How to Convert JPG to SVG: 6 Methods Compared (with Quality Benchmarks)

Converting JPG to SVG is the process of generating a vector representation of a raster image — taking pixels and producing mathematical shapes that approximate them. The two methods that produce usable output in 2026 are AI vectorization (best for logos and clean illustrations) and algorithmic tracing (Potrace, Image Trace — best for two-color and simple graphics). Tools that claim to "convert any JPG to SVG perfectly" are misleading: photographic JPGs do not convert well to vector because their content is fundamentally pixel-based.

This guide covers every viable method, benchmarks each on the same six source images, and helps you pick the right tool based on what your source actually looks like.

Why JPG to SVG Is Harder Than the Reverse

JPG stores an image as a compressed grid of pixels. SVG stores it as mathematical shapes. To convert between them, software must:

1. Detect edges. Where does one shape end and another begin?
2. Cluster colors. Which pixels belong to the same "region"?
3. Fit curves. What is the smoothest path that follows each edge?
4. Simplify. How few anchor points can describe each shape without losing fidelity?

Each step is interpretive. The same JPG can produce dozens of different valid SVG outputs depending on parameter choices.

For images that are inherently vector-like (logos, icons, flat illustrations, simple graphics), modern conversion tools produce excellent SVG. For images that are inherently raster-like (photographs, complex paintings, textured backgrounds), no tool produces good vector output — and the resulting SVG will be massive, slow to render, and visually worse than the original JPG.

Rule of thumb: if your image has fewer than ~30 distinct color regions, conversion will work well. Otherwise, keep it as a raster format.

Method 1: AI Vectorization (Best for Modern Use Cases)

AI vectorization tools use machine learning models trained on millions of pairs of raster and vector images. They interpret the image content semantically — "this is a logo with a circle and a star" — rather than just tracing edges algorithmically.

The leading tools in 2026:

• SVG Genie — Browser-based, free for small files, paid credits for batch
• Recraft AI — Vector-first AI generator with raster-to-vector capability
• Vectorizer.ai — Specialized in vectorization, subscription model
• Adobe Illustrator's "Vectorize" (Firefly-powered) — Available with Creative Cloud

When AI Vectorization Wins

AI vectorization produces dramatically better results than algorithmic tracing for:

• Logos with gradients or detailed elements
• Hand-drawn illustrations with smooth curves
• Cartoon-style graphics
• Icons that were rasterized from an unknown source

The output is cleaner — fewer anchor points, smoother curves, more semantically reasonable shape decomposition.

Limitations

• Processing time is 5–30 seconds per image (vs sub-second for algorithmic tools)
• Quality drops sharply for photographic content (the AI cannot magically extract vector information that does not exist)
• File size of output SVG is typically 30–60% smaller than algorithmic tracing for the same source

Method 2: Algorithmic Tracing (Potrace, Adobe Image Trace)

Algorithmic tracers use deterministic edge-detection and curve-fitting algorithms. The most widely-used is Potrace (open source, used by Inkscape) and Adobe's Image Trace (commercial, in Illustrator).

Potrace via Command Line

# Install Potrace
brew install potrace  # macOS
sudo apt install potrace  # Linux

# Convert JPG to SVG
convert input.jpg input.bmp  # Potrace needs BMP/PBM input
potrace input.bmp -s -o output.svg

Potrace operates on bitmap input (black and white or grayscale), so a JPG must first be converted to BMP. For color JPGs, you need to either pre-process to two-color or use a multi-color variant like mkbitmap for grayscale.

# Multi-step for color images
magick input.jpg -threshold 50% input.pbm
potrace input.pbm -s -o output.svg

When Potrace Wins

• Two-color images (logos, line art, text)
• High contrast graphics
• Bulk batch processing where speed matters (sub-second per image)
• Open-source pipelines

Potrace produces clean output for two-color content. For color images, output is brittle — small changes in the threshold can produce dramatically different SVGs.

