LuxeedRay opensource renderer

General project and community related discussions and offtopic threads.
User avatar
Dade
Developer
Developer
Posts: 5672
Joined: Mon Dec 04, 2017 8:36 pm
Location: Italy

Re: LuxeedRay opensource renderer

Post by Dade »

Sharlybg wrote: Wed Jun 09, 2021 8:34 pm Cycles x will have caustic of all sort without cache meaning you can animate it also.
Cycles x will have path guiding wich is superior to gi cache
They are vague promises, like OpenCL support in the past (and the future), they are not going to happen. Try to compare what you have today.

Comparing what you have today, on one side, and what they promise to give you tomorrow, on other side, it is quite meaningless.
Support LuxCoreRender project with salts and bounties
bestman8
Posts: 16
Joined: Sat Jul 25, 2020 6:45 pm

Re: LuxeedRay opensource renderer

Post by bestman8 »

Dade wrote: Wed Jun 09, 2021 11:11 pm
Sharlybg wrote: Wed Jun 09, 2021 8:34 pm Cycles x will have caustic of all sort without cache meaning you can animate it also.
Cycles x will have path guiding wich is superior to gi cache
They are vague promises, like OpenCL support in the past (and the future), they are not going to happen. Try to compare what you have today.

Comparing what you have today, on one side, and what they promise to give you tomorrow, on other side, it is quite meaningless.
isn't pathtracing able to do all sorts of caustics without a cache they are often very slow but it can do it (i think)
and as far as i know normal cycles can already do that.

also i will agree looking into path guiding is something different than going to use it
User avatar
Dez!
Posts: 368
Joined: Sun Apr 08, 2018 1:09 am
Location: Ekaterinburg
Contact:

Re: LuxeedRay opensource renderer

Post by Dez! »

There is the little detail they don't support GPU rendering ... at all: we win because they can not even show up for the match.
Let me be honest - I do not have the ability to use the GPU in two cases:
1. In setting up materials, lights, textures and rendering. The viewport on the CPU is slow. It takes even longer to run on the GPU. With the GPU activated on the viewport it becomes completely impossible to work. Every shift of an object stops work for too long.
Does the GPU help me with the Luxcore setup? No.

2. Final rendering.
I don't use the GPU for final rendering because I can't use the graphics card for that long. Luxcore takes a long time to render. I have to leave the switch on the CPU to have a working computer and continue to earn, simulate, etc.
This forum is full of GPU bugs, differences between CPU and GPU. I experience them myself all the time. And that's another very big reason why I have to render on the CPU.
Does the GPU help me in rendering Luxcore? No.

I do a lot of rendering. A lot of it. Nonstop, around the clock.
I have a separate desktop for that: https://i.imgur.com/m1ZUs5H.png
There the B-Renderon program and the Render+ addon keep my CPU constantly busy, without a break: https://i.imgur.com/Nn7VIfH.png
My stats on this are huge. I consider myself entitled to say that having Luxcore GPU rendering, for a freelancer producing an unlimited number of renderings, is not a working option.
Luxcore GPU is there, but you can use it in limited cases.
Does the GPU help make money? No.

It does. This is a situation that I would like to see changed. I'm talking about the problem openly. All the time. But it doesn't get solved. Year after year.



You are confusing what you need with what most people need: it is a common mistake.
I see myself as pretty much the only one doing product visualization at Luxcore. I look at Twitter, Blender Artist, the whole Russian-speaking Blender community - no one does product renderings in Luxcore.
Why?
Because there is no Cryptomate, no quick viewport, no bevel-node.

How do I do it?
I replace the bevel-node with high-end, expensive, non-competitive modeling. I lose out on that. And that's bigotry towards Luxcore. It's love. It's going against the tide of modern ways of fast modeling and making money with it. This issue became a hot, hot issue, even 3-4 years ago.
Replacing Cryptomate with an extra copy of a scene with a setup in Cycles to get Cryptomate on compositing. There is no product rendering without cryptomate technology. This issue has been open for years, too. It is not even discussed for so old anymore.
How do I get along without a quick viewport? You don't. That's how I work through the brakes. Losing an extraordinary amount of time on it.

How do I visualize products in Luxcore? Through fanatical love.
Why can I speak for most people? Because most people who try Luxcore product visualization come to me with questions about viewport, cryptomate and cad visualization without a bevel-node. And I have to say to everyone - love Luxcore so you don't have to endure and overcome the lack of these technologies. Or use Cycles.
Five totally awesome Cycles renderers just this month tried Lux. All five broke down when area-light became clumsy when setting up the scene.
I can speak for most because I see people try and it doesn't work, everything slows down, masks are hard to do.
I talk about it, all the time.
Most come to Luxcore consciously. They are cool professionals looking for an even higher level of quality renderings.
And I talk about it all the time - we're losing great renderers.

