Daz Studio Pro BETA - version 4.11.0.335! (*UPDATED*)
This discussion has been closed.
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
... Titan V user is back again ...
- I downloaded and installed the latest Cuda (9.2) and the Optix SDK (5.1.0) and run the denoiser program that comes with optix sdk. It works perfectly fine. But in Daz3D, either whole picture is greyed out or "blockized" (check my previous posts here with pictures). I followed some suggestions here from the users by setting denoiser kick-in at higher iteration, but those didn't work neither. Problem still persist. My question is, are there any other "titan v" users (not xp or others) with latest driver (or specific driver) that can work with denoiser?
- Rendering is (not feels like as fast as it should be... but) fine. I have an interesting finding though; in the past with 4.10, with my 980 gtx, enabling optix prime was increasing performance by %12 to %30 for me. Now, even if I leave titan v out and only run the 980 gtx with optix, nothing is gained. Actually it adds 1-3 seconds in total for 10min of rendering! When I test this with titan v or both cards enabled, results are similar, nothing is gained through optix prime.
- I'm heavily exploiting the usage of the shearing effect in dforce. An example would be a little object A passing through big object B, causing object B to tear apart and create a hole on it. The result is the "see the other side of the cloth". Please, for future dforce development, consider this as a use case. I can already use this to create knife/bullet effects on clothes and even character skins (but it requires aton of time experimenting damping/bending/simulation values).
- I'm also trying to accomplish same tearing with the solid objects like tables/chairs and basic primitives. I'm close to get this done but the desired effect is a very thin line between exploding simulation and simply not having the desired effect.
- The result is nearly everything in the scene is simulated dynamically. And nearly everything collides with each other. So, I need to be able to use all cards and cpu for dforce simulation. These "tearing" effects doesn't happen with default simulation settings and they need to be set (sometimes most of the time) much higher.
- I have a use case scenario where I need "keyframe" for dforce simulation. During my animation, if I add more frames, I am forced to simulate all necessary items from the first frame of the scene. Is there a possibility to "save" or create a dforce keyframe for what is simulated until certain frame number and then continue to the animation using "custom frame between X-Y frames" under simulation settings? The best option would be to add a "do not clear simulation" option somewhere for each object on the scene when "Simulate" is pressed.
- All my dforce usage scenarios are animations and not stills.
.daz format props that open fine in DS 4.10.0.123 will not open at all in DS 4.11.0.171. An error message is displayed and the log file indicates an access violation. Some examples are products Lisa's Botanicals - Potted Palm, SKU 9788, and Lisa's Botanicals - Carnation, SKU 8766. Both products had Daz Connect updates today, but neither product is working in DS 4.11.0.171.
Log file excerpt:
2018-08-10 15:10:25.991 Opening file pottedpalm07.daz...
2018-08-10 15:10:26.080 WARNING: QObject::connect: Cannot connect (null)::currentValueChanged() to DzSkeleton::handleDrawGLBonesChanged()
2018-08-10 15:10:26.080 Error reading file N:/DAZ 3D/Studio/Daz Connect Data/data/cloud/1_9788/props/lisa'sbotanicals/pottedpalm/pottedpalm07.daz : C Exception: Memory Access Violation - Attempted to read memory at address 0x00000000
2018-08-10 15:10:26.080 File read time: 0 min 0.0 sec.
I will submit a help request.
It might not be a beta glitch or a Connect glitch — I've had problems before with Lisa's Botanicals .daz files in earlier D|S versions. IIRC they'd load but the textures would never show. The fix I found was to load the Poser versions of the props and save in D|S format. Remember why DAZ changed to the .duf file format; the earlier format turned out to be only almost 100% stable, and every now and then it went cuckoo like a Swiss clock when you load it into a D|S version that uses .duf files.
I have had a couple of products fail to load in the beta as .daz files that were working in 4.10.0.123 - as Spotted Kitty says, in most cases loading the Poser files and applying the DS materials works though i've still reported them to Daz via a support ticket and they were updated.
It fails to render with it, so it is "removed"... Thus the conflict of notation, as I stated. Log excerpt below. "architectural sampler unavailable, caustic sampler disabled", both are selected in the render settings as ON
Also this...
followed by this...
You can see that only one card is being used for POST rendering of the "Denoiser", for some reason it is displayed twice, once with 8.5MB/487.2MB then again with only 4.4MB/249.2MB
That is after 200 iterations, where POST denoiser is setup to run. It is using CUDA device #1 and no other devices... But you see all three CUDA devices 0,1,2 are all rendering the rest.