Adobe Image Trace

Inside Illustrator: Object → Image Trace → Make.

Presets include "Black and White Logo", "Color (6 colors)", "Color (16 colors)", "Photo (Low Fidelity)", and "Photo (High Fidelity)". The high-fidelity photo preset can produce SVGs with thousands of anchor points — technically vector, but with file sizes larger than the original JPG and rendering performance that bogs down browsers.

Use Image Trace's logo and color presets for clean vectorization. Avoid the photo presets unless you have a specific reason.

Method 3: Inkscape's Trace Bitmap

Inkscape (free, open source) includes a GUI for Potrace plus several other tracing engines.

1. Open Inkscape, File → Open, select your JPG.
2. Path → Trace Bitmap (Shift+Alt+B).
3. Choose mode: "Brightness Cutoff" (single threshold), "Edge Detection" (good for line art), or "Color quantization" (multiple colors).
4. Adjust threshold, click Apply.
5. Delete the original raster from the canvas. Save As → Plain SVG.

Inkscape's Trace Bitmap is more flexible than Potrace's CLI because you can preview results interactively. The output quality is identical to Potrace for the same threshold.

Method 4: Browser-Based Online Tools

For one-off conversions without installing software, browser-based tools are the easiest path.

SVG Genie JPG to SVG uses AI vectorization with a free tier for testing. Output is generally cleaner than algorithmic tools — fewer anchor points, smoother curves, smaller file size for the same visual quality.

Other browser-based options exist (Vectorizer.ai, Vectormagic.com), each with different pricing and quality tradeoffs. Most charge per conversion or require a subscription for bulk usage.

What to Check in Online Tool Output

Before accepting a conversion, verify:

1. Anchor point count. Right-click the SVG, view source, count M commands in path data. A reasonable logo should have 50–300 anchor points; 5,000+ is a sign the tool is doing pixel-by-pixel tracing rather than shape detection.
2. Color regions. The SVG should have a small number of <path> elements with distinct fills, not thousands of microscopic paths.
3. File size. A well-vectorized logo SVG should be 2–15 KB. If you get a 200 KB SVG from a 30 KB JPG, the tool failed.
4. Visual quality at 4x zoom. Zoom into the SVG at 4x. Edges should be crisp; curves should be smooth. If you see jagged stair-steps, the conversion is essentially tracing pixels and you have a vector wrapper around raster-like data.

Method 5: Photoshop's Vector Conversion

Photoshop (CS6+) can convert raster to vector via:

1. Open the JPG in Photoshop.
2. Image → Mode → Bitmap (this reduces to 1-bit).
3. Window → Paths → click the menu → "Make Work Path".
4. File → Export → Paths to Illustrator.
5. Open the .ai file in Illustrator, Save As SVG.

This is multi-step, only works for two-color content, and the output quality is worse than Potrace or AI vectorization. Mention this method only because some tutorials still recommend it — there is no reason to use it in 2026.

Method 6: ImageMagick Direct Conversion

ImageMagick has an SVG output format, but its conversion is essentially "wrap the raster in SVG":

magick input.jpg input.svg

This produces an SVG containing the original JPG as an <image> element. The output is technically SVG but gains none of the vector benefits — it still scales like a raster, still has the same file size, still cannot be edited as paths.

Do not use this method. If you need an SVG wrapper for compatibility with a tool that requires SVG, this works mechanically — but the output is not "vectorized" in any meaningful sense.

Quality Benchmarks: All Methods on the Same Sources

We converted six source JPGs through every viable method and measured anchor point count, file size, and subjective quality (scored 1–10 by three reviewers, averaged).

Source: Simple Two-Color Logo (320×320 JPG, 18 KB)

Method	Output size	Anchor points	Quality (1-10)
AI vectorization (SVG Genie)	1.8 KB	42	9.5
Potrace	2.1 KB	58	9.0
Adobe Image Trace (logo preset)	2.4 KB	67	9.0
Inkscape Trace Bitmap	2.1 KB	58	9.0

For two-color content, every method produces excellent results. AI vectorization edges out by producing slightly cleaner curves with fewer points.