Dade, don't brush me off. I pay dearly to use your product. I love it. I love the picture I get at the end.
And I ask to be heard. Especially when the question is about the future direction of Lux.
In 5 years of using Lux I have asked:
Custom Normals - now everyone is using it. Every object has a normals edit modifier on it. On every object.
I asked the Constant Color node to bring out Values - many people use it.
I asked for the final rendering and viewport device delimiter to be brought back - many people use it.
I asked for a fix for uncalculated squares in Tile Rendering - no one uses Tile because there are still squares, but many are stuck with the lack of VRAM.

"You are confusing what you need with what most people need" is not true.

I'm not comparing the speed of AS and Luxcore.
I'm comparing their viewport. I am saying that if there is a movement to merge teams, then AS can give Lux viewport and Cryptomate technology. Those issues are more successfully solved there.
In general, Pro-render has the fastest viewport of all the plug-in rendering engines I've tried.

I'm not straying from the topic of this forum thread.
I'm saying that Lux could have taken from others to get better.
What Lux hasn't solved for years, but others have.
Linux Plasma | Ryzen 5, 32Gb, SSD M2, GT 590 RX | BenQ 27 | Wacom One | Microsoft Ergo | Tie Guan Yin tea
http://dezigner.tilda.ws/
User avatar
Sharlybg
Donor
Donor
Posts: 3101
Joined: Mon Dec 04, 2017 10:11 pm
Location: Ivory Coast

Re: LuxeedRay opensource renderer

Post by Sharlybg »

TAO wrote: Wed Jun 09, 2021 10:38 pm I do not think you should be worried about the progress of real-time renderers too much, the purposes of the offline renderers and real-time renderers are different. like all other renderers for example with Corona or Arnold you never going to see real-time rendering and you will need a ton of time for animation rendering and they sold out a huge amount of money. Also, the unreal engine projects are million dollars projects and really can not be compared to a free opensource project with a few developers.
bigger things to think about are publicity and popularity if users and industries know your product more you will receive more money and with more money, you can work more on your product. we should work on how to show it more and gain more users.
That's my opinion.
Yes Of course I know all that reason and i wrotte myself that realtime isn't going to replace high end quality renderer. But what I am trying to show you is the big picture. Realtime renderer is changing the market already :

1__ Seriously If a dynamic archviz animation to do for a client i have more reason to do it on Unreal than on Luxcore for sure
2__ If I am in competition to get a market against another artist using unreal exclusivelly How competitive i will be against him he is able to do the job for a fraction of the price.
3__ About quality there is lot of case where the lost is barelly perceptible. And when it is the level currently reach by realtime renderer is becoming enough for a lot of archviz scenario cases. Unreal isn't your common realtime renderer (Lumion / D5 / Enscape ). The Engine is already used in film production.
4__ Increase the resolution to 2k and 4k and it is a no brainer to choose between offline and realtime.
5__ increase the project size and Unreal win again (In the archviz Firm i'm working in we have to work on 100k m2 city project at least 5 time per years). Some require to be animated. It is fact and lot of artists will tell you that.

The market as changed and it impact every artists every CGI related rendering tools. Today you can remake the amazing the third and the seven quality animation in realtime : and how many client will cry for that ? the big majority of them : https://youtu.be/AQ-3aRhvFwU

The issue is that as a core developer it is hard to understand the strugle of people working with your tools day after days. you don't manage huge file or complexe 3D project on the viewport to see the strugle. But if lot of your experienced user are talking about something there should be a truth there.
Support LuxCoreRender project with salts and bounties

Portfolio : https://www.behance.net/DRAVIA
User avatar
Sharlybg
Donor
Donor
Posts: 3101
Joined: Mon Dec 04, 2017 10:11 pm
Location: Ivory Coast

Re: LuxeedRay opensource renderer

Post by Sharlybg »

That's a good idea to talk about. Of course, if everyone agrees and feel this is getting somewhere.
Years 2000+ => I was working on LuxRender and we were 15-20 developers scattered around the world. The major problem was to find a time for the weekly meetings. It was 8:00 am for the Brazil guy and late night for the one in New Zealand. They were the open source green days.
Todays most of the active contributor to the project are almost in the same area.

Dade == Italy
TAO == France
BYOB == Germany
Neo2068 == Germany
U3dreal == Germany
Sharlybg == Ivory Coast
Acasta == Nc

All theses name are separeted by 1 or 2 hours. And we are not forced to meet every month or every week. Only when it is needed. Like Elon say less time on powerpoint more time on making your product better.
Support LuxCoreRender project with salts and bounties

Portfolio : https://www.behance.net/DRAVIA
User avatar
Sharlybg
Donor
Donor
Posts: 3101
Joined: Mon Dec 04, 2017 10:11 pm
Location: Ivory Coast

Re: LuxeedRay opensource renderer

Post by Sharlybg »

Dade wrote: Wed Jun 09, 2021 11:11 pm
Sharlybg wrote: Wed Jun 09, 2021 8:34 pm Cycles x will have caustic of all sort without cache meaning you can animate it also.
Cycles x will have path guiding wich is superior to gi cache
They are vague promises, like OpenCL support in the past (and the future), they are not going to happen. Try to compare what you have today.

