P: LrC v15.5 on Windows Assisted Culling triggers crash and memory leaks

​@Sameer K  - I get no error message or crash notification.  Lr simply vanishes.  

I’ve seen some posts where others can turn off Assisted Culling after a restart.  I could not.   I could only keep restarting Lr until Assisted Culling finished.  I could switch to the folder I had set before starting Assisted Culling and that was it.   I had no ability to do anything else in Lr between crashes.

As for the behavior starting before v15.5 - This was literally the first time I’d used Assisted Culling.  Needless to say, my impression is not good! 😀

My PC is up to date with MS updates as well.   

If there is a way I can find crash logs or any other info you can use, please let me know how to get to it!  

# Lightroom Classic 15.5 crashes with Assisted Culling enabled — 0xc0000374 heap corruption

## Summary

Lightroom Classic 15.5 repeatedly crashes while working with a very large catalog when Assisted Culling is active.

After several controlled tests, Assisted Culling appears to be the trigger. Lightroom has now run overnight without a crash with **GPU acceleration enabled and Duplicate Detection enabled, but Assisted Culling disabled**.

## System

* **Lightroom Classic:** 15.5, build 202607291506-b8869fa7

* **Camera Raw:** 18.5

* **Operating system:** Windows 11

* **GPU:** NVIDIA GeForce RTX 4080 SUPER, 16 GB

* **NVIDIA driver:** 32.0.16.1088

* **System RAM:** approximately 64 GB

* **Catalog location:** local NVMe/SSD drive

* **Photo originals:** primarily stored on network-attached storage (NAS)

* **Displays:** 3

## Crash information

Windows Event Viewer repeatedly reports:

* **Faulting application:** Lightroom.exe

* **Lightroom version:** 15.5.0.8

* **Faulting module:** ntdll.dll

* **Exception code:** `0xc0000374`

* **Event ID:** 1000

The Lightroom console log records:

`AICulling:start`

shortly before the failure.

## Testing performed

### Initial behavior

With Assisted Culling active, Lightroom repeatedly crashed while I was working with the catalog.

The catalog itself could open successfully. In some cases Lightroom ran for a short period before crashing.

### GPU test

I completely disabled Lightroom's GPU acceleration and restarted Windows.

Lightroom still crashed with the same `0xc0000374` error.

This suggests the problem is not dependent on Lightroom GPU acceleration.

### Assisted Culling / Duplicate Detection test

I then disabled both:

* Assisted Culling

* Automatic Duplicate Detection

Lightroom became stable.

I then performed a more specific isolation test:

* **GPU acceleration: ON**

* **Duplicate Detection: ON**

* **Assisted Culling: OFF**

With this configuration, Lightroom ran **all night without any crashes** while Duplicate Detection continued working.

This is a major change from the repeated crashes experienced when Assisted Culling was active.

## Expected result

Assisted Culling should analyze photographs in the background without causing Lightroom Classic to terminate.

## Actual result

When Assisted Culling is active, Lightroom repeatedly terminates with Windows exception:

`0xc0000374`

Windows reports the crash through `ntdll.dll`.

Disabling Assisted Culling appears to prevent the crashes.

## Workaround

Disable Assisted Culling under:

**Catalog Settings → Metadata → Assisted Culling**

I can continue using Lightroom normally with:

* GPU acceleration enabled

* Duplicate Detection enabled

* Assisted Culling disabled

## Additional information

This is an unusually large Lightroom catalog. The catalog database itself is on a fast local drive, while most original photographs are stored on NAS storage.

The catalog opens successfully, so the failure does not appear to occur during initial catalog opening.

I can provide:

* Lightroom `lrc_console.log`

* Windows Event Viewer screenshots

* Windows Error Reporting information

* Lightroom System Info

The console log captured `AICulling:start` during the affected session.

Because Lightroom remains stable with GPU acceleration and Duplicate Detection enabled as long as Assisted Culling remains disabled, Assisted Culling appears to be the strongest common factor in the crashes.

@Sameer K - answering your question directly: no, this did not start with 15.5.

On my machine (Windows 11, LrC 15.5, GTX 1650, 32 GB RAM) the first heap-corruption
crash was 4 July 2026 on 15.4.1 - five weeks before 15.5 shipped. The install dates
below come from Adobe's own installer log, Install.log:

  Installed     Version    Days in use    0xc0000374 crashes
  ----------    -------    -----------    ------------------
  2026-03-19    15.2.1          30                 0
  2026-04-18    15.3            53                 0
  2026-06-10    15.3.1          10                 0
  2026-06-20    15.4             3                 0
  2026-06-23    15.4.1          45                 4
  2026-08-07    15.5             8                 8

96 crash-free days across four consecutive versions, then 15.4.1. Version 15.5 made it
roughly 11x worse - three crashes within five hours of installing it - but it did not
introduce the problem.

