← All articles 10 min read

How to Upscale an Image in Photoshop (3 Methods, 2025)

You have a 600 x 400 px image and need it at 2400 x 1600 px. Stretching it in Image Size gives you a blurry mess. Photoshop 2025 (v26.x) has three AI-powered upscaling methods that actually add convincing detail instead of smearing pixels. Each one works differently, produces different results, and suits different situations.

Here is what each method does, when to pick it, and exactly how to use it.

Comparison: 3 Photoshop Upscaling Methods at a Glance

Feature Super Resolution (Camera Raw) Preserve Details 2.0 (Image Size) Neural Filters (Super Zoom)
Max upscale 2x linear (4x total pixels) Unlimited (manual entry) Up to 4x
AI model Adobe Sensei (dedicated) Machine learning resampling Adobe Sensei neural network
Input formats RAW, JPEG, TIFF, PNG Any open document Any open document
Best for RAW files, photos needing maximum detail General-purpose enlargement, precise control Heavy upscaling (3-4x), recovering lost detail
Speed 5-30 sec depending on file size Instant preview, fast apply 10-60 sec (cloud-processed)
Non-destructive Yes (creates new DNG) No (modifies canvas) Optional (new layer)
Quality (2x upscale) Excellent — sharpest edges, best texture Very good — slight softness on fine detail Good — occasional artifacts on text/patterns
Requires internet No No Yes (cloud processing)

Method 1: Super Resolution (Camera Raw)

Super Resolution doubles the linear resolution of your image — a 3000 x 2000 px photo becomes 6000 x 4000 px. Adobe trained this model specifically on photographic detail, so it excels at recovering texture, hair, fabric weave, and foliage.

This method works on RAW files, JPEGs, TIFFs, and PNGs. It produces a new DNG file, leaving your original untouched.

Step-by-step

  1. Open the image in Camera Raw. If you have a RAW file, double-click it and Photoshop opens Camera Raw automatically. For JPEG/PNG/TIFF: right-click the file in your OS, select Open With > Adobe Photoshop 2025, then go to Filter > Camera Raw Filter (or press Shift + Ctrl/Cmd + A).

  2. Right-click the image preview and select Enhance from the context menu. Alternatively, click the three-dot menu (⋯) in the filmstrip and choose Enhance.

  3. Check the "Super Resolution" box in the Enhance Preview dialog. You will see a side-by-side preview — hover over detailed areas (hair, text, patterns) to evaluate quality before committing.

  4. Click Enhance. Processing takes 5-30 seconds depending on file size and your hardware. Photoshop creates a new DNG file alongside your original with -Enhanced appended to the filename.

  5. Open the enhanced DNG in Photoshop for further editing, or export directly from Camera Raw.

When to use Super Resolution

Limitations

Super Resolution caps at 2x linear (4x total pixel count). If you need more, you can run Super Resolution and then apply Preserve Details 2.0 on top, but expect diminishing returns.

Method 2: Preserve Details 2.0 (Image Size)

Preserve Details 2.0 is Photoshop's built-in resampling algorithm, available in the Image Size dialog. It uses machine learning to add detail during upscaling, and it gives you precise control over the exact output dimensions.

Unlike Super Resolution, this method modifies your document directly. Duplicate the layer or use Image > Duplicate first if you want to keep the original.

Step-by-step

  1. Open your image in Photoshop 2025 (v26.x).

  2. Go to Image > Image Size (or press Alt/Option + Ctrl/Cmd + I).

  3. Set the resampling method. Click the Resample dropdown at the bottom of the dialog and select Preserve Details 2.0. If you do not see this option, go to Preferences > Technology Previews and enable Enable Preserve Details 2.0 Upscale (it ships enabled by default in v26.x, but confirm).

  4. Enter your target dimensions. Type the width or height you need. Keep the chain-link icon locked to maintain aspect ratio. You can also enter a percentage — 200% doubles the size.

  5. Adjust the Noise Reduction slider. Preserve Details 2.0 adds a noise reduction control (0-100%). For clean originals, keep it at 0-10%. For noisy photos (high ISO, phone cameras in low light), push it to 30-60%. Over-reducing noise softens detail, so check the preview at 100% zoom.

