So I've got this still frame scene that Poser 5 just refuses to render. Apparently I am running out of memory and I crash to my OSX desktop about 5 minutes into the "Adding Objects" dialog. The scene is comprised of an "alley type" environment prop plus variety of figures (9 in total), V3s, M3s, Possettes and an assortment of props. My questions are with regards to rendering a complex scene in layers for compositing later in Photoshop or the like. Since I've never done this before I'm looking for some tips. And specifically, how to you handle shadows and raytacing...or are those options just out of the question when rendering in layers? Thank you

You could pull out the characters and render them over a blue background. Then render the scene without them. Then in photoshop make the blue color transparent and layer the scene under the characters. Then you can paint in the shadows with photoshop.

I render from the back to the front in layers. Save the PZ3 first. Now when ever you click to open the PZ3, every thing is exactly in place including lights. Now all you have to is render with less figures. Do it in sections. Tricks like render the scene with no figures for the Main background. Add a few figures, render those in the scene, you can get thier shadows working. Close Poser then open PZ3 again and render the other figure in scene. I render my figures without clothes and hair. Then render new layers with clothes and hair. All this helps poser render. I save everything as a tiff. You should see how big and how many layers I have in PSP. I also work form the back to the front in PSP to cut down on the size and import layers as I go along. That's one of the reasons I love HMANN hair's. I can render all the parts individually. Better efect and no poke throughs.

Thanks beryld - that's very helpful! Another question regarding the way Poser 5 handles memory, loading textures, etc. Is it enough to simply make the clothing and prop elements invisible for a render pass?... or must I delete them from the scene altogether in order to free up resources? thanks to both of you for taking the time to post.

If it's 1 character, you should be able to make things invisable. If you have a lot of stuff, I'd suggest delet. You don't lose anything because you have your saved PZ3/PZZ.

Once I have my basic backgrounds and figures rendered with the background for shadows. I delet everything but one figure including the background/scene and make all my layers for that, with the render with black background checked,then go on to the next. It makes for a lot of rendering and a lot of layers but I seem to get better results. Don't forget you can save smaller PZ3s from the main one to help with the rendering. Delete what you don't need and save.

Message edited on: 10/24/2004 05:09

That's the exact process I use too. It works wonders! :D

Great answer to the question beryld. I just got P5 and I haven't rendered that complicated of a scene with it yet but I'll definitely keep your advice in mind.

Ive used the back to front method before and it does work but I like using the blue screen and doing several different layers it allows you to add different lighting and shading effects in Photoshop to each layer.

I have always wondered if it would help to save the human and clothing figures out as .obj files delete the characters and bring back in the .obj for the render? I know hair would not work as I would loose the transperancy settings. I saw a tutorial once about using boxes (and other primatives I guess) to establsh shadows for compositing onto photos... I suppose you might try doing blue screening with blu primatives to take the shadows? The primatives would be placec where the background scene forms would indicate?

There is not 1 way of doing this, everyone will find what works best for themselves. I use the Black background with a Tiff. I find that all I have to do is load it into Gimp. Done, Nothing else to do, I don't need to play with the blue or what ever color. I just save the PSD file to use any where else. I don't have PhotoShop, I have PSP8.1 and Gimp. Different programs, different ways. All we can do is give as many tips as we can and hope some of it works for others. 1 last tip, render twice the size you want your finished image to be, better for postwork.

