Actually, that was my original goal: just to say thank you and give something back. With every step of development my project became more complex but also more confusing. There are currently different versions, with different problems and solutions. The node-setups are a huge mess and the handling of the relevant parameters is completely undocumented. As described elsewhere, due to the grown complexity and the already spent time and effort, I wanted to draw a line for the time being. The renderings I present here are the best I could get out of the status quo and are not necessarily representative of the craftsmanship of the file itself XD. In the current state, releasing a source file would therefore be more of an inconvenience than a pleasure for me.
I need some distance to the project for now, but I still keep the publication of a usable source file in mind. As soon as my motivation picks up again, Iām looking forward to share my project progress and know-how with the coummunity.
I hope that my thoughts on this are comprehensible for all interested people and therefore I ask for some patience.
Deep respect. I love what you have done, the results are very beautiful indeed. One thing caught my eye, the cloud shadows on image 03 seem too long compared to the apparent depth of the atmosphere seen at the horizon.
Wow, the difference is as impressive as it is frustrating XD. However, Iām very happy to see that the file doesnāt render as inefficiently on appropriate hardware as I experience.
On my hardware, with identical settings (but without Turbo Tools), almost 4h are announced:
Well, Turbo Tools let me use around 160 samples and for the most part it doesnāt even use that many. The 3080 Ti then does the rest, got to love those RT cores.
When using adaptive sampling and denoising itās common that the estimated numbers are high in the beginning.
But probably given the gradient in the sky for instance it might be that this part will need much less sample and adaptive sampling will work itās magic.
Anyway since itās a lot of volume with complex textures itās natural to get long render times.
But Iām sure with a few samples and denoising you can get good results after a few minutes.
I absolutely agree with you. However, in the case of the clouds, the denoising does not achieve very satisfactory results below about 500 samples in my opinion. If I want quality, I have to suffer XD . So, I guess, as long as I canāt afford a better graphics card, Iāll just have to deal with it.
Oh, and since this is some sort of benchmark, on a MacBook Pro (Intel QuadCore 2.3Ghz) the scene as downloaded announces an estimated render time of 12 hours 50 minutes.