The POST denoiser lines are never mentioned again, anywhere in the logs, but I see it working up to the end, or it is just using that single/double image as an "overlay", from that moment, as an underlay until corrected pixel-info appears. (It should be denoising, progressively, as each new X-iteration resolves, not just once or twice, in the beginning.)
Starting after 8 iterations is just bad... There is NOTHING resolved after 8 iterations, except the initial image-textures. You actually need some light resolved, before it can "guess" what the rest of the lighting will be. (Which is what it is trying to do, with the denoiser. It is guessing the "resolve" of the shades, of the unresolved pixels that light hasn't hit yet. But doing that only once, at iteration 8 is not good. There is not enough light resolved at that point.) Especially through "reflections and refractions", which takes 9+ to 250 iterations, before a "ray" has passed-through. (Like the eyes)
Are projection morphs broken in DS 4.11.0.171?Today at 22:35
Big Girl Morphs for Genesis 8 Female by SickleYield and KindredArts contains BGM Heavy Clothing Smoother morph and BGM Heavy Clothing Smoother Skirts morph. I believe these are projection morphs that should affect the clothing and not the figure, but I don't see any change in the clothing when they are dialed in in DS 4.11.0.171. In DS 4.10.0.123 I can see them move the clothing. Are projection morphs broken in DS 4.11.0.171?
Am I the only person who is getting better rendering times but suffering horrific performance in the software itself when setting up the scene? in 4.10 I could happily navigate my scenes and move items, sometimes a little bit of lag but nothing major. Now, my render times are significantly faster but i could have a cup of tea whilst waiting for the camera to move when trying to navigate simple scenes.
... Update from the Titan V guy ...
About my denoiser issue:
- Under the Nvidia Iray advanced render settings, if I uncheck my Titan V card and only leave out 980 GTX, denoiser works.
- If I only check Titan V, denoiser corrupts the image. If I select two of my cards, it chooses Titan V for deniser and it will corrupt the image. (What is the purpose of Titan V if I can't use the denoiser BTW!?)
- I need to be able to select which card to use for denoiser. But this time I'll be limited to GTX 980's 3GB of memory. But at least, it will be something.
Is anyone else seeing strange texture seam areas on human figures? I am experimenting with skin settings from over in the Fiddling thread, and with Genesis 2 and Genesis 8, the texture seams are visible.
hmmm.... evenly a freshly loaded Floyd shows evidence of this.
/edit and now I can not reproduce it. Perhaps just a glitch...
Someone add Turing support slaving over
Quadro RTX 8000
I'm confused about this new Invert Transmission Normal property. The default is on, but I've found that normal maps only look correct on skin surfaces (using translucency and Thin Walled off) with this property turned off. What exactly was the issue before with normals being inverted on certain surfaces?
(Edit: Never mind about the first part, the normal map generator I was using was inverting the red values for some reason, that's why the maps were messed up)
Is there any word on a fix for the distorted legs issue? Might not be the biggest of their problems but it does leave Daz pretty much unusable for anyone doing full body renders
Fixed for the next update. In the meantime, enable limits for poses.
I noticed that current OptiX implementation has a lot of bugs and causing crashes when more than one card is selected. It also feels somehow slow. Is it possible for dev team to update optix to the 5.1.0 for the next beta? I suspect that it is the culprit that prevents AI denoiser from working correctly for Titan V, since in Optix SDK 5.1.0, I can run the denoiser just fine.
Do I need to have "regular" Daz 4.10 and "regular" CMS installed in order to use the smart content in the beta version of DS 4.11?
Shouldn't be needed; both the beta and release program versions ought to use the same database program, database settings, and content settings. The only thing that might need re-installing is any extra plugins you've bought (if this is the first time you've installed a D|S beta). If you uninstall then reinstall these plugins, that will copy the necessary files to both the beta and release program folders.
I wholeheartedly concur with keyframing dforce simulations. This would be real helpful for animators. I would add keyframing textures and shaders to the mix. Adding non-linear animation as AniMate2 has would also solve the issue.
The G8 distortion leg issue is my biggest issue with this version. Using limits - restrains animation - where sometimes you have to exaggerate poses or break joints to get the right animation effect.
- In addition, also dforce values and some morph values that don't get recorded and official de-parenting/parenting support for keyframe data. For example, if something gets wet during an animation, I really like to change dforce values between keyframes.
