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
What were you expecting? It has beens tated multiple times - I'm pretty sure some of them in reply to your posts - that the 50x0 cards cannot be made to work in DS 4.x.x.x. There is nothing that the daz developers can do about this. I an sure when we get the next major version, even as a beta, it will be prominently announced - DS 4.x.x.x is not going to surreptitiously morph into that next major version whatever happens.
We have several Workstations in the office and 2 of them have the NVIDIA RTX 6000 - 48GB-Ram - And other similar. Why I mention this is because these are the best and fastest graphic cards out there for artists, especially if you want to do quick renders and make faster progres with Daz3D and other similar apps. However, it also depends on your system build. We only use dual CPU systems totaling 56, 48 and 40 cores on three difference workstations, and have a good 51GGB Ram. Then again, some design artists have less needs than others so to each their own. Mostly all Nvidia RTX cards are awesome. However, the RTX 5090 is mostly made for gamers, although you can use a different driver for designers on the Nvidia driver site. The RTX 6000 - 48GB Ram card is specifically made for fast, high end design, and way beyond. It's a beast and it's not cheap! Lol! Awesome graphics card, too! Cheers! - J
<drool>
I have been testing various AI Weights/Models on my local PNY GeForce RTX 4070 12GB, AM4 with 128GB RAM, and 2TB nvme SSD and from that I can say:
1) I will need a RTX 5080 or RTX 5090 with 32GB RAM.
2) Need a new consumer oriented AM5 socket motherboard with 16 core/32 hyperthreads and 256GB RAM.
3) 8TB nvme SSD x2, the size of those checkpoint can be very big.
Take my advice and read Richard Haseltine's recent comments in this discussion.. Yesterday I was just about to press the button on a new machine (that I can ill afford) equipped with a RTX 5070 and then I decided to check the forums.
He says (quite clearly and many, many times) 'the 50x0 cards cannot be made to work in DS 4.x.x.x.'
That's good enough for me. It's up to you what you buy but I'd hate for someone else to buy something they also probably can ill afford, only to find that it's not fit for the intended purpose.
Oh, I saw that. This PC would be a 2nd PC. For AI & development that would be offline. DAZ and such would be online with my current 4070 card.
This means, I'd be stuck with my 3060 12GB.
There is no 40xx RTX with 16GB or more availeble anymore and the 50xx generation won't work with DAZ Studio.
I am not a gamer, I don't even really care for computers unless it is about 3d stuff and DAZ studio is my most used app.
DAZ will better get a replacement for IRAY, that works with any GPU.
One other option would be to improve the Blender Link.
Oh wait.... paywall. Looks like I'*m out of here soon.
See you in the blender forum
does Filament and Filatoon work with it?
not that I can afford those cards but both actually work on my other PC without ar dedicated graphics card
so DAZ might be forced to develop Filament integration more fully if they want to stay relevent
...same here. The Ampere "A" series can also still make use of NVLink for memory pooling Yes the A6000 is expensive even though it is last generation, but the A5000 and even A4500 can also take advantage of NVLink.
I've been looking at the 4000 and 4500 Ada cards The single slot 4000 has 20 GB 6,144 shading units, and is currently priced about where some of the more higher end GTX cards are (I've even seen 16 GB 4070 TI Supers going for as much and even more).
The 4500 is a dual slot card (about 2,200 USD in price) for 24 GB and a shader unit count of 7.680
4000s are price and 1.200 - 1.350 USD on average while the 4,500 averages around 2,200.USD
Both "sip" power compared to the higher end GTX cards (particularly the 4090) with the 4500 having a TDP of 230 W and the 4000,130 W (about half of what my trusty old Titan-X uses at peak) True they are not as powerful (particularly for gaming) as their GTX counterparts. but could be an alternative given that GPU prices are on the rise again even for older refurbished cards.
For about double the price of the 4500m the Ada 5000 offers 32 GB of VAM (like the 5090) with 12,800 shading units and consumes a maximum of 250 W (same as my Titan-X) This is probably as close to a 5090 one could get that can still be used with Daz True the price is not for the faint of heart. or wallet
Both "sip" power compared to the the higher end GTX counterparts with the 4500 having a TDP of 230 W and the 4000, 130 W (about half of wha my trusty old Titan-X uses at peak) True they are not as powerful (particularly for gaming) as their GTX counterparts, but could be an alernative given that GPU prices are for some reason on the rise again even for older refrubished cards.