If the investigation is scoped to 15.5 only, I believe the regression will be missed.
I would suggest looking at what changed in the Assisted Culling / WML path between
15.4 and 15.4.1.

One clarification on symptom: mine is a hard crash rather than the import popup lock
described in the original post - exception 0xc0000374 (STATUS_HEAP_CORRUPTION) in
ntdll.dll, landing 5 and 30 seconds after AICulling:start on two separate occasions,
with no AICulling:complete. Full details, timeline and logs are in my post earlier in
this thread. Disabling automatic culling analysis does not prevent it; the second crash
occurred under manual Auto Analysis with AICulling_enableAutoCullingAnalysis = false.

On crash reports: I have now added my Community e-mail address to the Adobe Crash
Reporter. My previous 12 crashes were submitted anonymously - the Email Id field was
blank - so they will not be linked to this account. If it helps you locate them, the
crash reporter GUID on this machine is:

  fced2b08-11bd-487e-aff3-77043b35c688

Happy to reproduce on demand and submit a fresh report with contact details attached.

Moderators, ​@Sameer K,  ​@Rikk Flohr_Photography, please merge with the existing post:

P: Assisted Culling popup locks application | Adobe DME

LrC 15.5,  Windows 11
AMD Ryzen 7 3700X, 32GB Ram, RTX 2060 GPU (6GB)

I have all auto analysing turned off and LrC runs fine. I am able to do everything.

But, if I decide to start any analysing (i.e. trigger the culling process), it runs for a few seconds then I get a Windows dialog telling me that Lightroom has crashed.
I have tried turning off all GPU options and the problem persists.

I have attached a very long crash log - if anyone can review it.
Here is the tail of that log ...

 

2026-08-16T21:32:50.280795+01:00 |       Collection.cpp:1339  |    error | auto Collection::setPersonalizationRequestTimeout::<>::operator ()(void)  | Failed edge request on error | 
The thread 'session-13-intra-op-4' (10012) has exited with code 0 (0x0).
The thread 'session-13-intra-op-1' (31008) has exited with code 0 (0x0).
The thread 'session-13-intra-op-0' (21092) has exited with code 0 (0x0).
The thread 'session-13-intra-op-2' (4124) has exited with code 0 (0x0).
The thread 'session-13-intra-op-3' (16924) has exited with code 0 (0x0).
The thread 'session-13-intra-op-5' (7740) has exited with code 0 (0x0).
The thread 'session-13-intra-op-6' (27232) has exited with code 0 (0x0).
The thread 20788 has exited with code 0 (0x0).
2026-08-16T21:32:50+01:00: *** Warning: no ML mask found ***
Exception thrown at 0x00007FF9816B187A in Lightroom.exe: Microsoft C++ exception: dng_exception at memory location 0x000000005A77ADA0.
Exception thrown at 0x00007FF9816B187A in Lightroom.exe: Microsoft C++ exception: cv::Exception at memory location 0x0000000059A7D630.
Exception thrown at 0x00007FF9816B187A in Lightroom.exe: Microsoft C++ exception: cv::Exception at memory location 0x00000000617ED630.
Critical error detected c0000374
A breakpoint instruction (__debugbreak() statement or a similar call) was executed in Lightroom.exe.

Exception thrown at 0x00007FF983F12165 (ntdll.dll) in Lightroom.exe: 0xC0000374: A heap has been corrupted (parameters: 0x00007FF983FCC0E0).
Unhandled exception at 0x00007FF983F12165 (ntdll.dll) in Lightroom.exe: 0xC0000374: A heap has been corrupted (parameters: 0x00007FF983FCC0E0).

The program '[12940] Lightroom.exe' has exited with code 0 (0x0).

I never received a crash report dialog during any of my crashes - if I had, you’d have multiple reports already.

I recently enabled Assisted Culling Analysis after upgrading to 15.5, and I’m noticing silent crashes. Disabling the analysis process seems to prevent crashes. The analysis is actually making progress, as I have tens of thousands of photos, and I see the number ticking down. Sometimes LRC only lasts a couple minutes, sometimes it lasts longer (going on 10 minutes as I write). I also noticed that my GPU spins up just a little (not 100%) in cases where a crash ultimately comes (though it also spins up other times without issue). I would guess that this behavior might be consistent with a specific media item triggering the crash.