- Also, nearly everything in the dforce simulation settings should be private to the objects. For example, gravity and air resistance should be configurable for each object seperately. That way I can have my solid puffed up sofa (With negative gravity, no air resist) being properly animated and bent while a character (+1 simulated gravity) sits on it. I also think de-coupling these global settings from a single page will also help parallelize the cuda calculations in their own way, so devs won't have to limit to only single cuda device for dforce sim.
... and the current version of Install Manager makes this even easier; it provides Re-Install actions in the context/option menu for a given package on the Installed page. Check the boxes for the packages you wish to reinstall, then right-click and choose the Re-Install action. For packages that install like software or plugins, it will uninstall the package an immediately queue it for install... using whatever paths you have configured for each type. For packages that install like content it is slightly different in that it will use whatever target path you have selected for the destination (so that it can be used to mass move installations of content).
How would I find out if they got my bug report on the inability to use Dforce in the new version. I never got any kind of notification that it went through or the usual we will take a look at this email thing. I just need to know if I should resubmit or not.
Did a bunch of testing with the Deep Learning Denoiser / Post Denoiser feature. Found some very interesting and happy results in terms of render speed gains with it on vs leaving it off...
I wanted to test Post Denoiser myself to understand this feature better.
After all, we've all had (and likely frustratingly long ) renders only where unfortunately there was some bright specs or patches of pixel noise or grain. While often its user error that contributes to the render noise from placement of characters and light source(s) or lack thereof; sometimes its entirely intentional that we want details / characters cast in partial or full shadow, if only it weren't for render noise and grain.
For this series of tests and the sake of fairness I did only GPU + OptiX , leaving the CPU out of the equation.
My card: 1070ti 8gb , nothing special, none of the super clocked editions.
My scene setup starts with Urban Future 5, G3F default shape & texture set, Barbaric Hair for G3F, and Austrani clothing set - both of which use iray shaders by default. I choose a random walking pose, as well as selected a Sun&Sky Iray engine preset from one of the vendor's products and adjusted the dome's rotation so sunlight will shine down the street left to right. I place and rotated G3F so she's in some sunlight while the choice of hair hides her face in shadow. Good recipe for lots of render noise and grain.
My testing factors in a few different values:
Denoiser with GPU + OptiX
No Denoiser, GPU + OptiX only
Convergence % Low / High / Medium / Quarter
The idea here is start with a low convergence value (speediest and likely worst quality), default convergence value of 95%, and 2 mid-ground convergence ratios of 50% and again half that = 25%
The order of Testing:
Denoiser with Convergence set to 15%
Denoiser with Convergence set to 95%
Denoiser with Convergence set to 50%
Denoiser with Convergence set to 25%
no Denoiser with Convergence set to 15%
no Denoiser with Convergence set to 95%
no Denoiser with Convergence set to 50%
no Denoiser with Convergence set to 25%
Batch 1: G3F Body / Longshot with Denoiser
My method looks at the first batch starting with a very low Convergence ratio paired with Denoiser, then regular 95% default Convergence, then 2 middle ground values in between. Denoiser is set to start at Iteration 8, and upon inializing I make a screen capture showing render quality at Denoiser "Start" stage. Once at 100% render "Finish" completion, another screen capture is taken sowing the time and iterations. Also is a saved copy of the render for viewing the entire final output of 2560x1440
Batch 2: G3F Body / Longshot no Denoiser
Then for baseline comparisons, the tests are repeated at matcing Convergence values but without Denoiser, in the same respective order. The same steps are repeated except the "Start" stage since picture contains noise and grain throughout the render up until "Finish".
Batch 3: G3F Face Closseup / with Denoiser
The next batch of tests is the same scene, however the camera is a face closeup to examine finer details, and where likely is the chance for more concentrated noise and grain since majority of the G3F's face is being shaded by her long hair.
Batch 4: G3F Face Closseup / no Denoiser
Self explanatory.
G3F Stock texture, Barbaric hair, Austrani clothing, Urban Future 5
Max Path Length Path -1, no camera dof, no tone mapping
Batch 1: G3F Camera 1 Body Shot with Post Denoiser, Landscape: 2560x1440
1. denoiser gpu+opti 15% converge 10 iterations 35 seconds start
2. denoiser gpu+opti 15% converge 96 iterations 2 mins 52 seconds finish
3. finished render
4. denoiser gpu+opti 95% converge 10 iterations 38 seconds start
5. denoiser gpu+opti 95% converge 1592 iterations 39 minutes 37 seconds finish
6. finished render
7. denoiser gpu+opti 50% converge 10 iterations 34 seconds start
8. denoiser gpu+opti 50% converge 310 iterations 8 minutes 7 seconds finish
9. finished render
10. denoiser gpu+opti 25% converge 14 iterations 35 seconds start
11. denoiser gpu+opti 25% converge 132 iterations 3 minutes 43 seconds finish
12. finished render
Batch 2: No Denoiser, Landscape: 2560 x 1440
13. no denoiser gpu+opti 15% converge 95 iterations 2 mins 40 seconds finish
14. finished render
15. no denoiser gpu+opti 95% converge 1613 iterations 39 mins 41 seconds finish
16. finished render
17. no denoiser gpu+opti 50% converge 304 iterations 3 mins 30 seconds finish
18. finished render
19. no denoiser gpu+opti 25% converge 132 iterations 3 mins 30 seconds finish
20. finished render
Batch 3: G3F Camera 2 Closeup with Post Denoiser, Portrait: 1440 x 2560
21. denoiser gpu+opti 15% converge 9 iterations 59 seconds start
22. denoiser gpu+opti 15% converge 206 iterations 13 minutes 31 seconds finish
23. finished render
24. denoiser gpu+opti 95% converge 9 iterations 57 seconds start
25. denoiser gpu+opti 95% converge 1765 iterations 1 hour 52 minutes 30 seconds ** @ 95% completion
26. finished render
27. denoiser gpu+opti 50% converge 9 iterations 56 seconds start
28. denoiser gpu+opti 50% converge 453 iterations 29 minutes 5 seconds finish
29. finished render ** @ 95% completion
**
30. denoiser gpu+opti 25% converge 9 iterations 58 seconds start
31. denoiser gpu+opti 25% converge 273 iterations 17 minutes 26 seconds finish
32. finished render
Batch 4: No Denoiser, Portrait: 1440 x 2560
33. no denoiser gpu+opti 15% converge 205 iterations 12 mins 43 seconds finish
34. finished render
35. no denoiser gpu+opti 95% converge 1207 iterations 1 hr 23 mins 39 seconds finish
35a ** re-ran test, 59 minutes 15 seconds 945 itterations for 82% completion
36. finished render at 82% complete
36a ** re-ran test, 59 minutes 15 seconds 945 itterations for 82% completion
** Same issue as test #29, slows to a crawl, 5gb of 8gb video card memory used,
37. no denoiser gpu+opti 50% converge 445 iterations 28 mins 50 seconds finish
38. finished render
39. no denoiser gpu+opti 25% converge 271 iterations 17 mins 1 seconds finish
40. finished render
Next batch: The Closeup:
Same scene but lets get a closeup of her face...
File names corelate to the test numbers ie: test 22 = Untitled-22.jpg (screenshot), finished render image = 23.jpg
Notice anything? Key note here is Convergence. With Denoiser you can set it low and get a render finished in a fraction of the time vs default 95% convergence and no Denoiser ...
Continuing tests with same scene, however swapping out G3F for G38 and obviously different hair style and clothing...
50% convergence ratio:
25% convergence ratio:
Next series of test posts after this will be with G8F... Wanted to get G3F done first.
Hello,
Affects: DAZ Studio 4.10 and 4.11 BETA (and 4.10 Release)
Windows 10 64-bit
NVidia GTX 1080 Founders Edition Graphics Cards
In both versions 4.10 and 4.11, sometimes the iRay viewport preview will get stuck "loading" and lock up the application. It will not start rendering, and there is no way to cancel or return to the previous viewport mode (and of course, any work not saved is lost). The "progress bar" near the bottom of the window will read 0%. Task Manager will show about 25-27% utilization for Daz Studio Application. There is no way to generate a report.
Tried to submit a ticket via Help Desk, no positive feedback from clicking Submit so... not sure if it went in or not.
Also, when the hell will there be an autosave option? DAZ Studio should always save a backup before starting any render, and every 5-10 minutes during an idle period. DAZ should also have an option to create a backup copy of a any file when saving (in case the user saves over the wrong file).
Thank you.
Which driver version are you using? Have you tried a different one?
You should get a confirmation email after submitting a ticket.
mein auch. es ist ein Käfer beim DS 4.11.x dass soll in näschtes Version 4.11.x geflickt werden.
Dforce still crashes Studio on Mac.
I just installed the latest beta (- version 4.11.0.196) next to my general release 4.10 Pro. How do I also access my content library in the beta? I am not talking about Smart Content here.