"Distraction Removal> People" fails locally ("on device") on Intel Arc 140T in LrC 16.0 /ACR 19.0/ PS Beta (Works via Cloud & Nvidia GTX 1650)
Subject: Bug: "Distraction Removal > People" fails locally ("On Device") on Intel Core Ultra 9 285H / Arc 140T in LrC 16.0, ACR 19.0 & PS Beta (Works via Cloud & on NVIDIA GTX 1650)
Summary:
On a Windows 11 mini PC powered by an Intel Core Ultra 9 285H with integrated Intel Arc 140T GPU (Vendor: 8086, Device: 7d51), Distraction Removal > People consistently fails locally with "No distractions found" across Lightroom Classic 16.0, Adobe Camera Raw 19.0, and Photoshop Beta (set to "On Device"). However, switching Photoshop Beta's Remove Tool processing mode from On Device to Cloud detects and removes the exact same people properly. Additionally, local Masking > People works properly on the Intel Arc 140T, and local ("On Device") Distraction Removal > People works properly on the exact same software builds and .dng files on a secondary laptop (Dell XPS 15 7590) equipped with an NVIDIA GeForce GTX 1650.
Notably, Lightroom Classic's System Info reports GPUDeviceEnabled: false and Dedicated GPU memory used by Lightroom: 155,3MB / 128,0MB (121%) on this Intel Arc 140T adapter despite 24GB of available GPU memory.
System Comparison (Identical Software & Files):
- Machine 1 (Failing Locally):
- Hardware: Geekom Mini PC, Intel Core Ultra 9 285H (2.90 GHz, 16 logical processors), Intel Arc 140T GPU (24GB) (
Vendor: 8086, Device: 7d51, Revision: 3), 48 GB RAM - OS: Windows 11 Business Edition (
11.0.26300) - Intel Graphics Driver Version:
32.0.101.9175 - Lightroom Classic Version:
16.0 [ 202611041403-7ac14c66 ] - Internal Camera Raw Version:
19.0
- Hardware: Geekom Mini PC, Intel Core Ultra 9 285H (2.90 GHz, 16 logical processors), Intel Arc 140T GPU (24GB) (
- Machine 2 (Working Locally):
- Hardware: Dell XPS 15 7590, Intel UHD 630 + NVIDIA GeForce GTX 1650
- OS: Windows 11
- Software Versions: Exact same LrC
16.0, ACR19.0, and PS Beta builds as Machine 1
Steps to Reproduce:
- Open a
.dngphoto containing clear background people/bystanders in Lightroom Classic 16.0 on the Intel Core Ultra 9 285H / Arc 140T system and run Remove > Distraction Removal > People. - Observed Result: Immediately returns "No distractions found."
- Open the Masking panel (
Shift + W) on the exact same photo and select People. - Observed Result: Local people segmentation detects all individuals accurately.
- Open the
.dngin Photoshop Beta (via Adobe Camera Raw 19.0) and run Distraction Removal > People inside ACR. - Observed Result: Returns "No distractions found."
- Click Open to load the image into Photoshop Beta's native canvas, select the Remove Tool (
J), set processing mode to On Device, and click Find distractions > People. - Observed Result: Fails locally on the Intel Arc 140T.
- Switch the Remove Tool processing mode in Photoshop Beta from On Device to Cloud and click Find distractions > People on the exact same image.
- Observed Result: Detects and removes the background people properly.
- Test the exact same
.dngfile locally (On Device) in Lightroom Classic16.0, ACR19.0, and Photoshop Beta on the Dell XPS 15 7590 (NVIDIA GTX 1650) system. - Observed Result: Local ("On Device") Distraction Removal > People works properly in all three apps.
Relevant Excerpt from Help > System Info (Intel Arc 140T):
text
Lightroom Classic version: 16.0 [ 202611041403-7ac14c66 ]
Internal Camera Raw version: 19.0
Operating system: Windows 11 - Business Edition (11.0.26300)
Built-in memory: 48662,4 MB
Dedicated GPU memory used by Lightroom: 155,3MB / 128,0MB (121%)
DirectX: Intel(R) Arc(TM) 140T GPU (24GB) (32.0.101.9175)
Init State: GPU for Export supported by default
User Preference: Auto
Adapter #1: Vendor : 8086 / Device : 7d51 / Subsystem : 22128086 / Revision : 3 / Video Memory : 128
GPUDeviceEnabled: false
Expected Result:
Local (On Device) Distraction Removal > People in Lightroom Classic 16.0, Camera Raw 19.0, and Photoshop Beta should detect background people on the Intel Core Ultra 9 285H / Arc 140T (Device: 7d51) just as it does via Cloud processing and on NVIDIA GPUs.
