[TEST] Iray render fails in 4.24 but succeeds in daz beta 2026
ruin_seller
Posts: 0
[[TEST] Iray render fails in 4.24 but succeeds in daz beta 2026]
Introduction / Test purpose:
I performed a controlled comparison between DAZ Studio 4.24 and DAZ Studio 2026 (version 6.25.2026.6423 - 64bit public build) to evaluate Iray behavior in a borderline scene scenario.
The goal of this test was to verify whether both versions behave consistently when rendering the exact same scene under identical conditions, and to identify potential differences in memory handling or render initialization.
I am posting this report in the hope that it can reach the DAZ developers and be useful in some way, since the difference in behavior between 4.24 and daz 2026 was consistent and reproducible in my tests.
I used HwInfo for monitoring during the test; unfortunately, I cannot attach a .csv file to this post due to the site’s restrictions. However, I am attaching relevant data relating to the tests, via text.
--------------------------------------------------------------------------------------------------------------------------------------------
Test conditions (controlled variables):
The following parameters were kept strictly identical between both versions:
- Same scene (.duf)
- Same camera
- Same render settings
- Same resolution: 1536 × 864
- Same Iray configuration:
- Max samples: 1000
- Render Quality: OFF
- Denoiser: OFF
- Same hardware: GPU: RTX 3070 Ti (8GB)
- Driver: 591.74
- Scene configuration: Heavy environment / textures/lights + 3 Genesis 9 characters (with clothes, hairs dforce strand based)
No changes were made between tests - between tests, the software was closed and the VRAM cleared.
---------------------------------------------------------------------------------------------------------------------------------------------
Results:
DAZ Studio 4.24
Result: FAIL (2/2 attempts)
Render does not properly start
- Observed hardware behavior during failure:
- GPU memory allocated max: ~7132 MB
- Dedicated D3D GPU memory max: ~6963 MB
- GPU memory usage max: ~87.1%
- Minimum available GPU memory: ~1060 MB
- However, despite using a large amount of VRAM, the GPU never enters a true render workload:
- GPU core load avg: ~7.1%
- GPU core load max: ~29%
- GPU power avg: ~45.7 W
- GPU power max: ~111.9 W
In practice, DAZ Studio 4.24 appears to fail before reaching a normal full render phase.
DAZ Studio 2026 (Public Build / Beta)
Result: SUCCESS (2/2 attempts)
Render starts correctly and completes successfully
Visual result is correct and consistent
- Observed hardware behavior during successful render:
- GPU memory allocated avg/max: ~7634 / 7647 MB
- Dedicated D3D GPU memory avg/max: ~7465 / 7477 MB
- GPU memory usage avg/max: ~93.2 / 93.3%
- Minimum available GPU memory: ~545 MB
Unlike 4.24, DAZ 2026 clearly enters a normal render state:
GPU core load avg/max: ~91.8 / 99%
GPU power avg/max: ~224.8 / 239.4 W
In short: 2026 renders the scene normally, while 4.24 does not
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Additional notes:
This was a controlled comparison: same .duf / same scene / same camera / same render settings / same hardware / same optimization level
The behavior was reproducible:
- 4.24 failed twice
- 2026 succeeded twice
The most important point is that 4.24 does not seem to fail because of a simple total VRAM overflow.
In the failed 4.24 runs, there was still roughly 1 GB of available GPU memory left.
The more relevant difference is that:
4.24 never transitions into a proper high-load render phase, 2026 does.
This suggests the difference may be related to:
- scene/render initialization,
- memory handling,
- or how borderline scenes are prepared for Iray rendering
A separate lighter test also showed that 2026 reports iterations differently than 4.24, but in that case the final image quality and render time were effectively the same.
In this heavier scene, the more important difference is no longer iteration reporting, but actual success vs failure.
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Conclusion:
Under identical test conditions, DAZ Studio 2026 consistently renders a borderline scene that DAZ Studio 4.24 consistently fails to render.
This is significant because:
- the same .duf file is used,
- the same hardware is used,
- the same camera and render settings are used,
- and the result remains reproducible across multiple attempts.
Based on these results, DAZ Studio 2026 appears to handle this specific borderline Iray scenario more successfully than DAZ Studio 4.24.
The main difference does not appear to be simply total VRAM availability. Instead, the evidence suggests that DAZ Studio 2026 is able to enter and sustain a real render workload in this scene, while DAZ Studio 4.24 fails before reaching that stage.
This may be useful for investigating differences between both versions in:
- Iray render startup behavior,
- scene initialization,
- and memory handling near practical GPU limits.
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
© 2026 Daz Productions Inc. All Rights Reserved.
Comments
Please make sure to have your question in the text/post box as well. What graphics card do you have?
Is CPU Fallback enabled in both? Hardware tab of Render Settings.
Nvidia RTX 3070 Ti (as mentioned in the post) – I’m currently running some tests to check the stability of DAZ 2026 before buying a nvidia 50-series card.
What do you mean by “question in the text/post box”? I’m sorry, but I’ve never posted on this forum before
I forgot to mention this important detail: the CPU is disabled, and so is the CPU fallback. I've attached a screenshot of the settings.
Please don't put part of the post, in this case the question, in the subject-line only - as a convenience for readers, especially if they see only a quote of the opening post, make sure the post is complete in itself (but do make the subject line informative, as you did, so that people who might be able to help will see it).
That is strange, it is possible that you are getting the benefit of some optimisations in DS 2026 or the newer Iray version that mean there is enough free memory there but not in DS 4. The log files (Help>Troubleshooting>View Log File) should tell you if there was an error, or if not why the render stopped.
This post doesn’t contain any questions, it is a report/feedback for the build "daz 2026", I’m not sure if this is the right section for feedbacks, so I apologise if I’ve posted in the wrong place.
What this post mean, shortly, is that under the same conditions, DAZ 4.24 fails to render, whereas the "new DAZ 2026 build" starts and finishes the render correctly. This feedback may be useful for investigating differences between both versions in: Iray render startup behavior, scene initialization, and memory handling.
The post should be complete without needing to read the subject line, whatever that is.
Yes, as I said this may be a memory issue. I don't think it is widespread, judging by other posts.
While this may be your intent, you're not preensenting anything useful, and don't seem to even understand what you're doing or looking at.
Without the scene file itself, or a list of all assets used, as well as light, camera and all render settings, your data is uselss.
There's just too many variables to determine if your conclusion is even close to accurate.
Which, even if you provided that information, it's not.
The main problem here, you don't understand what you're looking at in HWINFO, ,in relation to relevant information, and a particular issue it has in relation to iray.
The only things that matter are current and max vram usage, but max only matters if the render is or has been running.
This is where the issue with HWINFO is. When iray exceeds vram and drops out, HWINFO, doesn't understand what just happened, and will report signifcantly less vram usage.
In your test scene, HWINFO reports 7.1GB max, but in reality, iray tried to allocate over 8GB of vram.
If you watch the graph, you'll see a spike then a massive drop in Vram usage. In some cases, it'll be even less than was allocated before you started rendering.
FYI, 2026 does seem pretty stable and if you're getting a 50xx card, it is mandatory to use 2026 because of the updated IRay engine. You may notice that some older scripts will no longer run or fail. I think it has to do with the naming convention of the Daz model inside, I'm not 100% sure. Vendors may have fixes available for their scripts.
It's not just a problem of naming conventions. The code base and the scripting engine have both changed significantly in DS2026 (hence why it's a new major version), so some things need to be coded differently.
For the record: