What are the minimum system requirements??

in New Users
Hello, I recently downloaded Daz Studio pro 4.9+ (64-bit), all packages installed except for 3D bridge for photoshop (due to some error about the path). But what i would really like to know is; what are the minimum system requirements to run Daz Studio 4.9 as smoothly as possible?? What would I need to have the best possible expirience with this 3D software?? Thank you in advance.
Comments
https://helpdaz.zendesk.com/hc/en-us/articles/207530513-System-Recommendations-for-DAZ-Studio-4-
The Photoshop Bridge will not work with Photoshop Elemental, and you have to set the path to your Photoshop software in DIM.
The question is, are you just testijng this out as a new hobby, or are you intending to continue rendering? Because, the answer vastly differs depending on if you just are trying this out,. or if you want to do serious rendering.
If you are into serious rendering, check out the system recommendations above. You're looking at a system around $3000+.
If you are trying out... 8 GB is the absolut minimum for RAM. You will be able to render a single person in Iray, if you do not have a Nvidea graphic card.16GB is better, but will still give you crashes when you have more than two people in it. 3DL and Iray both like multicore fast processors. If you want to invest money into a better graphic card, the minimum would be something like a GTX970 (4GB). As all textures have to fit into the graphic card's VRam, that means that you are not be able to use if you load more than two people with that card. Larger scenes need bigger cards, or multipass renders...
I would have to disagree with some of that, I have 16 GB of ram and no nvidia card and I can render 3 sometimes 4 figures with out crashes. If you don't have a Nvidia card you can still render, it just takes a lot longer
Well, you can render 3-4 figures, but once you add HD and subdivision into the game, and dress the figures into something that has large maps, you really quickly reach the limits in Iray. Like, when I have 7-8GB before render, I can be almost 100% certain that the system will crash. I'm using Simtenero's Iray memory calculator, finding it very accurate for the purpose.
Interesting thread. How can a person tell how big their scene is? I have 8GB RAM (16GB more arriving today) and a GTX 960 4GB and some renders turn out fine, others not so much. Some render fast, others...not so much. So is there a way for me to check what my scene is and edit accordingly until I am not overloading my GPU? I always Select ALL and give everything a whack with the Uber Base Shader but not everything in my scene is in my camera. Should I limit that to only the objects that are going to be in my render? I figured that my GPU would only be working on developing the things inside my camera view but maybe I am taxing it unnecessarily.
I'm using this product: http://www.daz3d.com/iray-memory-assistant
It's also pretty useful in that you can use it to "switch off" certain parts, and it tells you how many normal maps each item uses. Norm al maps are extra heavy on memory, so the first thing to do would be stripping mid-to distant characters/items of normal maps (displacement maps can go, too). It also gives you pretty good usage numbers of your VRam, so you'll know if a scene will fit or not (works in 95% of all cases).
But what also works fairly well is, simply open your Task Manager. See how much memory you are using before you load your scene. Then, load your scene. The memory usage will go up. Then, do the maths. Like, I have 2.1GB usage before loading, and 8.3GB usage after loading my complete scene. that makes 6.2GB for the scene. Add that to the 8.3GB = 14.5GB. In my system, 14.3GB is usually the treshhold where my 16GB RAM PC will crash, so I would have to toss out maps and characters (basically making several versions of the scene and put everything together in Photoshop). Keep in mind that if you use lots of Subdivision, memory usage also will increase, even if you haven't reached your treshhold.
EDIT: Also, it's worth saving the scene, close DS and then open the scene again. Especially when you are setting up scenes, DS doesn't always "clean" the memory properly, so you'll have "ghosts" in the system.
EDIT 2: About the GPU - it would need information of all the items in the scene, to calculate raytraces correctly. For example, if you have a mirror outside of your carmera, which reflects some light on your character, the GPU will need the mirror's information to correctly calculate that light. I'm not sure about items which are not "visible" in terms of being blocked from the light paths.
Thanks Bee! I just want to make sure I'm understanding the math correctly.
X= Current Memory Usage Prior to scene. <2.1GB>
Y=Memory Usage with prior usage + Scene <8.3GB>
X-Y= Scene Memory Usage (Z) <6.2GB>
Then you Z+Y? Isn't that counting the scene memory twice? Y includes Scene Memory and Z IS Scene Memory. I'm not arguing. I am learing all of this computer performance stuff as I go and just want to make sure I'm understanding. Thanks again!
Yes, it is basically loaded twice into the RAM if you render in CPU mode. The scene is prepared for upload to your nvidea card, and if it doesn't fit into it (or you don't have one), the RAM pretends to be the VRam, and the prepared scene is 'uploaded' there instead.
Okay so since I have a 4GB NVIDIA and 24GB of CPU RAM, I have 28GB to work with? Or say my scene is 4.2GB does the other .2GB go to CPU or does it say 'Oh that's too much, CPU take it all from here!" ?
