So I am running Poser Pro 2014 on Windows 7 Pro, with an I5 processor and 8 GB of RAM and a 4 TB hard drive. I don't remember the specific graphics card, but it's a middle tier Nvidia card.

Recently I made probably the largest and most complicated scene I've ever done: a warehouse filled with containers, pallets, loose wires, boxes, shelving, etc. There's probably well over 100 objects between figures and props.

This scene is basically killing my system. It takes close to 5 minutes to load the scene, the preview window is very slow and jittery, and it takes roughly an hour to render one image at 1600 X 1200.

I know the render times are directly related to my processor, but I am wondering what speeds up the preview window? Is that also processor based, or is it more the RAM or graphics card that would help with that?

And what about scene load times? I assume that is purely related to hard drive performance, and maybe RAM, but has nothing to do with processor speed? I'm saving the files in compressed format, so I know the scene has to uncompress to load, but 5 minutes seems like a very long time to wait before Poser can actually be utilized.

I'm planing some upgrades, but I'm just wondering what will provide the most bang for my buck because I'm on a budget. Otherwise I'd just build a whole new rendering rig from scratch.

1.  Buy the GameDev upgrade ($40) 

2.  Read the GameDev Addendum.

Use the Polyreduce features along with the ability to bake figures to a single mesh.  I suspect that your scene is in the multi-gigabyte range - you need to make it smaller.

After that - get more ram.

I'd thought of that as well ssgbryan, but that route has too many unknowns for me right now. 

How many figures/objects can be baked into one?

Do they all have to be conformed to one another?

Do all morphs work from the original figure?

My current understanding of the new features would be that I would have to individually reduce the polygons on each figure as they are not conformed to one antother and probably can't be. I can't conform the freight container that's been moved way over into the corner to the warehouse floor.

Not to mention the polygon reduction would affect how good the figure looks.

Thank you for the suggestion, but I don't think that is a viable solution.

First your MB/system

What socket is you MB?

What version of PCIE does your MB support?

Get the biggest/best i7 that will play on your MB/PS combo

Now your work flow (or stoppage in this case)

You start by separating your working scene from the rendering "scene"

You only need morphs in a scene when your sculpting a character

you save as the file (new name) then apply all the poly reductions and other techniques (removing morphs etc.)save and shut down Poser (maybe restart) reopen Poser and the new file , then render

Quote - Get the biggest/best i7 that will play on your MB/PS combo

Depending on the chipset on the mobo, and I7 may not be a huge improvement other than core count. It will still be stuck at whatever the thruput is on the board. Sort of hard to guess without knowing what motherboard it is. If the I5 is an LGA1150 socket, the bios may support an I7, possibly even a Xeon.

It would probably be a better route to build a twin processor Xeon system than waste the money on an I7. A top of the line I7 can be over a grand. Yes, that is a lot of money out at once either way, so It is a big decision over plugging parts into what you have.

For less than the price of a fast I7, you can buy 2 Xeons, the mobo, and the memory for it. The Quad core (8 threads) Xeons are fast, up the core count and they slow down due to the TRL (turbo ratio lock) differences between an I7 and a Xeon. With all core active, like rendering, the Xeons win hands down.

The E3-1241 V3 Xeon is faster than any quad core I7, but it is a single socket chip. For a 1/4 the price of a comparable I7.... Ironic isn't it.... Yes it uses a slower memory controller, but it makes up for it in other ways.

The E5600 Westmere Xeons are not as fast clock speed wise, but the thruput is still just as fast as an I7. And they are scalable. (multi cpu capable) Motherboards for them are not that bad pricewise either.

Lot of choices, but we need to know what you actually have now to make any type of informed decision about upgrading or replacing it..

A ) poly decimators will load scene faster.
http://render.otoy.com/ and a realtime nvidia 3D Card.
will get you Realtime renders .

B ) buy badest PC known to man

C ) Load & Render in a faster app

D ) render partial scenes peace renders together in PhotoShop

Temp fix change the camera from openGL to Sreed display. Then your scene won't wait around to recalculate and display the shadow details if you move the camera or a figure or prop even just a nudge

Use this all the time when a scene I am creating gets over populated and starts to bog down.

If you're not planning on replacing the entire system and just want a speed boost and reliability increase with large scenes get more ram if your mobo will allow you to. Large scenes eat up ram and if you hit the pagefile your HD will be swapping and you'll lose a large chunk of performance unless you have an SSD.

Try lowering your preview texture resolution to 256 or 128 from the default 512 if your preview window is acting very laggy.

Should have mentioned the Motherboard. Haha!