Nvidia 4090 on Windows 11.

Reposting the report below in plain text - the original paste lost its table
formatting and picked up my draft heading. Please read this version instead.
Moderators: this was posted under Questions by mistake, could it be moved to
Bug Reports? Logs are attached to this reply.

------------------------------------------------------------------------------
SUMMARY
=======
Lightroom Classic crashes with STATUS_HEAP_CORRUPTION (0xc0000374) in ntdll.dll on
Windows 11. Twelve occurrences since 4 July 2026. The crash is strongly correlated with
Assisted Culling analysis.

This machine ran 15.2.1, 15.3, 15.3.1 and 15.4 across 96 days with zero heap-corruption
crashes. The first occurrence appeared 11 days after 15.4.1 was installed, and the crash
rate increased roughly 11x after upgrading to 15.5.

ENVIRONMENT
===========
Lightroom Classic : 15.5.0.8 (crashes also on 15.4.1.1)
Camera Raw        : 18.5
OS                : Windows 11, build 26200
CPU               : Intel Core i7-12700F (12C/20T, AVX2)
RAM               : 32 GB
GPU               : NVIDIA GeForce GTX 1650, 4 GB VRAM
GPU driver        : 32.0.16.1088 (Studio 610.88, clean install)
Catalog disk      : WD Blue SN570 NVMe SSD, 216 GB free
Display           : single 2560x1440

VERSION HISTORY AND CRASH CORRELATION
=====================================
Install dates taken from Adobe's own installer log:
C:\Program Files (x86)\Common Files\Adobe\Installers\Install.log

  Installed     Version    Days in use    0xc0000374 crashes
  ----------    -------    -----------    ------------------
  2026-03-19    15.2.1          30                 0
  2026-04-18    15.3            53                 0
  2026-06-10    15.3.1          10                 0
  2026-06-20    15.4             3                 0
  2026-06-23    15.4.1          45                 4
  2026-08-07    15.5             8                 8

15.5 was installed at 09:51 on 7 August. Three crashes followed the same day at
14:25, 14:29 and 14:35.

Windows Error Reporting on this machine retains records back to December 2025, so the
crash-free period before 4 July is genuine and not an artefact of log rotation.

CRASH SIGNATURE
===============
Exception code   : c0000374 (STATUS_HEAP_CORRUPTION)
Faulting module  : ntdll.dll, offset 0x112165 (identical in all 12)
Affected builds  : 15.4.1.1 and 15.5.0.8
WER buckets      : StackHash_3394 (x4), StackHash_dfbf (x3),
                   StackHash_496a (x2), plus 0000 / f104 / c040 (x1 each)

Lightroom.exe is the only process on this machine that has ever raised 0xc0000374,
verified across all Windows Error Reporting records.

REPRODUCTION - MOST RECENT OCCURRENCE
=====================================
From lrc_console.log, captured before relaunch (15 August 2026):

  12:35:57.848   Module Switched: develop -> library
  12:35:58.464   isConstrainedEnvForCulling: false
  12:35:58.593   AICulling:start
  12:36:28       CRASH - no matching AICulling:complete

Assisted Culling was invoked MANUALLY via Identity Plate > Auto Analysis.
AICulling_enableAutoCullingAnalysis was false in that session, so automatic analysis
was not involved. Disabling automatic analysis is therefore not a workaround.

In the 30 seconds before the crash, the Camera Raw log records 142 ML inferences across
8 models:

  adobeone-mobileclip, mobileclip-based-classifier, subject-lite-culling,
  undesirable-blur, face-detector, eyes-open-close, bfs-frame-score, bfs-face-score

A second crash on 14 August followed the same pattern: AICulling:start at 19:44:43,
crash at 19:44:48, again with no AICulling:complete.

It is not deterministic. On 14 August, five culling cycles completed normally in the
minute before the sixth crashed.

NOT MEMORY EXHAUSTION
=====================
Heap telemetry at the moment of the crash:

  Private commit  : 6.956 GB
  Heap regions    : 10,025
  Camera Raw CPU  : 1.515 GB

32 GB installed, 16 GB free. This is corruption, not exhaustion.

ALREADY RULED OUT
=================
GPU driver            Clean install of Studio 610.88; stale 591.86 package removed
                      from the driver store. Crash persists.
GPU acceleration      GPUManagerPref = "off", all useGPU* flags false. Crash persists.
Auto culling analysis Disabled. Crash persists under manual invocation.
Disk space            216 GB free on NVMe. Crash rate actually INCREASED after 242 GB
                      was freed on 8 August.
System RAM pressure   See heap telemetry above.
Camera Raw 18.5       Shipped 5 August; crashes began 4 July.

