E-Cycles - Faster rendering for CUDA GPUs and better AI denoising

(Lumpengnom) #283

Not for my workflow. It allways depends.


(smilebags) #284

I’m not debating the value of the work he is doing, I’m sure there are many people willing to pay a monthly fee for access to what he’s doing. My comments are regarding the possibility of alternate methods of working with the community.

I think there are parts to that problem that you aren’t fully aware of or are not considering. A pointless 20% drop in performance is a big deal, but I have many doubts that is the case. Remember, Blender 2.8 is not released. Shifts in performance are expected. He is clearly saying he will make an effort to fix it for release.

In the worst case, progressive refine is 20% slower because the developers deemed another feature more valuable than this. That is highly unlikely to be the case, and this is definitely not a case of developers being unreasonable with merging changes.

My point above was this: if there are clear changes which the developers can see improve performance without hindering image quality (which it seems like most of the optimisations are, given that he can get identical results with a 170% speedup) and can see why those changes help things, they will happily merge it. They won’t reject a straight up performance boost. They likely will reject changes which, as I said before, break temporal or spatial consistency with sampling, or change the resolved image.

1 Like

(rawalanche) #285

Of course it is. Chaosgroup employs over 200 employees, many of them top experts in their fields with salaries appropriate to their skill. You can’t do that if you don’t get paid for the product you make. Especially not if the main source of your funding is a community that for the most parts tend to be dicks about having to pay for stuff. :wink:


(smilebags) #286

I don’t believe working for a long period of time without income is a good idea. I certainly wouldn’t be able to do it, I’d be homeless in a matter of months. My understanding (which may differ from others) of contributing to open source software is that you so under the understanding that you’re not getting paid. Work a job, and contribute in your spare time. If you can find someone willing to pay you to work on it, then great! But it is unfortunate to lump genuinely beneficial changes to Cycles in with mostly unrelated (but still very valuable and well worth its price) learning material.

If you can find 100 people to pay a subscription to you in order to develop Blender for them, why not contribute those changes back into the community at large? The people still receive the exact same improvements regardless, plus, more people will find out and likely fund further development. I think that’s the essence of my suggestions.

I have thought long and hard about how to make a living from contributing to open source software. I have a few theories such as Patreon-style subscription based ‘employment’ where people subscribe to keep you developing, and an alternative feature-based crowdfunding style, where you put your price to develop a certain feature, then if you get enough backers, you do the work and get paid. Both of these allow the development to benefit everyone, which was the key prerequisite to my theories.

I haven’t considered what you’re doing as an option because I honestly don’t feel like it is healthy for the community. Noted, these are only theories, and I haven’t seen anyone try to work in this way, but I feel like there is a better arrangement than what you are doing now.


How to fund specific features/enhancements in a viable way?
(Grzesiek) #287

I recommend you read some of Ton’s own comments related to making money from open source.

Note that many companies like Red Hat, Microsoft all work on Open Source and make money of it… A lot.

I’m happy that Bliblubli is managing to perform these levels of optimizaiton, and as you stated he should be compensated for this.

To me this is no different then Blender Developer Fund., except that in this case you can put your money into specific function.

There was another developer who also did a lot of OpenCL optimization in the past, and I was glad to pay to get that extra performance. After a while he submitted his work back to Blender.

Now, there is also a possibility for Blender Institute to hire him, but that would mean Blender Institute would need a larger Blender Development Fund to be able to actually pay him.

Now I personally would rather supprot Blender Dev Fund mainly to ensure Blender Stability is top notch. Extra performance I can wait, but fixing crashes and such I can’t

Still in the end, it is up to each and every one of us to decide what we do with our money, you can go and support a specific developer like BliBluBli or Blender Dev Fund or Hire your own developer. No one is forcing anyone to support BliBluBli.

1 Like


In my benchmark above Cycles GPU beats V-Ray Next GPU (brute force). And currently as I know a lot of V-Ray’s feature simply not available with GPU.

Grzesiek: I think you got the point.


(smilebags) #289

I think we agree on almost everything we have said. Open Source projects can generate money. People should be paid for the work that they do (or inversely sensible people do work which will get them paid). People are free to spend their money how they wish.

Would you have been as happy to pay if the changes were merged into master as soon as they were deemed an improvement and reliable? I suspect most reasonable people would be, or maybe even more motivated to contribute.

And only a limited few see the benefits of it with this setup. This is really my only concern with what he is doing.

I don’t see any problem with freelancers finding funding to work on OSS. I feel like we as an industry do need to support this with our attitude, wallets and technology if we want it to succeed. I would love to do it one day. I feel we should be advocating for benefit for the masses rather than segregating people into those with the ability to pay (and therefore get better software) and those who can’t. That’s a recipe for spreading effort out needlessly.

