Well, the thing is, doing it in layers just gives so many options, especially since stuff like Speedtrees don't let you adjust the textures. You can do all the depth of field stuff inside Hitfilm and get more lighting effects to boot. It was how I had to work in Poser, but anyhow, until either hair or DOF gets an overhaul, I do think I will just be planning out my shots with that in mind. It really is the way to save on a lot of unnecessary rendering anyhow, if you just need a 10 second background clip that is blurred behind characters that gets cut back to every few seconds as the camera moves, why render it in whole anyway? That way you can add more to the background anyway. I wish it were possible in iClone to put different items into display groups and then be able to toggle the visibility of the groups as opposed to individual items, correct me if I'm wrong, but this isn't currently possible. I think on the rendering end at some v7 point release, it would be great to see much more flexibility, including rendering a depth map automatically. But maybe all of that will happen with the Python support and open API. One thing I'm curious to see is how rendering using GI but no sky/background will work, will water get the benefit from the PBR, I guess we'll know soon enough. Anyhow, when in the history of film making have you NOT had to use work-arounds? The whole illusion of cinema is a work around that tricks you to thinking it is real. 

~~~~~~~~~~~~~~~~~~


My system: i7 3770 Gtx1060 6GB

Yeah. A gaussian blur is going to sample from stuff around it, so it might not do a very nice job with hair. I suppose you could use a depth map for cutting your clip up for layering. Maybe that's an easier way to get your layers?

I've been messing with the depth map method using HitFilm Lens Blur, but I don't get good results, the blur isn't very nice. It's also terribly slow. I've gotten much nicer results using layers, the Hitfilm blurs especially of trees look horrible. Maybe I'm not doing it right, it works, but it has its own set of issues. 

~~~~~~~~~~~~~~~~~~


My system: i7 3770 Gtx1060 6GB

You might want to make your own depth map so you can do your blurring to simulate DOF in your video editor. 

Set your rendering mode to "quick mode". This turns off anything that is shiny or glowy.
Turn off your lights.
Set the ambient light to black.
Turn fog on, set to white. Click on the "effect sky" box. Start it at something close like 200, and extend it to your own preference.

Render your movie clips again with this method, and then use it as a mask for blurring farther or nearer the camera. This has the benefit of a proper progressive gradation, rather than discrete layers. I think it's probably easier than doing layering as well.

You can see it in her ponytail as it crosses the background. The greater the DOF the more this is an issue. 

~~~~~~~~~~~~~~~~~~


My system: i7 3770 Gtx1060 6GB

I figure these are the most obvious of options:

don't use it
try to avoid shots when it looks bad, even if it means you miss the shot
render in layers and add the depth of field in HitFilm
make hair props etc that don't have transparent fibers
Use Alpha Threshold a lot

So I'm just wondering, how do you work around the issue? I am kind of thinking the Hitfilm approach is the best for me, it does separate everything and give you a lot more freedom with composition, but I would love to hear from more experienced users how they approach this. The main issue with that approach is that iClone doesn't do this automatically, and how different objects interact with light and shadows is an issue. Anyhow, thoughts?

~~~~~~~~~~~~~~~~~~


My system: i7 3770 Gtx1060 6GB