To increase the "drool factor", Nvidia is in development of it's first pro grade GPU based on the new Blackwell architecture with their flagship RTX 6000 workstatio ncard which is expected to have 96 GB of GDDR7. Time to buy a Powerball ticket or two.
They should work, but whether they would greatly benefit I don't know.
Well, we have a couple of things in common. I’m not a gamer either and in spite of being a retired systems manager I no longer care much for computers either. My current PC is a 12 year old HP Envy with 16Gb memory and a NVIDIA GeForce 750 Ti with 2Gb. Needless to say it no longer passes muster. I can render a single clothed figure but I have to save the scene, shut down Daz and occasionally restart the PC and then open Daz and reload the scene to get a result. I also use Zbrush (kosher, pre annual fee version) which still plods along quite nicely.
I’m not sure you are right about the 40xx – I’m looking at this supplier as a source for a new machine
https://www.pcspecialist.co.uk/
and they list a
16GB Zotac GEForce RTX 4060 Twin Edge – HDMI, 3 *DP
and also a
16GB MSI GEForce RTX 4060 Ti Ventus 2* Black OC – HDMI, 3 *DP
That probably makes more sense to you than it does to me. There’s a bit of a lead time on these cards but nothing too bad.
I think Richard also says that you need at least 4 times more System RAM than you have GPU memory. That sounds like good advice too. If you haven’t already done so, I should take a look at some of the better considered opinions in these discussions. A lot of what I see posted here is fluff and white noise. some of which will quite likely lead some people to buy a not fit for purpose 50xx
I don’t know where you are in the world, if you’re not in UK it might not be much help and I don’t really know anything about the supplier so I can’t recommend them, but their website is worth a look.
That makes sense. Darn, I wish I could afford 2 machines. I'm looking at a 4060 in my next machine. Do you have any idea how that would compare to your 4070?
I would love to see this happen! I can't speak to the relevance issue, but I did request improved Filament integration as my closing request on the last survey.
The X870E motherboards (which support AMD AM5 sockets and the 7000 and 9000 series Ryzen CPUs) claim to support up to 256 GB RAM (4 x 64), but I expect that this would be very difficult to get to work in practice. DDR5 RAM is much more demanding on the CPU voltage controllers than DDR4 was, and so most of the boards will struggle to keep all 4 DIMMs running in sync to keep the RAM stable. I just upgraded my PC to an X870E board from MSI (X870E Gaming Carbon Pro) and I chose to run 96 GB DDR5 6400 RAM (2 x 48 GB) with a Ryzen 9 7950x CPU (16 cores/32 threads) and it is rock solid. I saw on the MSI message boards that someone mentioned that they got 192 GB DDR5 (4 x 48) to work with this board, but no mention about how they set the RAM up and what frequency/latency they are running with. The board I am using has 4 NVMe slots (2 Gen 5 and 2 Gen 4), but Gen 5 M.2 slot #2 shares PCI lanes with the PCI 5.0 slot for the GPU and so will depreciate the GPU from x16 to x8 if you populate it. I know there are 8 TB Gen 4 NVMe SSDs out there but prices look to be $600 USD and up (and don't know the quality of the manufacturers). I note that Samsung just announced a line of Gen 5 NVMe drives coming out this month (1, 2 and 4 TB drives will be available in March and 8 TB drive in 2nd half 2025), so something to look at.
I have one set up like that - but yea, you just have to run the RAM slower (it's 5200 I think running at 4800). But it's been pretty solid - if you want the speed though you're stuck at 2 sticks really. Yes, there are some who get 4 working, but it's a whole thing and who really has time for that these days.
If you want to go into AI stuff, my vote would be to go for a decent server/workstation board (TR if you want the PCIE lanes) and just load it up with DDR4. Much easier and predictable... and EEC too.
Are there no interim solutions through the use of a shim or wrapper for DS 4?
From Iray 2024.1.0 manual - The public API requires now C++ 11. All code paths for C++ 03 have been removed
The long wait for the next major update (for DS 5) with an unknown date, while more average users will buy 5xxx series cards and the lack of support, is clearly unsatisfactory.
