Key Takeaways for ON1 Portrait AI Users
- ON1 Photo RAW 2026 integrates Portrait AI for automated face detection, skin smoothing, eye enhancement, face sculpting, blemish removal, and subject masking, all running locally on your GPU without cloud dependency.
- Face sculpting uses per-face warp meshes with sliders such as Slim Face, Nose Size, Chin, and Forehead, so you can adjust each subject in a group portrait without affecting adjacent faces.
- Batch processing applies consistent Portrait AI settings across large RAW sets through sync operations, although overall speed depends on your local GPU performance rather than shared cloud servers.
- Skin smoothing, eye enhancement, and individual face control rely on automatic masks that sometimes need manual brush refinement, especially for darker skin tones or distant subjects in group shots.
- To generate unlimited, consistent portrait content without reshooting, use Sozee for on-brand portrait generation alongside your RAW retouching tools.
How ON1 Photo RAW Handles Face Sculpting
Face sculpting in ON1 Photo RAW 2026 activates automatically once Portrait AI detects a face in the frame. The engine maps facial landmarks and builds a warp mesh for each face before you touch any slider. The primary sliders, Slim Face, Nose Size, Chin, and Forehead, apply localized geometric distortion rather than global warping, which helps protect background elements from unintended shifts.
The recommended workflow starts with Slim Face at no more than 15–20 units to avoid an artificial look. Once you set face width, Chin and Forehead adjustments then balance overall proportions without fighting the initial slim setting. This sequential approach matters in group shots because Portrait AI generates an independent warp mesh for every detected face, so each subject needs its own proportional balance. For diverse skin tones, the sculpting layer affects geometry only and does not touch color channels, so it introduces no skin-tone bias at the sculpting stage.
Manual refinement remains available when the automatic mesh boundary is imprecise. You can paint the warp mask with a brush to exclude areas such as ears or the hairline where you want zero distortion.
To move beyond single-tool retouching, use Sozee to generate new portrait variations and keep your RAW edits focused on captured images.
Batch Portrait Editing in ON1 Photo RAW
ON1 Photo RAW 2026 supports batch processing of large RAW image sets through Portrait AI. You can apply skin smoothing, eye enhancement, and other retouching adjustments consistently across an entire folder. The batch trigger is a sync operation: edit one hero image, select all target images in the filmstrip, then use Edit > Sync Settings to push the Portrait AI panel values across the selection.
The main constraint is hardware dependency. All AI processing runs locally on the user's GPU, so batch throughput scales directly with GPU capability instead of shared cloud infrastructure. AI batch editing that once took days now finishes in hours, and photographers using AI culling and editing tools saved an estimated 89 million hours collectively in 2025, but ON1 users only see those gains up to the limit of their local machine. The 2026.3 update restored GPU acceleration on macOS after a temporary CPU fallback, which meaningfully improves batch speed on Apple Silicon hardware.
The same 2026.3 release also improved Auto Level consistency across similar images. Running Auto Level before Portrait AI sync helps normalize exposure in event photography batches and reduces visible variation between adjacent frames.
ON1 Photo RAW AI Skin Smoothing in Practice
ON1 Portrait AI automatically detects faces and generates a separate skin mask that isolates skin from eyes, eyebrows, lips, and hair before any smoothing occurs. The primary sliders are Smoothing, Texture, and Pores. Smoothing reduces mid-frequency skin variation, Texture controls fine-detail preservation, and Pores targets high-frequency micro-texture.
A reliable starting point for natural results uses Smoothing at 30–40, Texture at the 50 default, and Pores at 20–30. Smoothing above 60 without compensating with Texture often creates a plastic appearance. For darker skin tones, the automatic skin mask can under-select high-contrast shadow areas, a known edge case that needs manual brush refinement to expand the mask boundary. Painting the mask at 50% opacity in shadow zones instead of 100% helps preserve tonal gradation.
The skin smoothing layer is non-destructive. You can toggle or delete it at any time without affecting other Portrait AI adjustments.
ON1 Photo RAW Eye Enhancement Controls
ON1 Portrait AI generates a dedicated eye mask that covers the sclera and iris independently from the skin mask. The three primary sliders are Eye Brightness, Eye Whitening, and Eye Detail. Eye Brightness lifts overall luminosity within the mask, Eye Whitening reduces warm and red tones in the sclera, and Eye Detail adds local contrast and sharpening to the iris.
For single portraits, the automatic eye mask remains accurate under standard lighting. In group shots with subjects at varying distances, face detection accuracy drops for faces that occupy fewer pixels in the frame, so the eye mask may be missing or imprecise for background subjects. Manual addition of a brush mask targeting distant faces corrects this issue. Eye Detail above 70 introduces visible sharpening halos on subjects shot with shallow depth of field, so keeping it between 40 and 55 avoids that artifact.
The eye enhancement tools do not interact with the skin smoothing mask. Adjustments to one region stay isolated and do not bleed into the other.
ON1 Portrait AI Controls for Each Face in a Group
ON1 Portrait AI generates an automatic subject mask that lets you adjust background brightness or blur while keeping the portrait subject untouched. When an image contains multiple subjects, the per-face architecture assigns each detected face its own mask set, including skin, eyes, and mouth. You can apply different smoothing or enhancement values to each person without affecting adjacent faces.