I agree with all the above cool ideas if poser internal rendering is really your only option. I don't do a lot of P5-only assembly work or rendering. (And admittedly, my machines have enough firepower this is rarely a problem on my home jobs.) In Renderman (and some MAX and LW renderers) there's a "receive cast shads only" shader you can apply to the whole scene for one specific pass, and it just outputs an all black frame with a white alpha where shadows fall. You then use this alpha to darken through as an effects layer on your cast-shadowless (fast rendering) background and beauty passes -- only self-shadows are needed. It's FAST, because it doesn't load a bunch of textures, just one shader for all over every object. (Maybe there's a P5 equivalent I'm not aware of?) Professional FX houses do a LOT of multipass rendering with a final compositing step to mass-assemble and tweak trick-shadow, crowded 3D scenes. I have Combustion and AfterEffects, which both have "two-and-a-half-D" tricks and cheats that help with cross-layer shadow making -- at least on shadows that are not cast across critical areas of subject interest. (When two main objects inteact that closely, they should probably be in the same render pass anyway.) Those tricks generally involve taking a copy of the alpha outline of a character and either doing actual quickie 3D z-offset one-lights on it, or else 2D "squishing" or "squeezing" it according to the direction of light source in the scene, and then using the result to darken any layers behind the object. And yes, you CAN do this same thing manually by eye with photoshop or PSP and their various shape distortion tools (I have done so on small jobs) but the trick-lighting in the upscale composite programs is comparatively pretty spiffy and quick, esp. for many-frame anims. The absolutely critical thing always is do these two objects cast shadows on each other? If not, and render power is limited, they should always be in a seperate pass. (Note in an animation, which object is or is not interacting with others often CHANGES, and your passes sometimes have to be frame range specific!) Does this really save time if you have to do many multiple passes? YES. Each part summed is is still almost always less than the whole shebang at once. Not only that, multiple passes let you do color (or blur, or other) tweaks to seperate elements in your composite easily without affecting others. If it's all one piece, your changes often help one thing while hurting another, and you have to figure nasty ways to split them up and treat them seperately after the fact. Seperate multiple passes are a GOOD thing. Sometimes this even means passes with some lights on and off, to be composite-mixed-and/or-tinted, or even rendering your specular highlight passes seperate from your diffuse light passes to be treated differently. (Note these things are less critical in a still frame whech can be rather easily paint-tweaked, but become critical in animation sequences of hundreds of frames.) For masses of ground shadows when the light source is high above, Rather than actually render a ground shadow, I almost always just take a layer copy of the alpha shape outline of the characters, squish it vertically flat (to the degree that seems like a reasonable shadow "length" for the scene and light angle) and then maybe offset it or blur/distort it a little. I then use it to darken the ground-only render layer beneath them in the comp, maybe adding some color tint based on the ground surface. Like it's already been said, There's always more than one way to UV skin a Poser Cat. - - -

Another trick is to make reduced resolution Textures Maps for objects further in the background. 4000x4000 photo realistic texture for background people is insane. I sometimes make reduced lower res copies of textures for this purpose. Photoshop and about 15 seconds reduces render times alot.

Although a couple posts mention using TIFF renders, the reason for doing so is that Poser saves an alpha channel of just the rendered figures/objects in the scene. That means, when brought into a paint program, a selection can be loaded and automatically select everything that was rendered from the default background. If an entire background scene is rendered, it will be part of the selection.

You can take japes' great "lower res textures" point a step farther by even using lower res geometry for things that will remain distant in the scene as well. You have to be careful about things that will approach camera eventually in an animation, obviously -- e.g., the Vicky Reduced Rez figures in the BG (or even posettes, or super low rez substitutes for extremely far distant people) and the full rez Vicky3 only as the foreground "hero" character(s). Also, is your render going to have fairly narrow Depth of Field? If only one plane of the images is going to be in clear focus, all objects closer or nearer than that plane which will be blurry anyway should get lower resolution textures and/or geometry -- why expensively render detail only to expensively blur it away later? - - - --

Whoops... (that post above should read "closer or farther", obviously)

Stewer made a material in P5 that you could apply to the ground plane that would make the ground plane invisible, but still receive an objects shadow. By doing this, you could render your person with a shadow, and export the picture to use the alpha channel to extract the figure and shadow. Then you could layer your pic in a paint program with shadows.

Compositing works fine for shadows, but doesn't work if you're using raytraced reflections - delete the foreground figures to render the BG, there's a reflective surface in the BG, the now deleted figures don't get reflected - bit of a bugger Only solution I found was to upgrade to Shade Render qualities better too

Stewer made a material in P5 that you could apply to the ground plane that would make the ground plane invisible, but still receive an objects shadow.

Here it is

Life is better with links :)

Actually, regarding Rubermatt's comment on compositing and raytraced reflections, I've managed to do lots of neat tricks by making copied layers of the beauty passes of object renders, flipping or flopping them, messing with their size, distortion, and transluceny, and then blending them back into the inside of the alpha outline of the object they are supposed to be reflected into -- SO -- comp tricks can in fact, take care of some reflection issues as well as shadows (inverted and modified "reflections" of people standing on shiny floors or rippling ponds come to mind, both of which I've often done as a 2D trick rather than a raytrace render). It all just depends on how clever you want to get with your compositing tricks and how much time you need to save by omitting raytrace in your complex render passes. --- Thanks for the mat link, BTW -- that's a great and useful material. --