Quote from: Hetzen on October 07, 2011, 09:09:53 AM
Good efforts Martin. I agree with Seth unfortunately, I prefer the original, I think there's too much saturation in the processed version.
In the end, we went with blue sky and GI on the NE plate, it just looked better, and I had absolutely no time to finess due to the client changing the Cad data 1 1/2 weeks before delivery. Even so, I wasn't really happy with it.
We want to go back and do a few selected shots to show the client what they could have had if they had pulled a bloody finger out and gave us more than a few weeks to turn around the project. I've had a tentative go at doing an overcast version, but it's a nightmare to balance getting any light into the scene through overcast skies. It would be incredibly helpfull to have a "cast shadow opacity" in the cloud node that didn't rely on cloud density or effect the shadow in the cloud itself. The only other way to get around this would be to render the scene twice and create a horizon mask. This is also a good example where the alpha channel which didn't include clouds in the atmosphere or the atmosphere itself would be more useful than what we have at the moment. Anyway. Rant over.
Ghehe, well I never thought of agreeing with Seth as being "unfortunate"
just kidding
Funny observation though, because I didn't touch any colour/saturation slider in the proces as far as I remember.
The raw output I posted is from the .bmp output, while I used the .exr as a base for the first posted image.
Not sure if this contributes to it.
I like that image by the way. The overcast clouds add much to the scene, they also look pretty realistic themselves and the hint of bluesky really works for me.
I terribly agree
with you about the cloud opacity function you described. Unfortunately, and also a bit illogical if you'd ask me or to me, the light propagation settings in the clouds don't have any effects on the casted shadows from the clouds on the surface. It is density dependant though as far as I know. What one can try is to keep density low and increase edge sharpness. This way you can still simulate dense clouds without having actual dense clouds, so keep the benefits of brighter shadows beneath.
You can play a bit with the 2D cloud map function, but the softer you make it the less accurate the shadow becomes, so not ideal.
I also agree improved alpha functionality would greatly help in these situations.
Ok back to real work again
sigh