Hi,
Janusz, thanks for your comments and advices!
My setup is not Phsky+Phsun, so as I read your advices, I was very happy to hear about rendertime boosting otions
I made some test:
01 - 8:08 (488sec)
I removed all the vegetation, the original setup is the same:
- PathTracing fg 100/1600
- Global Photon Power: 1.5 (Yes, like Janusz said - I also realized this)
- Photons: 0 - ONLY PathTracing, surface shaders also use only PT
- Light setup: GI dome (200% lumi sphere 7000K, unseen by cam, no shadows) + Sun (LW Area, 1km distance, 100%, 6500K), NO other light or lumi poly
- tone mapping: Gamma 1.0/1.0 (I want high contrast. maybe I could rise it to 1.2 or 1.4, but it's OK for me) - this results the native LW render look

, Limit DR Before Tonemap
- walls surface shader: force PathTracing with fg 1000/2000 (to remove the noise from the most visible surface)
- leaves (removed from this test, but: surface shader: force PathTracing with fg 0/0) - I know it results noise, but the deadlines....
- AA high (for this test I changed the grid size from 3 to 2 - The original images were made with grid 3)
02 - 3:45 (225sec) - this will be the new base image
To speed up test renders, I modified some parameters:
- global fg down to 100/500
- surface shaders removed fro walls
Now the speed is ok, but the noise is not - but now I dont't care about noise, I want to test as fast as I can.
03 - 2:28 (148sec)
- GI dome (200% lumi dome 7000K, unseen by cam, no shadows) turned off
- Sky (ball with spherical image, unseen by Rad., no shadows) turned off
- Tree images (images on planes, unseen by Rad., no shadows)
- Sun (LW Area, 1km 100%, 6500K) turned off
The only lightsource was the white LW background - The boosted renderspeed is obvious
04 - 3:26 (206sec)
- Phsky with sun turned on (sun position almost matches the original)
Conclusion in terms of rendertimes: If I use Phsky + Phsun instead GI dome + LW Area I gain 10% on render times, but the noise level is the same