Core I5 2500K, Socket 1155

Asustek P8Z68-V LE, Intel Z68 chipset

8 GB of DDR3 with two full expansion slots open. Plenty of room for more RAM.

Graphics is an AMD Radeon HD 6700 (my bad, I was sure it was a GFX card of some kind)

Windows 7 64 bit, Poser Pro 64 bit, and yes I'm running the 64 bit version of the Firefly rendering engine. I didn't even install the 32 bit Poser files, but I've verified that the system is calling the 64 bit version of Firefly when I'm rendering.

I was planning to upgrade to an I7 4770K, and probably 8 to 16 GB of more RAM. I figured that was a quick, and fairly cheap, upgrade that should provide a noticeably improvement. The odd part about the RAM is that it never seems to use more than 4 GB, even though I have 8 GB available. Even when the processor is maxed out rendering, it's still only using 4 GB.

I thought I'd tried Sreed, with no change, but I'll try that again. Thanks!

I hadn't thought of changing the texture size for Preview either. I had to dig some to find that setting, but I did find it and I'll try that too. Thank you!

Is the 1155 interchangeable w 1150

The wattage requirement for this one is pretty upgrade friendly and will max out the processor dept.

Intel Core i7-3770K Ivy Bridge Quad-Core 3.5GHz (3.9GHz Turbo) LGA 1155 77W Desktop Processor Intel HD Graphics 4000 BX80637I73770K

My approach is to set the base scene (in your case, the warehouse) and set up exteriors (stuff visible through windows), skydome, sunlight. Do test renders until I am satisfied with the lighting. At that point, when I start populating the scene, I change to wireframe mode. I have scenes with over 500 objects and characters (many are between .9 and 1.2Gb files) and can move around in it just fine. I could do so on my i5 laptop as well. The problem is that Poser needs to load the textures for everything you bring in to the scene, and if you move around in the scene, you are asking Poser to recalculate light and shadow for everything. That would be my guess as to why the preview screen is slow and jerky. Take away the lights and textures by using wireframe, and that problem is gone.

I also save my scenes in wireframe, so when I open them, Poser doesn't have to load textures. The scene loads an awful lot faster.

Before upgrading anything, give that a try and see if it speeds things up. To be honest, I hated wireframe at first, but I soon got used to it and find that I now have a better idea of what's going on in a scene than I dod before with full shadows and textures.

just a thought... before reeling in new HW and SW and tossing existing stuff out of the window...

 - what's the size of your Poser file compressed / uncompressed
So we can compare, I'm running win7pro 64 and Poserpro2014 as well

 - how many texture files are you using, at what size?
I think this is the one. All files have to be imported and converted to HDR (exr) format into your texture cache. Try menu Render > Reload Textures and see what happens. Does it take that long too?

 - is your Texture Cache big enough (general prefs). If not, than only a limited portion of your textures survives between Poser sessions, and all the other ones have to be reloaded. The default 500MB will probably not be enough - I had scenes requiring 10Gb or so.
Just load the scene, then see what's in your %temp%PoserPro10PoserTextureCache

 - check your user-mem usage, in Taskmanager (Processes), especially when rendering. Do render as a separate process anyway.

food for thought: when you need 10GB for textures and you've got 8Gb ram, I guess something gets slowed down somewhere.

I do think polycount hardly matters. Andy2 is 20k verts and requires 10Mb in the scene. Complex hires chess piece: 15k verts. Growing hair on Andys head with 1 hair per cm2: 170k verts.

textures do matter, 1500x1500 exr file: 10MB. So each 4000x4000 map will do about 90-100MB.

A 100 objects is a lot. Whatever machine you'll buy, the loading most likely will not speed up at all. Unless you're going to use a solid state hard drive, but that would be kind of throwing away money.

aRtBee has some good advice and I like to add something to it. with such a scene I can image your preview being slow, there's a lot to handle by your graphics card. Go into render settings and select the preview tab. There's a feature called Texture Display and bring the number there down to 256 or even 128 if needed. Also make sure Enable Harware Shading is off. That should speed up your display a bit. You can also go into your graphics cards settings program/drivers (not sure what that is for Nvidia, only have ATI) and set the 3D settings to speed instead of quality. Standard it's set most likely to balanced or so, but go for speed in this case. You can set it back later when done. That helped me a lot to get rid of the slowness is some scenes.

Then there's the i5...... it isn't the best rendering cpu there is. If your motherboard can handle a better then go for it I would say and adding more memory is never a bad idea. But you have to ask yourself this question first..... is this a one time project or will it happen much more often? If it is something you'll do occasionally then don't bother doing all kinds of upgrades. If you are struggling with your system more often then consider saving up for something new or better.

Thanks everyone!

I will try to look at this stuff this evening.

Your motherboard supports the Intel Xeon E3 series from bios 3901 and up.