  6. Check the preview. Click inside the preview window and drag to inspect critical areas — edges, text, fine textures. Compare against the original at the same zoom level.

  7. Click OK to apply.

When to use Preserve Details 2.0

Limitations

Preserve Details 2.0 modifies the document in place — there is no separate output file. Heavy upscaling (beyond 3-4x) introduces visible softness. For extreme enlargements, stack it with Super Resolution or use Neural Filters instead.

Method 3: Neural Filters (Super Zoom)

Neural Filters use cloud-based AI models to upscale images. The Super Zoom filter specifically targets upscaling and can push beyond 2x while attempting to reconstruct fine detail that the other methods might miss.

This method requires an internet connection because processing happens on Adobe's servers.

Step-by-step

  1. Open your image in Photoshop 2025 (v26.x).

  2. Go to Filter > Neural Filters. The Neural Filters workspace opens.

  3. Find Super Zoom in the filter list. If it is not downloaded, click the cloud icon to download the model (one-time, approximately 150 MB).

  4. Enable Super Zoom by toggling it on. Set your zoom level — options range from 1x to 4x. Start with 2x to evaluate quality.

  5. Toggle "Enhance Face Detail" if your image contains faces. This sub-model specifically targets facial features and produces noticeably sharper eyes, skin texture, and hair at the cost of occasional over-sharpening.

  6. Choose the output mode. Select New Layer to keep the original intact, or Current Layer to replace it. New Layer is the safer choice.

  7. Click OK. Processing takes 10-60 seconds, depending on image size and Adobe's server load. Progress displays in the bottom-left corner.

When to use Neural Filters

Limitations

Neural Filters require an active internet connection and an Adobe Creative Cloud subscription. Processing time is unpredictable — during peak hours, expect delays. The AI can introduce artifacts on repeating patterns (brick walls, fabric weaves, chain-link fences). Always inspect at 100% zoom before committing.

Which Method Produces the Best Results?

The short answer: Super Resolution wins for 2x photo upscaling. It produces the sharpest edges, best texture recovery, and fewest artifacts. Adobe trained this model specifically for photographic enlargement, and it shows.

For a more nuanced breakdown:

At 2x upscale (the most common scenario): Super Resolution > Preserve Details 2.0 > Neural Filters. The gap between Super Resolution and Preserve Details 2.0 is small — you will only notice it at 100% zoom on fine textures like hair and fabric. Neural Filters trail slightly, with occasional over-sharpening on edges.

At 3-4x upscale: Neural Filters > Super Resolution + Preserve Details 2.0 stacked > Preserve Details 2.0 alone. At extreme enlargements, Neural Filters' cloud model handles detail hallucination better. Stacking Super Resolution (2x) and then Preserve Details 2.0 (another 1.5-2x) is a solid second choice.

For RAW files: Super Resolution is the clear winner. It works in Camera Raw's native pipeline, preserves the RAW advantage (more color data, less compression), and produces a DNG you can continue editing non-destructively.

For illustrations and graphics: Preserve Details 2.0 with noise reduction at 0%. The other methods are trained on photographic content and can introduce unwanted texture into flat-color areas.

For faces: Neural Filters with "Enhance Face Detail" enabled. The dedicated face model recovers eye detail and skin texture that the other methods smooth over.

When Upscaling Is Not the Answer

AI upscaling is impressive, but it has hard limits. The algorithm is predicting what higher-resolution detail might look like — it cannot recover information that was never captured. Be honest with yourself about these scenarios:

The source is tiny. A 100 x 100 px thumbnail upscaled to 800 x 800 px will look AI-generated no matter which method you use. Below approximately 500 px on the long edge, upscaling produces more hallucination than recovery.

You need pixel-accurate detail. Legal documents, medical imaging, scientific photography, evidence photos — any context where fabricated detail is a liability. AI upscaling invents plausible detail. Plausible is not accurate.

The image is heavily compressed. A JPEG saved at quality 10 has already destroyed most of the information. Upscaling amplifies compression artifacts (blocking, ringing, color banding). Garbage in, larger garbage out.

You can reshoot. If the subject is still available and you can take the photo again at the resolution you need, that will always beat upscaling. A real 4000 px photo contains real detail. An upscaled 4000 px photo contains predictions.

You just need a different size for web. If you are resizing images for a website, email, or social media — where the output is displayed at screen resolution and does not need enhanced detail — you do not need Photoshop's AI upscaling at all. A standard resize does the job faster. Pixotter's resize tool handles this in your browser in seconds, no Photoshop required.

Quick Resize Without Photoshop

Not every resize needs AI upscaling. If you are preparing images for the web — blog headers, social media posts, email banners, e-commerce thumbnails — you need fast, accurate resizing without the overhead of opening Photoshop.

Pixotter resizes images directly in your browser using WebAssembly. Drop your image, pick your target dimensions (or choose a preset for Instagram, Twitter/X, LinkedIn, etc.), and download the result. Processing happens client-side, so your images never leave your device.

It also handles compression, format conversion, and background removal in the same pipeline. One upload, multiple operations, one download.

When to reach for Pixotter instead of Photoshop:

Frequently Asked Questions

Does upscaling in Photoshop actually add detail?

Yes, but with a caveat. Photoshop's AI models (Super Resolution, Preserve Details 2.0, Neural Filters) predict what higher-resolution detail would look like based on patterns learned from millions of images. The added detail is convincing and usually indistinguishable from real captured detail at normal viewing distances. At extreme zoom, trained eyes can spot AI-generated textures. For print, web, and social media use, the results are production-ready.

What is the maximum upscale Photoshop supports?

Super Resolution caps at 2x linear (4x total pixels). Preserve Details 2.0 has no hard cap — you can enter any dimensions, though quality degrades past 3-4x. Neural Filters support up to 4x. For extreme enlargements, stack methods: Super Resolution first (2x), then Preserve Details 2.0 for the remaining factor. See our guide on resizing without losing quality for more techniques.

Can I upscale a JPEG, or does it only work with RAW files?

All three methods work with JPEG, PNG, and TIFF files. Super Resolution was originally RAW-only but Adobe expanded support to JPEG, PNG, and TIFF in Photoshop v24.x and later. That said, RAW files produce better results because they contain more color data and have no compression artifacts for the AI to amplify. Learn more about image resolution and how it affects upscaling outcomes.

Is "Enhance" in Camera Raw the same as Super Resolution?

Yes. When you right-click an image in Camera Raw and select Enhance, the Super Resolution checkbox in the resulting dialog is the feature. "Enhance" is the menu label; "Super Resolution" is the technology name. They are the same thing.

Does upscaling increase DPI?

Upscaling increases pixel dimensions. DPI (dots per inch) is a print setting that maps pixels to physical size. If you upscale a 3000 x 2000 px image at 300 DPI to 6000 x 4000 px and keep DPI at 300, your print size doubles. If you keep the print size the same and let DPI increase, you get 600 DPI at the original physical size. You control this relationship in Image > Image Size by unchecking Resample and adjusting the DPI field. Check out our guide on how to increase image DPI for detailed steps.

Should I upscale before or after editing?

Upscale first. Edits like sharpening, noise reduction, and color grading work better on higher-resolution files because they have more pixel data to work with. If you sharpen first and then upscale, the AI may amplify sharpening halos. If you upscale first and then sharpen, you get cleaner results.

Wrapping Up

Photoshop 2025 (v26.x) gives you three distinct paths for upscaling: Super Resolution for the best quality on photos, Preserve Details 2.0 for precise dimension control, and Neural Filters for heavy enlargements and face recovery. Pick the method that matches your source material and target use case — and remember that reshooting beats upscaling every time you have the option.

For everyday resizing that does not need AI enhancement — web images, social media, email headers — skip Photoshop entirely. Pixotter's resize tool handles it in seconds, right in your browser. If you are working in Photoshop regularly, our Photoshop resize guide covers non-AI resizing workflows in detail, and our photo enlarger guide explores dedicated enlargement tools beyond Photoshop.