Bucket size is about memory. With a GPU use the largest bucket size possible without running out of memory as only one bucket will render at a time (IIRC). With CPU as there will be one bucket rendering for each thread it's better to use a smaller bucket size. CUDA cores are about speed - the more there are, the faster the card will render.

As I understand things, if you're using GPU rendering it is dependent on VRAM, and if you're using more than one GPU the VRAM available is whatever is on the largest card only. If you exceed that memory the render will immediately fail. The log will give stats on timing and memory usage. It's up to the user to manage the scene - if the scene is too large, switch to CPU or do things like reducing texture size, replace maps with procedurals etc to optimise it.

The built in way is to use Dependent Parameters. Create a new Master Parameter by clicking on the little white triangle at the side of the Parameters palette and name it. Go to Keyed Dependencies, teach the maximum first, then zero everything, then teach the minimum if required. So press Start Teaching, set the Master dial to 1, set whatever controls you want at their maximum effect, then set the Master dial to zero and make sure all those Parameters are zeroed. Test it and adjust the Settings for your new dial. Then you can make your own dial groups, name them how you like, drag them into an arrangement you like, nest them, and so on.

Refer to the manual - Parameter Dial Groups p219 and Editing Master Parameters, p822.

https://vimeo.com/mattpinchbeck/gogmagog

Embedding doesn't seem to work, oh well ...

I made that video nearly 3 years ago for the first commercial product I made for Superfly. It's step by step from default startup scene to final render (a demonstration, not tutorial as I know that my knowledge is limited and that 3d, which to me is essentially a problem-solving exercise, offers more than one way to get similar results). The product I made it for is called Gogmagog and it is currently available at Hivewire free of charge as part of their Stay Home Spread Hope category (I am considering making it a permanent freebie and and uploading it here as well). The set includes a couple of high-res 16k panoramic skies (I spent nearly £500 on hardware and software to produce those!).

@FVerbaas - would some controls for offsetting scattered objects in x, y and z be useful? I might suggest that in the thread at HW. Also, the polygon reduction feature might be useful to change the arrangement of vertices on an object used for a scatter plane?

@ockham - I've been using the construct with my own pano skies for quite a while, as the screenshot shows, and not had any problems so far though I haven't rendered very large reflective surfaces. What issue are you seeing with the construct?

Do you have either the Construct or the original Poser Ground in the scene? I think it looks for one or the other of those - they don't have to be used though, or visible. I'm not sure of that but I'd suggest try adding either one of those and see if it makes any difference?

@Rhia474 - I'm sure Ken will get this sorted for you and hope you're up and running soon!

I think it'll be very useful. There are a lot of options but the videos on the product page will cover them plus Ken is working on a volume scatter option which looks good so far. I've just done a very quick test with a few of my own props - 20x trees at about a quarter million polygons each, 40x grass props at a few hundred polys, rendered in Superfly with only 8 samples (v. noisy) at 1400x600 pixels, i7-3930K, 145 seconds at 10Gb RAM. Superfly did take a few minutes to load the geometry :D

Nice one Ghostman, thanks!

I tried exporting fibremesh using GoZ; imported as a hair prop; converted to a figure using the Fitting Room and get the same issue - it is visible in Preview but disappears when rendered. Using Poser File Editor to look at the pp2 I can see the hair prop is in there but there is no geometry so it is not recognised by PFE as a figure. I would guess that the render engine has the same trouble - it expects geometry and can't 'see' the hair prop.

Indeed ;)

I'm not sure if they were generated by EZSkin but they certainly draw on it. Personally I think that the limiting factor with skin is now more likely to be with the maps rather than the shaders.

EZSkin and other scripts were created to make it easy to update older content by adding SSS and Superfly-compatible materials. La Femme and L' Homme already have up-to-date materials built for P11.

If you want to convert the fibremesh hair into a conforming figure I think - though I'm not certain - that you might need to export the fibremesh as geometry (i.e. using Export from Zbrush to get an obj rather than GoZ). I think that when you use GoZ with fibremesh it becomes static strand hair in Poser, and I don't think strand hair can be turned into a figure.

Normal maps can produce different results in Superfly, Firefly and Preview as the tangent basis in each is different - see this link http://wiki.polycount.com/wiki/Normal_Map_Technical_Details#Tangent_Basis (Superfly uses MikkT which is the most common; Blender is probably using this for baking). Tangent basis is not the same as tangent space, and nothing to do with OpenGL/DirectX either.

Much simpler to use a height map in Poser ;)

Baked normals are really useful for making other maps to use when texturing though, e.g. curvature.