Daz 3D is part of 
Connect
DAZ Productions, Inc.
7533 S Center View Ct #4664
West Jordan, UT 84084
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2025 Daz Productions Inc. All Rights Reserved.
Comments
Ouch, that would be painfully annoying. Almost makes me wonder if the scene isn't fitting onto the 1030 and just dumping to the CPU render instead...
That being said, when not using the 1030 for rendering, how well does it work/handle daz studio for loading in characters, posing etc before rendering? If it handles that aspect decently, then you could still use it as your primary graphics display card (with all your moniters plugged into it) and leave the 1070ti as a secondary card (no monitors plugged into it) and select only the 1070ti + cpu for iray rendering.
With my 1070ti, its my only graphics card, and it also drives my 3 monitors, as well as display everything in daz studio's viewports. Windows10 takes a small ~500mb (1/2 gb) chunk out of that 8gb for itself, so i only have 7.5gb effective usable vieo ram for iray rendering... tempting me to want to pickup a lesser 10xx like a 1030 or 1050 series card just for desktop display & daz studio functions.
I do have a handy helpful tip here for everyone though -
Doing this saves daz studio from displaying all the geometry that gets loaded into your videocard ram, and therefor gives you that much more usable vram for iray to use.
This can mean the difference between the scene fitting onto the videocard or dumping to the cpu.
It's not quite as elegant as what KyotoKid mentions the Pros of 3Delight's render to RIB standalone Render option... But it does work.
You can check this for yourselves:
You should see a difference between rendering normally like you do vs the Wireframe Bounding Box method.
Other tips are to make sure all other running programs like webbrowsers are closed as they too take a chunk out of your total available vram.
I posted this some time ago in another thread but it seemed to go unoticed... I've got a few tips for increasing renderspeed as well posted that nobody noticed.
Is 3delight / standalone render to rib still capped at 8 cpu threads or less? I don't mind 3Delight, I find it still has its uses for a few things on occasion.
Also wanted to say thanks, your mention of Octane in your other "hd crash" thread really got me looking at Octane again especially Out of Core render option in 3.0 and enhanced & expanded in 4.0. Its pretty exiting, and if it can use / autoconvert iray shaders without too much fuss, it'll be pretty awesome. It'll mean serious competition for Iray, and it'll be interesting to see if Iray will adopt that method or not.
Thanks for the info regarding how a 1030 handles in studio putting scenes together before rendering. Might pick up a 1050 or something down the road for that purpose as I mentioned earlier regarding settin the 1070ti as a secondary card for rendering only.
Also take a good look at Octane render, it's "Out of Core" rendering mode lets your GPU access and use your system ram for rendering large scenes that won't fit on your card's vmem. Octane 3.0 right now supports Out of Core for textures, and 4.0 coming out soon also includes Geometry Data, and possible other features, possibly adding CPU support as well. I havn't tried Octane yet but its looking to be quite a game changer...
I do like iray and daz content is iray optimized out of the box, it's just annoying we're limited for GPU rendering to what fits in vmem. I've got my iry render settings tweaked so i can get a finished complete render in less than 10 minutes - that 1070ti is certainly a speedy powerful gpu. I really have to gripe when it dumps to CPU mode and it takes an hour or few to finish rendering and lose of AI Denoiser capability (GPU render feature only) ...
For now I will be using 1070 only as my current comp only supports 1 card. Once my new computer is up and running I will be using 2 like you suggest.
Honestly my PC may not be the ideal test for a 1030 setup. It only has 6 gig ram. That may be hindering studeo just as much or more than the graphics card.
...I know when rendering on the CPU, reducing the scene to wireframe in the viewport reduces the system memory load of having the scene open while rendering on the CPU (something I do as a standard process). I don't see how that affects VRAM usage as that still needs to crunch data for all the polys in the geometry as well as the textures with all their various maps and settings, otherwise all you would get a nice render of the wireframe.
Give it a try when you get around to rendering a few "doodles"
Tech PowerUp GPU-Z is a free utility, and popular along with CPU-Z, for hardware and system information, also shows you other information like GPU load, temps, fan speed, whether or not SLI is active/engaged, etc...
In daz studio, all that textures and geometry displayed in the viewport still take video card memory to display, no different than setting the viewport mode to Nvidia Interactive.
When you set to Wireframe Bounding Box, the only thing your video card is displaying is the daz studio program itself, as it it were a blank document / blank slate. All your characters props buildings (Geometry data), textures etc still get loaded into the System ram however, just not loaded into the Videocard ram. This is what and why you save on the video card ram.
The saved Daz Studio scene also loads faster as a side effect bonus. Again, its only loading into system memory, but not to the video card.
But it is VERY important to close daz studio to clear the video card ram and reload studio, viewport set to Wireframe Bounding Box and the scene saved in Wireframe Bounding Box for this to be effective.
In 1 scene I had brought right to the limits of my 1070ti 8gb of barely keeping on card memory to render. Just barely...
I added another character fully clothed and accesorized, hit render and sure enough it exceeded the card memory dumped to cpu rendering.
I set the viewport to wireframe bounding box mode, saved the scene, exited studio (clears the video card memory) , reloaded studio as i described, loaded in the scene
basically looking at a blank and empty daz studio viewport, aside from a bunch of white short right-angle lines or checkmarks, GPU-Z reported less memory usage. Much less.
I hit render... Watched the GPU-Z memory used bar... Sure enough, I was able to render the entire scene on the GPU.