The I7 2700 Sandybridge appears to be the highest supported I7, and the Xeon E3-1245 will be faster even thou it is clocked a little slower due to the TRL and math unit differences when all cores are active. It is less expensive than a I7 2600 as well.

The ES-1270 V3 will probably work on the board as well, but I would ask ASUS if that is the case just to be sure.

On the flipside of this, the I7's are overclockable with proper cooling and the Xeons are limited in what you can get away with overclock wise.

Those motherboards are also know for memory issues above 16gig. (second channel doa syndrome)

grichter - Switching to Sreed improves things noticeably, but it's still pretty choppy.

piersyf - I'm not a fan of wireframe at all. I've tried it. I can't see poke throughs, and (more importantly) I can't fix them with the morph brush in wire frame. Maybe that would change with practice, but I'm too busy learning the material room right now. :P  Good suggestion though.

However, changing to wireframe makes the preview Window just fly even with the scene I'm talking about loaded. So it's almost certainly a texture loading issue.

JohnDoe641 - Lowering the texture resoution for the preview window to 128 improves things slightly, and by slightly I mean barely noticeable. :)

ArtBee - At its largest, the save file is only 40 Mb compressed. The current iteration of the scene that I'm working with is 30 MB compressed, and 135 uncompressed. While loading the scene, RAM usage jumps to 4 GB then slowly increases from 4 GB to 6.5 GB and evens back out at about 5 GB. Texture cache has 164 items in it with the scene loaded, totalling 3.06 GB in size. Reloading textures takes forever - render/reload textures.

shvrdavid - Maybe it shouldn't be, but this is my main day to day and gaming system at home. So while an Xeon may render faster, I think the I7 would be a more appropriate choice for this Rig until I can build a dedicated rendering machine (which will be a long, long time). What are your thoughts though?

There is one other thing I've noticed. Because this scene is slow sometimes I'll open a new scene with File - New. I do this to work on morphs or materials, then save the modified object and reload the original scene. I add the modified object back in from my library. This saves me trying to work on it with a super slow preview window. When I do that a couple times though, and try to reload the original scene, then Poser becomes non-responsive and I have to End the Process. I then open Poser again and load the scene, and I'm back to just slow instead of non-responsive.

So this really sounds like a texture loading problem to me.

to me as well...

Solutions:

 - using procedural colors/textures instead of maps

 - re-using the maps over multiple objects

 - shrinking map sizes.

Note: a texturemap needs about twice as much pixels (in its visible area) as shown in the render result. With your vast amount of objects, each one only can take a small portion of the render image, and hence won't need a large texture. It's some work, but a real saver.

just to be of some help, let me do the math in reversed way: how much texture does one need? Just to give you some clues in advance.

a) texture maps need to be at least twice the resolution of your render, so 1MP (megapixel 1000x1000) render requires 4MP texture visible.

b) Each object on the avarage is at least 25% visible (50% when full in camera, 75% with mirrors in the back etc), so 1MP render result requires 4MP /25% = 16MP full texturing.

c) using texture on texture, like decals, diffuse + specular + bump but also dress over figure etc, makes another factor 2 to 6, so let's say each 1MP render takes 16x2 to 16x6 or 30 - 100MP texturing.

d) in my first post, I mentioned 1500x1500 EXR requiring over 10MByte, so say 1MP texturing takes 5MB storage, and as a result 1MPixel render takes 30x5=150 to 500MB Texture Cache.

So when facing < 100MB per MP, one has to rethink his texturing for quality reasons. When facing > 500MB per MP, one has to rethink his texturing for performance reasons. Especially when doing fine print (5000x7000=>35MP/image) or real large screen cinema, texture management means something (for instance, the Jurassic Park T-Rex took 4GB on textues alone).

BTW: I dedicated a (4GB flexible) portion of my (24GB) physical RAM to a RAM-Disk which holds my texture cache. You can't get it faster. Sometimes, a new massive background set does not fit. Then I copy/save its texture folder and run Photoshop on the original folder, halving all texture resolutions. and again, if required. Only when I need the hires ones for the grand final poster size result, I a) swap texture folders and b) point the TextureCache folder to somewhere on HD. Will take forever, but so will the render (overnight job) so who cares.

just thinking with you, take your pick to suit your needs.

I thought Poser automatically re-used textures when and where it could? And by that I mean, if you have the exact same texture map on multiple objects then doesn't it only cache that once?

How do you move your texture cache to a different drive? I'm going to pick up some more memory today and configure part of it as a RAM drive.

Interesting side note - My work just gave away free outdated desktops to the staff. So I grabbed two small systems with an I5, 4 GBs of RAM, and a 125 GB SSD each. So I've configured these as two network render nodes.