ADDITIONAL FINDING: GPU PREFERENCE DOES NOT DISABLE WML GPU USAGE
=================================================================
With GPUManagerPref = "off", the Camera Raw log still shows:

  Initialize GPU system
  Loaded GPU system: DirectX
  GPU4 Soft Status Result: fail_insufficient_memory
  Supports FP16 for DX on WML: YES
  WML: Try use GPU? yes
  259 compute pipeline states built
  Current memory usage (GPU): 0.698GB  Budget: 2.613GB (pool) 1.699GB (cache)

nvwgf2umx.dll, nvgpucomp64.dll, nvldumdx.dll and nvppex.dll remain loaded in the process.

So the Windows ML inference path continues to schedule work on the GPU even when the
user has disabled graphics acceleration. There appears to be no supported way to move
Assisted Culling inference onto the CPU.

This exact behaviour was independently reported by JINGORO8816 on 30 July 2026, who
observed 262 "Building compute pipeline state" lines and the same
fail_insufficient_memory result with all GPU preferences set to off.

VRAM capacity does not appear to be the determining factor: AD-Tran reports the same
Assisted Culling crash on an RTX 4060 Laptop GPU with 8 GB of VRAM.

RELATED EXISTING REPORTS
========================
AD-Tran, 10 Aug 2026 - Win 11, RTX 4060 Laptop 8 GB, 32 GB RAM
  Same symptom: 15.5 crash during Assisted Culling / Auto Analysis.
  https://community.adobe.com/questions-675/lightroom-classic-15-5-crashes-during-assisted-culling-auto-analysis-on-windows-11-rtx-4060-laptop-gpu-32-gb-ram-please-investigate-1636144

JINGORO8816, 30 Jul 2026 - Win 10, RTX 3050 Ti
  Same exception (0xc0000374 in ntdll) and the same GPU-not-disabled finding.
  Trigger is a 143 MP import rather than culling.
  https://community.adobe.com/questions-675/lrc-15-4-1-15-1-14-5-2-100-reproducible-crash-ntdll-heap-corruption-importing-100mp-images-disabling-gpu-has-no-effect-1634663

kgdizainslv, Jul 2026 - macOS
  15.4/15.4.1 AI Assist Culling memory leaks and freezes. Marked Released.
  https://community.adobe.com/bug-reports-674/p-lightroom-classic-v15-4-15-4-1-causes-ai-assist-culling-memory-leaks-mac-freezes-and-denoise-failures-1630421

naemick, Jul 2026
  15.4.1 persistent crashes and AI Culling memory issues.
  https://community.adobe.com/questions-675/lightroom-classic-v15-4-1-causes-persistent-crashes-launch-failures-and-ai-culling-memory-issues-1629986

Caveat offered in good faith: JINGORO8816 reproduced his crash on 14.5.2.1 and 15.1.0.5
as well, so the underlying heap defect may predate 15.4.1. On this machine, however, the
15.4.1 boundary is sharp - 96 crash-free days across four prior versions.

ATTACHMENTS
===========
lrc_console-aug15-culling-crash.log   Captured before relaunch; culling timeline and
                                      main-thread stack
lrc_console-aug14-crash2.log          14 August crash
cameraraw-aug15-culling-crash.txt     ML inference timings and heap telemetry

Crash dumps available on request.

WHAT WOULD HELP
===============
1. Confirmation of what changed in the Assisted Culling / WML path between 15.4 and 15.4.1.
2. A supported way to force Assisted Culling inference onto the CPU.
3. Whether the WML GPU path is expected to remain active when graphics acceleration
   is disabled in Preferences.

ATTACHED TO THIS REPLY
======================
lrc_console-aug15-culling-crash.txt   Captured BEFORE relaunching Lightroom, so it
                                      still contains the culling timeline and stack.
                                      Lightroom overwrites this file on every launch,
                                      which is why it is rarely available.
cameraraw-aug15-culling-crash.txt     142 ML inference entries and heap telemetry at
                                      the moment of the crash.
lrc_console-aug14-crash2.txt          The 14 August culling crash.

Crash dumps available on request.

Same on Windows 11 PC. Crashes every 5-10 minut