutorials and free scenes (we never had official free scenes before and relied on 7 year old free scenes hosted by users).
The "taking its place alongside the other products" will continue too, just as it does for Enscape for example, as we seek to make sure that anyone looking into Chaos products are fully aware of what options they have within our ecosystem and when they'd want to be making use of any particular product, as well as links between more and more products (Corona to Vantage being the prime example at the moment). Hope that little bit of background proves useful!
-
Hello Tom,
thank you very much for such a detailed response. This is very helpful information that gives a better understanding of how everything is happening.
Please understand me correctly, in my previous post I shared my thoughts because, in these turbulent times, it's important for users to understand how the company is doing and what the approximate plans are, to avoid ending up with knowledge that becomes obsolete, as happened recently with Adobe XD, for example.
On those two pages - the docs page is exactly where you linked to, under the old corona-renderer.com URL (and no, it was never a page that was linked in the site map when that was our home page, it was always pretty much a "if you know where it is, you know where it is" page and that hasn't changed yet, til we find a good place to put it on the chaos.com site - maybe then it can even get a link to it from someplace to make it more discoverable :) ).
As for the web Corona UI module, I think it would fit perfectly if there were a link under the "Built-in help" text in the features section. Please see my attachment.
Best regards
-
I totally understand the reasons why you were asking and sharing thoughts, and do please continue to do so :) We always welcome hearing how things look from the outside, it does help us share information that helps the community know what is going on and why, etc. I am sure if you were wondering about these things, other people were too - and I hope the answers help shed some light on it all! TY for the mock ups too :) Not sure when the next round of website page updates will happen, but good to have ideas and suggestions for when it does!
-
Continuing with the offtopic ;) - remember that you can use https://docs.chaos.com/category/corona in addition to https://corona-renderer.com/doc
-
We have been fully transitioning to UE5 for producing architectural animation since this year. Actually, I am not very fond of the production effects of UE5. UE5 does not have much optimization for architectural animation production, and many functions need to be explored by myself. For example, the PCG and brush of the tree planting plugin are difficult to use. We miss forests and chaos scatter very much. The production effect is also very strange, with many dark areas being dead black. Now occasionally using Corona feels particularly comfortable. However, the speed of UE5 is really fast. Based on the animation path feedback just sent by the customer, we estimate that we can send the sample to the customer for confirmation in one or two hours. This is difficult to achieve with an offline renderer (3Min animation)
-
Out of curiosity, have you tested Vantage as a way of getting GPU rendering speed, since conversion from a Corona scene is far easier than getting something over to and working in Unreal? (And then you have the best of both worlds, from one scene). Would be interested to hear thoughts from someone who is experienced with using UE.
-
Out of curiosity, have you tested Vantage as a way of getting GPU rendering speed, since conversion from a Corona scene is far easier than getting something over to and working in Unreal? (And then you have the best of both worlds, from one scene). Would be interested to hear thoughts from someone who is experienced with using UE.
We have previously used Vantage to confirm the camera path for animations with clients. However, the rendering speed of Vantage is not very fast. For a 1-minute camera path with a sampling count of 100, it takes 5-6 hours, using a 4090D graphics card. I feel that the speed of Vantage is similar to that of UE5's path tracing
-
Out of curiosity, have you tested Vantage as a way of getting GPU rendering speed, since conversion from a Corona scene is far easier than getting something over to and working in Unreal? (And then you have the best of both worlds, from one scene). Would be interested to hear thoughts from someone who is experienced with using UE.
We have previously used Vantage to confirm the camera path for animations with clients. However, the rendering speed of Vantage is not very fast. For a 1-minute camera path with a sampling count of 100, it takes 5-6 hours, using a 4090D graphics card. I feel that the speed of Vantage is similar to that of UE5's path tracing
Indeed, it is a GPU ray tracer with many cutoff options, and it does not have anything similar to Lumen in UE, which in UE works instantly.
But at the same time, it’s a trade-off -you don’t need to optimize materials, redo scatters, or deal with advanced, unsupported parts of the scene.
-
In terms of both quality and speed, D5 Render is clearly beginning to dominate the architectural visualization market. Chaos Vantage, while technically capable, lacks the content depth and aggressive marketing to match what D5 currently offers.