Imagine Nvidia hires a dev to work on Blender for a year, but only releases their version of blender internally and under the table to other organisations. Now, this is unlikely, but you can see that it would indeed get Nvidia access to the software they paid for, and the developer would indeed get paid, but the community at large would not benefit at all. Luckily this isn’t happening yet (even with studios which are developing Blender for internal use) since developers and studios also are contributing their efforts back into the community.


How to fund specific features/enhancements in a viable way?
(Lumpengnom) #290

Why? I´m pretty sure this happens all the time.


(smilebags) #291

Proof or it didn’t happen. Maybe it does. If studios are modifying Blender for internal use and paying their devs, good on them. This is different. This isn’t a studio earning money by using software which they happen to make some improvements to.

Regardless, whether it happens or not doesn’t change the validity of my statement.


(Lumpengnom) #292

Well, not necessarily Nvidia but there must be some compainies doing this.

I as a freelancer am modifying Blender for my purposes. I think it is absurd to assume that no company ever modifies Blender for internal use and pays somebody to do so.


(Grzesiek) #293

Well as he stated and others who 've done somethign similar he will be submitting the change, just not on day 1.

The issue is that most people get so used to Blender being free that they expect everything related to it for free on day 1.

Now Nvidia and AMD have a big investment in Blender, lending hardware and developers to ensure their hardware works best, to ensure that if I want to spend my money on new hardware I will choose one over the other.

Now again, if someone will develop a new Game Engine for Blender, there will be people who expect it on day one for free. But if that team charge money we’ld be back to this discussion of free vs paid. It is unfortunately the aspect of the Open Source community.


(smilebags) #294

This is almost certainly the case, but it is also irrelevant. People will modify Blender for personal use and that’s completely okay.


(Nitram_2000) #295

The only thing I don’t like about this is the subscription service. It’s one of my huge bug bears with software companies over the last few years.

As a freelancer in television production, my work isn’t always guaranteed. That’s the life of a freelancer in any field. So when Adobe switched to their CC subscription model I held off for a long time. I had already paid for PS and AE and their version 6 held me in good stead for a few years afterwards. But now, alas, they have me cornered. So now I have to fork money over to them every year regardless of how much work I have on. Luckily I have a few regular clients so I’m never really stuck, but it’s still something that’s in my mind every financial year.

If this was made as a one time payment build (I’m not too concerned about learning to program my own builds) then I’d happily pay for it at a reasonable price like I do other Blender addons that I’ve purchased.


(Lumpengnom) #296

It is not for personal use, though but for business purposes. A freelancer is basically a one man company.
Anyway, do you really assume that no companies modify Blender without sending the patches to the Blender Foundation?


(Grzesiek) #297

Good point. Wasn’t that the case with the team that made that Netflix “Next Gen” movie using Blender?


(smilebags) #298

I understand that he intends to release the changes in a year. It is good that he’s doing that.

I agree that many assume that improvements to Blender ‘just happen’ and that they can get it all for free. They can. For them, everything really does just happen, and they can get it all for free. These people don’t play any part in this discussion.

Let’s go with the idea of a new game engine. If everyone expects the development to happen for free, then it simply won’t happen. If the team charges money for those who want to use it, they will either get enough interest for it to be viable, or they won’t. The deciding factor is how many people with money want this.

Now, if a team was to propose development for a new game engine and state what it will cost to make, and those same interested people contribute the same amount of money, then it gets built and everyone has it. The only difference between these situations is the exclusivity for those with money. I don’t think that’s a healthy thing to have in any community.


How to fund specific features/enhancements in a viable way?
(Grzesiek) #299

Agreed, the small difference here is that after a while everyone will get these improvements. If he wouldn’t commit to this, I would be fully behind you. :slight_smile:

1 Like

(smilebags) #300

I expect they do. I’m just saying this is irrelevant. No money is moving between the public and an organisation or individual, therefore it isn’t part of the answer to the question “What is the best way to stimulate independent development of Blender between the community and developers, freelance or in studios?”

Apologies, personal use was the wrong terminology. I should have said “internal use”.

1 Like

(smilebags) #301

And this is why I don’t see the motivation behind delaying that unnecessarily. That’s the essence of what I’m trying to contribute to the discussion. Whether bilibili decides to take my opinion on board is up to him.

Yep, it was great to see them contributing their significant improvements back in as soon as it was viable to do so. I understand they might have had to wait for a while not to unintentionally release any information about the film, but this sort of community contribution is exactly what I would love to see more of, both from studios and individuals. People getting paid and contributing their efforts back to everyone.

1 Like

(smilebags) #302

Would you still be happy to pay for the build if you knew that everyone would have the same changes in a few weeks/months? Would you still contribute money if you knew your money was helping getting the feature developed but that you weren’t going to be the only one using it?

If the answer is no, then I need to reconsider my ideas of how my ‘ideal situation’ would work in reality.