Page 5 of 5

Re: LuxeedRay opensource renderer

Posted: Thu Jun 10, 2021 6:52 pm
by TAO
maybe it's for the best to move this thread to the private section. I'm sure the developers will do their best to get the best results of the LuxCore anyway.

Re: LuxeedRay opensource renderer

Posted: Thu Jun 10, 2021 6:58 pm
by JulianoLisboa
TAO wrote:
Thu Jun 10, 2021 6:52 pm
maybe it's for the best to move this thread to the private section. I'm sure the developers will do their best to get the best results of the LuxCore anyway.
If it's not too impertinent of me, I'd like the topic to continue here so I can follow along.


Re: LuxeedRay opensource renderer

Posted: Sun Jun 13, 2021 3:25 pm
by daros
I'm sorry, but I would like to add a couple of opinions because only just now did I find the time to write a few lines. I completely agree with TAO, there is no point in being dazzled by the advances in realtime.
Rendering has always been played out on the terrain between speed and quality. Why abdicate quality all of a sudden?To a Luxcore that resigns itself to pursuing speed as the only goal I contrast an Unreal that pursues the quality of Luxcore. Furthermore, a game played on the terrain of quality is more suitable for luxcore than a game played on the terrain of speed.Speed is an objective factor and therefore appreciable by everyone and therefore suitable to cover large slices of the market; it is for this reason that the big fish, the multinationals, move here. Quality, on the other hand, is more subjective for the simple reason that it is not easily measurable and as such lends itself to more niche markets, exclusive or at least with different objectives.Many don't even see quality. The one that aims at quality is a market that moves smaller amounts of money but for this very reason is suitable to host microorganisms like Luxcore. Why fight against the giants if you can be comfortable among those who need you? I have the feeling that a lot of talk about the future of luxcore revolves around this node.
I have clients who would never use cycles or luxcore without bidir; these are clients who see the quality of good light distribution and appreciate this quality and are willing to wait a moment longer for this quality.
When you deliver renderings for a museum of modern art you want to use bidir, if you want to do decent lighting simulations you want to use bidir, if you want to do images for magazines you want to use bidir, if you want to work in the area of high-end furniture catalogue images you want to use bidir.There are so many markets that are not covered by GPUs.
The important thing is to perfectly align the pathcpu with the bidir; in this way you can use the pathcpu as a preview and the bidir as a final; this would be fantastic. But now it seems to me that they are quite incompatible...this is counterproductive because it distances two things that could be strongly complementary.

With regard to the economic problem, Dade, I understand your point very well. It is difficult to make open source a sustainable business model. I'm trying to imagine a system whereby users "invest" in a development, and as the development progresses and new users are added, those new users in turn fund the development. But as the users and investments increase, more and more of the investment goes to compensate the initial investors. The aim is to create a system of micro-funding that converges to cover development costs and even out the investment effort. Example: Today I put in 3,000 euros to cover part of your salary, but in two years time, when there will be 3,000 "investor" users, a typical investment will be 3 euros, of which 2 go to you and the other 1 goes to those who, like me, made the greatest effort in the first phase. It would be a business model made to create wealth in the form of "progress" rather than capital. A mathematical model has to be created, a system so to speak. This is something that is missing in the open source world and it would be cool. But for now it's embryonic and incomplete but I think it's doable. And it is especially with innovative business models that battles against giants are won.

Re: LuxeedRay opensource renderer

Posted: Mon Jun 14, 2021 5:47 am
by Dez!
A wonderful formulation of Luxcore's power.

Re: LuxeedRay opensource renderer

Posted: Mon Jun 14, 2021 8:16 am
by Sharlybg
About that quality vs performance and realtime i don't think you get my point. It isn't that we have to fear UE5 or Unity or Eevee. What i mean is that each render engine have a relative Performance/quality ratio or balance linked to a set time. And no matter you're on realtime or offline this Ratio is perceived according to the best average ratio reached by other software in the same field. Higher ratio tend to be easily adopted while lower one are depreciated until they die. Example :

Before Corona. Vray was seen by artist like the best option for Pro artist and because of Vray cycles was look as too slow for archviz for example.
After Corona Vray user already a bit unsatified by Vray setting complexity Start to complain and migrate to corona faster and easy to use. Same happen to Octane at certain Level because of Redshift. You can even extend the concept to blende vs Autodesk & Co.
But this ratio don't only concern 2 software only the industry is moving at a giving pace with a certain standar you have to fit or overtake to catch attention.
To go even far this Ratio impact the way your product is perceived among artist in the field it is basically your reputation. So my point is that if the standar move up you have to level to that and this standar isn't static it isn't set by a particular softaware it evolve with time and competition. I don't mean Luxcore should become a realtime software.Cycles X isn't realtime and Unreal quality isn't at Luxcore level yet But Both get higher ratio compared to the past. And in our case we are in direct competition with Cycles for sure.

People always compare Nvidia performance relative to AMD.because of AMD CPU performance INTEL small incremental upgrade can't be supported by COnsumers. And Now people start to wonder About apple M1 arch against X86.

Now back to another concept : Quality

I agree that not everyone can see it.And also we are all in this community because we like Luxcore output quality. It isn't about sacrifice Speed for quality I mean .But in our field Performance is part of Quality. Let me explain :

How for example you can a Birdirectional render to look better ? There are two ways :

1__ Increase bounce level from default 10+10 to 32+32 or 128+128 ( it cost you speed it cost you time it performance )

2__ Add feature like SDS cautics and Spectral capabilities ( Spectral can cost performance depending of your code approach/ and SDS cautics can speed up your render considering that sometime noise come from unresolved caustics )

In fact Bidirectional is already a well known precise and qualitative engine i'm a fan of it myself ;) . The thing Bidir miss the most is Speed. I mean both speed and performance are connected in our field. What happen when when performance is low ? most user tend to sacrifice quality in the settings for speed boost (Lower max bounce very low Clamping value ....)

About Funding I agree we need to put a strategy in place. But things are to be clear for people to decide. The community should know what is missing for the project and what system to follow to fix that.I think people will not be against fund to help Devs get more time on the project .What about a roadmap with taged funding goal ?