Engines like Unreal Engine may appear fast and flexible for Archviz workflows, but they often become impractical when dealing with large-scale, detailed, and serious projects. That’s why we believe D5 will continue to gain momentum in the coming months. As the largest visualization office in Turkey, we've already shifted around 45% of our pipeline to D5.
We are now entering a market where seamless technical control and ultra-optimized workflows are more important than ever. Corona Render will always remain an artistic tool that I personally can’t give up — it has soul. But it’s also true that the market is changing at an incredibly fast pace.
Unfortunately, Corona has begun to lag in technical innovation, much like what happened with Mental Ray and even V-Ray at certain points. There’s been little movement recently in terms of speeding up workflows (e.g., no integration with Gaussian Splatting, minimal real-time development, lack of AI-driven enhancements, etc.).
Today there are dozens of real-time enhancement tools — very few professionals even open Photoshop anymore. At this stage, I believe Corona may face a long-term challenge if it doesn’t evolve fast enough.
And no, I won’t go down the path of complaining “why Corona doesn’t have GPU support.” That would be like arguing whether Tesla should make a diesel model. It’s irrelevant. The discussion is about pipeline efficiency and market direction, not nostalgia.
-
In terms of both quality and speed, D5 Render is clearly beginning to dominate the architectural visualization market. Chaos Vantage, while technically capable, lacks the content depth and aggressive marketing to match what D5 currently offers.
Engines like Unreal Engine may appear fast and flexible for Archviz workflows, but they often become impractical when dealing with large-scale, detailed, and serious projects. That’s why we believe D5 will continue to gain momentum in the coming months. As the largest visualization office in Turkey, we've already shifted around 45% of our pipeline to D5.
We are now entering a market where seamless technical control and ultra-optimized workflows are more important than ever. Corona Render will always remain an artistic tool that I personally can’t give up — it has soul. But it’s also true that the market is changing at an incredibly fast pace.
Unfortunately, Corona has begun to lag in technical innovation, much like what happened with Mental Ray and even V-Ray at certain points. There’s been little movement recently in terms of speeding up workflows (e.g., no integration with Gaussian Splatting, minimal real-time development, lack of AI-driven enhancements, etc.).
Today there are dozens of real-time enhancement tools — very few professionals even open Photoshop anymore. At this stage, I believe Corona may face a long-term challenge if it doesn’t evolve fast enough.
And no, I won’t go down the path of complaining “why Corona doesn’t have GPU support.” That would be like arguing whether Tesla should make a diesel model. It’s irrelevant. The discussion is about pipeline efficiency and market direction, not nostalgia.
Very interesting. Thank you for the insights.
I'm using Vantage for the first time and I find it a little buggy EDIT: and the camera animation part is not quite ready yet IMO, nor the export of camera animations from Max.
I'm interested in D5 but I, like you, I don't want to give up Corona for still images (EDIT: not yet at least).
How is your workflow in that case?
Do you have a Corona scene and a separate scene for D5? Because AFAIK D5 is not compatible with Corona shaders, no?
-
Envision may be another tool to consider: https://www.chaos.com/envision
-
Envision may be another tool to consider: https://www.chaos.com/envision
The problem with Chaos products, it seems, is that they are spreading too thin to the point that they seem to be neglecting a percentage of each product they acquired. I don't like that and would rather work with a software that is unique to the company that owns it. Sorry, but I'm being honest here.
-
Envision is a 100% in-house Chaos product.
-
Envision is a 100% in-house Chaos product.
I didn't mean that. I meant they seem to be neglecting their products in general or just not having enough manpower or enough responsiveness to keep such a large amount of products.
At least that's my feeling with Corona and Vantage, and I read in this forum that the Anima forum isn't getting responses from developers.
https://forum.corona-renderer.com/index.php?topic=44916.msg233263#msg233263
I mean, it's not a secret that core Corona development got very slow. Its development seems to be focused on the compatibility with other Chaos products and their ecosystem or additional services they can offer rather than true improvement.
Not a good sign.
Edit: Also the difference in development between D5 and Vantage, although they both started practically at the same time, is huge.
Edit 2: Also at least here in NY, I find it impossible to get an Nvidia RTX 5090 or even 5080 for a decent price. They're mostly sold out since their launch. D5 seems to support AMD without issues. I may be wrong though.