No, you can fit up to 4GB on your card's VRAM, but the RAM and VRAM can not combine. It is either RAM or VRAM. So. If your scene is more than 4GB (or what ever GB is left from screen, etc), the video card will bow out of the rendering, and you will be back to CPU render.
This means that a fast nVidea card like GTX 780 is only good for renders where everything fits into its VRam. Beyond that, your renders will slow down, because you no longer render GPU, but CPU.
In addition, having several video cards also do not combine thei VRAM, but the render data has to be loaded into each card separately. So 4GB and 4GB VRAM are still 4GB. I recommend you read the link I posted above for system requirements, it is explained better there.
*sigh
Thanks Bee, I think I get what you're saying there. That other link ended up explaining it too but then confused me about other things. I'm done for today lol. But thanks for your responses and time.
I loaded The Franklin Loft (figured it's medium-ish) 2 characters, 1 rug, 3 primitive planes for outside the windows and a 3pt light from Architectural Lighting Rig for Iray as well as another room light rig.
I have a i7 3.3 GHz processor, 24GB of RAM and a GTX960 4GB
X: Computer on= 2.2 GB Memory being used
Y: Scene loaded= 4.2 GB
Z: Scene Usage= 2.0 GB
Y+Z= 6.2 GB
Do I really need 8+GB GPU to render this scene on the GPU? It's two characters, wearing clothes. No candles, art, props, dragons, appliances all on at the same time. That link Bee posted indicated that 2 GPU's can work in lieu of a more expensive single card but doesn't increase GPU RAM, just combined CUDA Cores. But I also have been told that SLI doesn't work in Iray. So does one just sit there while the other one works or does it contribute it's CUDA cores to the one that is working?
What am I doing wrong? I don't get it! It's like when your toddler cries because he's hungry but then refuses what you give him. "I'm trying! Speak why won't you speak?!!" lol
Thank you all so much.
I must admit I have never tried to work out GPU memory usage in this way as there are to many variables - Textures - Sub D etc but what you describe may well go over 4GB especially if some of your GPU memory is used on your operating system and if it does exceed 4GB it won't use your card for this render
I use GPU Z to be sure the scene is fitting in & using the GPU - The negative side of that is you don't know until you try to render it
https://www.techpowerup.com/gpuz/
As you noted 2 GPU cards of 4GB will not increase the amount of GPU ram you have available for rendering you would just have more cuda cores so any scene that fits into 4GB will render faster - Anything greater than 4GB will fall back to CPU and the cards won't be used at all.
The way I understand it is that iray will try to load the full scene into all available cards & if it doesn't fit into one of them then that card & its cuda cores will not be used at all in that render - If it falls back to CPU no cuda cores will be used at all.
The "Add up" only happens when you are not using GPU, so if you are using GPU, the first calculation would be
X: Computer on= 2.2 GB Memory being used
Y: Scene loaded= 4.2 GB
Z: Scene Usage= 2.0 GB
Z would go to the GPU, in addition with whatever extra memory is needed for Subdivision. Only if Z doesn't fit into the GPU's VRAM, it would be send back to the RAM. But it's not a precise calculation. Here's an example of where tha actual VRAm usage is higher than what the plain Task Manager maths suggest:
My scene has a scantily clad character with hair in a very nice, Iray ready Courtyard scene. The maths go:
X: Computer on, DAZ STudio open= 2.7 GB Memory being used
Y: Scene loaded= 3.9 GBZ:
Scene Usage= 1.2 GB
My GTX660 has 2GB of VRAM, so I should be fine, right? Unfortunately, that's not the case. If I run Simtenero's tool, I come up with additional VRAM needed for Subdivision and the way the maps are handled, so instead of 1.2 GB, I am looking at 2.2 GB being send to my poor GPU, which will wave the white flag and send everything back to my CPU's RAM. The CPU's RAM will see 4.8 GB increase (instead of the 2.2 GB that go to the GPU, because of handling, etc), so once my GPU bows out, my calculation is:
X: Computer on, DAZ STudio open= 2.7 GB Memory being used
Y: Scene loaded= 3.9 GB
Z: Scene sent back from GPU (NO-GPU-rendering!): 4.8 GB
X+Z= 8.7GB
Note that in this scenario, your 960 would still be fine and happily chugging away, as the 2.2 GB would still fit into your VRAM, if you don't have too much other load on it.
EDIT: Here's an example of a scene that would actually fit into my card:
Don't forget that if your card is the display adapter soem of its RAM (qyite a bit, in fact) will be used for the desktop and open windows so you will have less than the total RAM avaialble for rendering. I'm currently using 211MB of my 4GB for Thunderbird, Skype and Firefox (plus the desktop). Opening DS and loading a naked and bald G3F for my clothes-testing scene bumps that up to 364MB.