Comparing what you have today, on one side, and what they promise to give you tomorrow, on other side, it is quite meaningless.

Let's say you are right even if no one know the futur and nothing can stop a milion euros fondation to reach that goal :

1__ We currently sttrugle to still old cycles users base cyclesX isn't going to make it easier (we don't need too much only 10% will be fantastic)
Today blender user base is arround the 5 milions and in 2019 it was estimated at 1 to 3 milion people. Do we have at least 10000 active users ? even 1000 active users will already feel better : https://www.blender.org/press/blender-b ... mmunity%3F

2__ Cycles X alpha path tracer is already faster than Luxcore strongest feature GI caching by 30% in a highly favorable project (interior with lot of diffuse surfaces )
3__ Old Cycles is much responssive than luxcore in the viewport no matter it is gpu or CPu. With Cycles X it is even Worse as they add this strong viewport adaptive rendering.

We are the underdog why should we count on competition inactivity to progress ? We already lost meaningfull advantages to cycles X project because of that why we need to repeat this error. Matching Cycles X performance will not be enough for most users to switch they always ask this question what do we have to offer to justify such a move ?
Performance/feature is really key there i have 2 videos benchmark on the channel wich are among the most performing video i have for a reason , both doing arround 25000 vues.

AND now how worst it is going to be if one of the above mentionned feature get through ? :?:
Last edited by Sharlybg on Thu Jun 10, 2021 10:03 am, edited 2 times in total.
Support LuxCoreRender project with salts and bounties

Portfolio : https://www.behance.net/DRAVIA
User avatar
Dade
Developer
Developer
Posts: 5672
Joined: Mon Dec 04, 2017 8:36 pm
Location: Italy

Re: LuxeedRay opensource renderer

Post by Dade »

bestman8 wrote: Thu Jun 10, 2021 12:54 am isn't pathtracing able to do all sorts of caustics without a cache they are often very slow but it can do it (i think)
and as far as i know normal cycles can already do that.
Nope, try to render the caustic of a point light source with Cycles.
Support LuxCoreRender project with salts and bounties
User avatar
Dade
Developer
Developer
Posts: 5672
Joined: Mon Dec 04, 2017 8:36 pm
Location: Italy

Re: LuxeedRay opensource renderer

Post by Dade »

Dez! wrote: Thu Jun 10, 2021 1:22 am Luxcore GPU is there, but you can use it in limited cases.
The vast majority of the our users use GPU rendering (and have NVIDIA GPUs). The source of your problem is in having picked an AMD GPU.
Dez! wrote: Thu Jun 10, 2021 1:22 am I see myself as pretty much the only one doing product visualization at Luxcore.
This is just disillusion. Your claim is obviously wrong.
Support LuxCoreRender project with salts and bounties
User avatar
Dez!
Posts: 368
Joined: Sun Apr 08, 2018 1:09 am
Location: Ekaterinburg
Contact:

Re: LuxeedRay opensource renderer

Post by Dez! »

Most users render 1 picture a day. Or per week. To indulge.
Someone can make an interior. One in a week. With 2-3 visualizations.
I'm talking about large volumes. Serious work.
Apparently, I'm not knocking on the door, but on the wall.
That's too bad.
Linux Plasma | Ryzen 5, 32Gb, SSD M2, GT 590 RX | BenQ 27 | Wacom One | Microsoft Ergo | Tie Guan Yin tea
http://dezigner.tilda.ws/
User avatar
Sharlybg
Donor
Donor
Posts: 3101
Joined: Mon Dec 04, 2017 10:11 pm
Location: Ivory Coast

Re: LuxeedRay opensource renderer

Post by Sharlybg »

bigger things to think about are publicity and popularity if users and industries know your product more you will receive more money and with more money, you can work more on your product. we should work on how to show it more and gain more users.
That's my opinion.
The problem is what we are going to advertise as advantages over main competition. Don't think caustics and Light groups are enough for an artists to convert all his library and habits from Cycles to luxcore. With cyclesX speed boost it is more true.
With more money you can do more. but to increase investment you have to attract users first with your product. And currently the attraction rate isn't high enough to support futur and current competition. Why it isn't that high ?

1__Performances
2__ Features
3__ Comunication

In the past before Cycles X people do prefer to uses Ecycles instead of learn and re adapt to luxcore. If we want this to change we need a bigger performance/features difference.
Support LuxCoreRender project with salts and bounties

Portfolio : https://www.behance.net/DRAVIA
Post Reply