welcome and thanks for the feedback.
Which level did you use for the AI denoiser? I would need a file to reproduce and investigate what is going on here.
2.8x now has the exact same quick settings panel as 2.79x (using the latest builds). So the UI should look about the same apart from vertical (2.8) vs horizontal (2.79) tabs. If it’s still unclear with the latest builds, I can redo a tutorial for 2.8 specifically.
As noted in the first post of this thread, the noise compatibility option is a bonus for 2.79 only for now. If their is enough demand, it could be a feature update for 2.8x later.
I’ll try to extend the number of case the addon work with. In the mean time, you can:
hit B to box select the render layers node and all the denoising nodes,
hit shift+D to duplicate
switch the render layers node to the View layer you want.
reactivate the layers like in the default one (it will automatically restore the links for you)
There’s another bug I found. When you hit the node creation button and there’s no “Composite” node, an error pops up. In my scenes I often just have a “File Output” node and nothing else.
Thanks for the file.
2.79x users are mostly professionals who need stability and also require lot of support for specific use cases. To give you an idea, where I work, an hour of support is sold 129€. For E-Cycles, the first month is the one where the most support is required. The best I could do would be to offer a base build with one hour of support for 89€. Then 79€/hour of support or per feature update.
Noise compatibility can be ported to 2.8x for the next feature update in 2 weeks if their is enough demand for it.
I’m working on a fix for the case their is no composite node.
It should be faster out of the box. Maybe you have your CPU enabled in the cuda devices? If not and it’s still slow, you can send me a file per PM and I’ll have a look.
I’m not interested in your support rates to be honest. I’m interested in buying a product. If that product works and comes with proper documentation, support is reduced to a minimum anyway. As you can see here, the support I needed was for things that are buggy or not explained in the accompanying files. And if you speak of your rates you should not forget about the rates of the people who buy the product. If the product is faulty, time goes into it for testing, fixing and workarounds. That’s not a one-sided issue.
Paying 170 Euros for testing a product just is too much. That’s why I wish there was a monthly option. 14 Euros for a quick test is OK. But as I’ve seen now, working with the 2.8 version is too problematic still and the time saved doesn’t make a big difference. So I hope the 14 Euros are well spent and the features you work on will make it into the master one day.
Hey Andreas,
you asked, I answered honestly. Now you know why I don’t offer a 2.79 monthly, only a handful people want to pay that.
The addon is a bonus I kindly made to make the experience better in most cases. Even the denoise node was a bonus the real feature update was a 5% speedup. That the addon doesn’t work in 100% of the cases after a few weeks of release is I think pretty usual. Graswald and many other addons I bought are out since a long time and still have a lot of rough edges in 2.8 since the API changes a lot.
Please stay cool, I was actually working hard right now fixing the 2 bugs you found and you can already find a fixed beta version on the product page. Most bugs I get reported get fixed withing 24 hours, you got it fixed in about an hour.
I appreciate honesty, that’s why I’m honest as well. I never stated otherwise.
I’m not sure how you can tell how many people would by the monthly option of 2.79 if you don’t offer it. I’m sure many users are shying away when they see the price tag. That’s a lot of money for a shot in the dark.
I agree that software can be bumpy after release yet users have to pay a premium price. That’s what a beta phase or lower price phases are for. If I pay 170 Euros I pretty much should get a rather polished product.
I think it would have been better to have offered the addon for 30 to 50 Euros. A lot more people would have bought it (including me) and you might have got the same income - maybe even more.
Kudos for your quick updates. Less to worry about for you. Those issues would have come up at some point anyway.
Actually I’m a coder, not sale person. Yes, I could have sold the addon instead of doing a free bonus. I hope most of my customers are happy with the extras. I can do longer testing phase, but then their would be less bonuses. Regarding beta, it was indeed with 40% off. I still have lot of customers at 6€/month. I think it’s a good deal to have the bonuses early with them working in 80% of the cases at beginning and slowly getting better through feedback.
Although I won’t use the addon, I’m glad that you pointed out the hidden potential in the Cycles engine. I hope that developers will pick them up and use them in the master one day.
Can you implement the ability to click a button to tell blender to “stop rendering the current bucket, it looks good enough, move on to the next bucket !”
Yes it will automatically move on, but I am talking about the human on his own free will, decide this portion is visually good enough and command the renderer to move on to render the next bucket instead. I don’t care if the renderer says “no no I am all about the adaptive sampling and I will tell you human when I think is enough”.
What you are asking does not sound practical to me. Most of the time you may end up with a resulting image where the eye could distinguish the difference between the tiles or boundary between tiles.
I understand where you are coming from, I am just requesting that we have the option where I can be the judge of that.
Recently I was rendering a large scale still image on 10,000 samples, Cycles fxxking spend a good 5 minutes or so rendering the part/tiles where it was just a still image background instead of focusing all the processing power on the 3D model composite !
I can understand if say we use a matte shadow material because obviously that would mean sampling is needed for the matte shadows but not when I didn’t use that.
I am telling you, Blender really need a “move on to the next tile” function during rendering.
it’s all could be solved by adaptive sampling.
10к is too much, in production we rarely go higher than 128-256 spp.
“Move onto the next tile” button it’s a regression to manual sampling