It may be possible to use Iray Server, so a semi-external solution supported by a bridge from DS (in the Advanced tab of Render Settings).
I tried and it doesn't work with my 5090. The renderer in Iray Server 2024.1.0 falls back to cpu rendering. As Daz Studio 4 uses an older version of Iray, only a suitable version is used in Iray server.
For a DS release 5, it might be time to leave IRAY behind and start integrating another, better, more supported render engine.
My personal favorit would be Octane Render. Not as some additional Plug In, as it is now, but as part of DAZ Studio, as IRAY is now, with a huge library of shaders and materials.
Leave iray behind for octane? haha. nearly all my content for the last 10 years has been bought for iray and I am not alone. you would have a mutiny on your hands.
I doubt Octane would agree to give us a free version without watermarks, even if they were willing to agree to bundling in principle at a price Daz was willing to pay. And of coruse any switch of engine would require a massive amount of material conversion.
These! ^ This does not nullify the idea of better Filament integration, though.
Why waste money on a expensive video card when you can do something else with the money >> Like take a vacation? If you need an upgrade wait till it is not overpriced!!!! I have found that upgrades rarely are that good both software and hardware, LOL they sometimes break everything including your bank account.
Was worth trying.
AFAIK
IRAY isn't free either and someone could come up with shader converter.
Hwever thanks for the answers.
Omniverse support would seem a better option.
IMHO People need to prepare for the possibility that NVIDIA could easily EOL IRay.
Who ,besides, Daz is still using an old brute force Path tracer with no modern RTX support?
Reallusion Dropped it’s Iray plugin and has an exporter to omniverse where people with RTX cards can actually get the best use of them.
This is the problem with being dependent on third a party for a major feature of your program with no apparent option to buy it and or have permission continue to update yourself even if you “buy” it.
Hence the “Daz Studio” lipsync still being 32 bit only. requiring animators to buy third party options that eventually get broken by core DS updates ( DS5)
Hence aniMate 2 not fully functional with figures after G2 and ,again likely will be broken by core DS updates ( DS5).
Thanks guys, I'll keep this in mind as I move forward. I'll also hope that they'll upgrade the capabilities of the consumer boards because 256GB RAM, 2x8TB nvme SSDs, and a RTX 5090 is already crazy expensive for me.
As of Oct. 2024. Is deprecation same as EOL?
https://forums.developer.nvidia.com/t/rtx-accurate-iray-deprecated/310178
That message is just saying that Iray is being removed as one of the rendering engines from that 3D software (Omniverse Composer) not that Iray development has been abandoned by Nvidia.
https://docs.omniverse.nvidia.com/composer/latest/index.html
https://blog.irayrender.com/
https://www.nvidia.com/en-us/design-visualization/iray/
"The key difference from "deprecated" is that EoL is more final. While a deprecated feature might still work and have a replacement planned, EoL software is essentially abandoned by its creators, making continued use potentially risky due to a lack of security updates and support."
source: https://blog.invgate.com/deprecated-software
Iray is a useful tool for Nvidia and product marketing. The cost to maintain and develop it is Infinitesimal in their overall development costs for the corporation. It is something AMD can not share in. So what would be the business case for getting rid of it? Plus it allows them to experiment with render software/hardware development in an environment they completely control.
Well Iray was essentially a loss leader product to sell NVIDIA graphics cards
to the Arch/vis still render market back in the pre RTX days.
Now the FREE omniverse
(with it’s Full RTX support) fills that role
and offers realtime path tracing
for animation as NVIDIA no longer sells non RTX cards AFAIK.
Thus there is no incentive to update the old Iray for RTX support even if they don’t actually “get rid of” it because they want users buying RTX GPU’s to be using omniverse as the RTX GPU’s are best optimized for the many things that omniverse can do beyond render nice stills
I'm in the same boat. My 4090 suffered a terrible fate, and right now I'm just limping on a 1660Ti...which is back where I began my Daz journey. I'm not going to have the bones to get a newer video card right away...probably not till summer time or a little later. It would suck if there were no new 4090s left anymore, and Daz still didn't work with the 50-series. All those thousands of dollars spent on assets...
Who knows? Maybe this will spur the dev team to getting Daz5 out?