So doing this is pretty much worth a free extra G8F character loaded into your scene.
:)
I second CoolBreeze wich regards to closing and reopeng daz. I never thought about switching to wireframe. But when I restart daz it does sometimes render faster, especialy when its a second or third render.
I also find Out of Touches Genesis 8 hair, while prabably the best arround, is be a big time culprite. It tends to double my render times, prabably because its a memory hog compared to other hairs. I find other hair venders like Kool, Swam, Goldtassel, and AprilYSH to be much more render friendly. That being said some of my favorite hairs are OOT hairs and I dont let the render times stop me from using them when the situation calls for it.
...explained that way makes sense. However if you use a dedicated GPU for rendering (either using the MB graphics or a second GPU for the display) it shouldn't be an issue then unless the process dumps from VRAM to the CPU.
The restart is also a bother. We need a script of some sort that lets us manually clear the VRAM buffers after running a render process. A big scene can take long to clear from system memory and load back in. (I've had to wait upwards of ten minutes or more for Daz and a large scene to clear from the process queue in Task Manager before restarting [old system] and just about as long to reload the scene after restarting - that seriously impacts workflow if you are running multiple render tests).
Really wish we could batch render in Iray like Lux and Octane do so that Daz and the scene can be closed down during the process to preserve system resources.
Fair point. I wasn't sure if you had 1 or 2 cards, one as a dedicated rendering card... I don't have 2 video cards, just a single 1070ti to shoulder all the load of display duty and rendering. With GPU-Z you would be able to see whether or not there's any vmem load on your dedicated rendering card, as GPU-Z detects lets you select which card to display the Sensor info on.
Still, this is applicable in 2 scenarios:
At the very least saving, closing and re-loading Daz Studio after a while and before rendering a scene does help clear both vmem and system memory of any left-over / residual geometry & texture data from adding / changing / removing content and shader presets during the course of building your render scene.
I'll second you on the point regarding the batch rendering or lack tthereof. Lux render as a batch rendr also allowed for further saving / pausing / resume of renders, and a tool to "paint prioritize" areas needing noise removal. Carrara also has a batch render function but no iray (and no support for g3 / g8), but it does have a lux plugin. I'm waiting to see how well Octane 4.0 Beta release goes.
Instead of using two different apps to track GPU and CPU, MSI Afterburner can do all of that from one screen. You can also map a key to toggle a HUD that pops in some windows. It does work in the Daz window.
If you are truly rendering any sort of decent size scene, then almost certainly that 1030 is sitting there doing nothing most of the time. With only 2GB that fills up very quick. A single G3 or G8 character can hit that cap with all the textures hair and outfits have these days. Also, even when a 1030 runs, there actually CPUs that can render much faster than the 1030, LOL. That is seriously the bottom of the bottom tier GPU.
There is not an Iray benchmark for the 1030, so you can totally try doing if you like to compare. The Sickleyield scene benchmark is in my sig. Download her scene from the first post, simply load it up and render. We try to tell people to use the Iray preview mode before rendering. I also made my own bench scene...but do not bother, you will never finish it with a 1030. It might time out at 3 hours (the time limit.) To test your CPU you can uncheck the 1030 and render SY's scene on CPU. Then you will be able to directly compare if the 1030 is helping much when it runs.
The reason for the bench is to see where you stack up right now. You can look at how long your 1030 takes and look at what other people are getting on their machines. With that information, you can estimate how long your typical scenes would render on other GPUs. So you see somebody has a GPU and renders the bench 10 times faster (which is being conservative looking at a 1030) then you can estimate a scene you have done before to be 10 times faster. Iray does scale pretty well in this regard. This might help with your buying decision.
If you want to render OOT hair faster, the easiest way is to compress the textures. What I do is make a backup folder in the location where the original textures are, and make copies of all the textures for the hair in that folder. Then I take a tool called Image Resizer for Windows.
http://www.bricelam.net/ImageResizer/
Image Resizer is awesome because it adds a right click option to any image you ca right click on in your folders. You can also select multiple images, which is why this tool is such a huge time saver. You can select the size and jpg texture quality. You can take OOT rextures down to 3000 by 3000 pixels, and reduce the jpg quality to the low to mid 90's. You already made a back up, so you can check the box to over write the original images. (Or not, the choice is yours.) If you it do it this way, then the hair will load with the new optimized textures automatically. If you choose to keep the original textures, then you will have to manually add the new textures in from the surfaces tab. If you feel you lose too much quality, you have the original textured backed up and you can try a different setting. I have even taken some textures to 2000 pixels with new real impact to the render made.
You can use this on more than just hair! Any texture can be resized, it is just a matter of how much you want to resize it. I have noticed that some characters these days have massive 95 MB specularity textures as tif files. I convert those to jpg and reduce the quality. Without resizing I can still get that file down to 9 MB, a full 10 times reduction! And I can't see any difference in the render.
Doing this will not only make renders faster, but save a ton of VRAM. Like I said earlier, even when a scene already fits your GPU VRAM, optimizing it can make a big difference in render times. Anything that is far from the camera can be compressed.
Iray does do some texture compression, but it could be better.
...Carrara's render engine on it's own is pretty solid for a CPU/Biased ray trace one. The nice part is you can also network render so instead of just 8 threads and 24 GB I can have 20 threads and 56 GB of memory at my disposal (Carrara can support up to 100 CPU threads). There is also an Octane Render plugin available for Carrara as well.