The workflow for individual face control in a group portrait follows a simple sequence. Open Portrait AI and confirm face detection thumbnails in the panel, where each detected face appears as a selectable chip. Select the target face chip and adjust sliders for that face only. Then select the next chip and apply different values as needed. Faces not detected, typically those turned more than about 45 degrees from camera or partially occluded, require a manual brush mask addition.
ON1 Photo RAW 2026 also includes a Mask Layers system that supports stacking and combining multiple masks per filter. This extends individual face control to non-Portrait-AI adjustments such as local color grading or dodging.
If per-face control at scale slows your delivery times, Sozee removes the mask-refinement step so you can generate consistent, on-brand portraits across unlimited subjects without touching a single slider.
What Changed in ON1 Photo RAW 2026
The 2026.3 update delivered three workflow-relevant changes. It restored GPU acceleration on macOS for AI features including Resize AI, Generative Crop, and Generative Erase after a prior CPU fallback, which is detailed earlier in the batch processing section. It improved Auto Level consistency across similar images for batch and event workflows. It also added custom color label renaming with subtle background tints in Grid View to speed culling in large portrait sets.
The 2026.4 update introduced Restore AI, available in the Photo RAW MAX edition, which targets repair, colorization, and restoration of degraded or archival portraits. The 2026 release also added one-click Subject and Background masks and improved AI edge detection for hair and fine detail compared to prior versions.
Real-World Limitations of ON1 Portrait AI
Batch consistency in ON1 is bounded by local GPU throughput. Unreviewed AI pipelines at high volume frequently show lighting drift and color inconsistency between batches, and that risk applies to any locally synced Portrait AI workflow when source images vary in exposure or white balance.
Skin-tone bias appears at the mask-generation stage. The automatic skin mask can under-select shadow regions on darker complexions, which requires manual brush correction. This pattern does not belong to ON1 alone and reflects a broader industry issue, but photographers serving diverse client bases must budget time for mask refinement per image instead of assuming full automation.
Restore AI's face recovery option frequently replaces original facial features, including gaze direction, eye color, tooth shape, and lip form, rather than restoring them. That behavior creates a significant limitation for heirloom or archival portrait work where likeness fidelity is the primary requirement.
For high-volume workflows that exceed several hundred portraits per session, the local versus cloud processing model becomes the main bottleneck. Evoto AI processes 1,000 photos in approximately two minutes through cloud batch tools, compared to ON1's local GPU constraint. AI batch editing tools can process 1,000 photos in under 10 minutes, and ON1 typically sits between those poles depending on hardware. When the goal is generative AI portrait output instead of RAW retouching, ON1 does not fit the required tool category.
Frequently Asked Questions
Does ON1 Photo RAW have AI features?
Yes. ON1 Photo RAW 2026 includes multiple AI-powered tools across its editing suite. Portrait AI handles automated face detection and generates local masks for skin, eyes, and mouth regions, which enables targeted skin smoothing, eye enhancement, teeth whitening, blemish removal, and face sculpting without manual brushing. The software also includes AI-powered Subject and Background masking with improved edge detection for hair and fine detail, Resize AI for upscaling, Generative Crop, Generative Erase for content-aware object removal, and Restore AI for archival photo repair and colorization in the MAX edition. All AI processing runs locally on the user's hardware using GPU acceleration, which is fully restored on macOS as of the 2026.3 update.
Is ON1 as good as Lightroom?
ON1 Photo RAW 2026 and Adobe Lightroom serve overlapping but distinct use cases, and neither works as a universal winner. ON1 operates as a standalone RAW processor with integrated Portrait AI retouching, local GPU processing, and a non-subscription perpetual license option, which benefits photographers who prioritize offline operation, portrait-specific automation, and cost predictability. Lightroom's AI tools, powered by Adobe Sensei, include subject recognition, AI skin softening, and eye emphasis sliders with deep integration into the Adobe ecosystem, Creative Cloud sync, and a larger third-party plugin library.
Lightroom's cloud infrastructure enables faster batch processing on lower-spec hardware. ON1's Portrait AI offers more granular per-face control, including individual face chips, stacked mask layers, and dedicated sculpting sliders, than Lightroom's current portrait toolset. The practical choice depends on workflow. Photographers already in the Adobe ecosystem with cloud-heavy batch needs generally find Lightroom more efficient, while photographers who prioritize local processing, portrait-specific automation, and perpetual licensing often find ON1 competitive or preferable for portrait-focused work.
Conclusion: When ON1 Fits and When Sozee Helps
ON1 Photo RAW 2026's Portrait AI delivers a capable, locally processed portrait retouching system with per-face masking, dedicated skin, eye, and sculpting sliders, and batch sync across RAW files. The 2026.3 and 2026.4 updates improve GPU performance, batch consistency, and edge detection in real workflows. Photographers still need to account for skin-tone mask limitations on diverse complexions, hardware-bounded batch throughput, and Restore AI's tendency to replace rather than restore facial features.
For workflows that focus on retouching existing RAW captures, ON1 functions as a credible tool. For workflows that require unlimited, consistent portrait content generation without a camera or reshoot, a different platform category delivers better results. Upload three photos to Sozee to generate unlimited on-brand portraits and bypass hardware constraints and manual mask refinement entirely.