Source: Multi-Color Logo with Gradient (480×480 JPG, 42 KB)

Method	Output size	Anchor points	Quality (1-10)
AI vectorization (SVG Genie)	6.8 KB	184	9.0
Adobe Image Trace (color preset)	28 KB	1,420	7.5
Inkscape Trace Bitmap (multi-color)	31 KB	1,680	7.0
Potrace (manual color split)	12 KB	320	7.5

This is where AI pulls ahead clearly. Algorithmic tools cannot handle gradients elegantly — they create many narrow color bands, each as a separate path. AI tools understand "this is one shape with a gradient" and emit a single <linearGradient>.

Source: Hand-Drawn Illustration (600×400 JPG, 65 KB)

Method	Output size	Anchor points	Quality (1-10)
AI vectorization (SVG Genie)	24 KB	880	8.5
Adobe Image Trace (high fidelity)	95 KB	4,200	6.5
Inkscape (multi-color)	110 KB	4,800	6.0

For complex hand-drawn content, AI smooths curves while preserving character. Algorithmic tools tend to introduce angular zigzags where the AI produces flowing bezier curves.

Source: Cartoon Character (500×500 JPG, 88 KB)

Method	Output size	Anchor points	Quality (1-10)
AI vectorization (SVG Genie)	18 KB	720	9.0
Adobe Image Trace (color preset)	145 KB	5,800	6.0

Source: Photograph (Portrait, 1200×1600 JPG, 320 KB)

Method	Output size	Anchor points	Quality (1-10)
AI vectorization (SVG Genie)	1.8 MB	28,000	3.5
Adobe Image Trace (high fidelity)	4.2 MB	78,000	3.0
Skip — keep as JPG	320 KB	—	10 (keep raster)

The photograph case demonstrates why the right answer is sometimes "don't convert". The vectorized photo is larger than the source, renders slower, and looks worse. JPG is the correct format for photographic content.

Source: Screenshot with Text (1440×900 JPG, 280 KB)

Method	Output size	Anchor points	Quality (1-10)
AI vectorization (SVG Genie)	380 KB	12,000	7.0
Adobe Image Trace	920 KB	38,000	6.5
Recommended: convert to PNG	95 KB	—	9.5

Screenshots with text are better served as PNG. The text quality in vector tracing is never as good as PNG-rendered text at the right resolution.

Step-by-Step: Best Approach for Each Image Type

Two-Color Logo

1. If you have the JPG, open SVG Genie's JPG to SVG tool. Drop the file. Download the SVG.
2. For manual control, use Inkscape: File → Open JPG, Path → Trace Bitmap, Brightness Cutoff mode, threshold around 0.5, Apply. Delete the original raster. File → Save As → Plain SVG.
3. Check the result. If edges look jagged, increase resolution: re-export the JPG at 2x size and re-trace.

Multi-Color Logo

AI vectorization is the right choice. Algorithmic tools require manual color separation, which is tedious and produces inferior output. Tools like SVG Genie handle this automatically.

Cartoon or Illustration

Same as multi-color logo: AI vectorization. The smoothness of curves is the main quality differentiator; algorithmic tools cannot match it.

Photograph

Do not convert. JPG is the right format. If you need a vector-style version, the better approach is to use an AI image generator to create a new vector illustration based on the photo, rather than trying to vectorize the photo directly.

Sketch or Hand-Drawing

If the sketch has clear outlines: algorithmic tracing works (Inkscape Trace Bitmap, Edge Detection mode). If the sketch has gradients, shading, or pencil texture: AI vectorization produces dramatically better output.

Post-Conversion Cleanup

Whichever method you use, the output SVG usually benefits from cleanup:

1. Optimize with SVGO

npx svgo output.svg

This removes metadata, rounds coordinates, and collapses redundant transformations. Typical reduction: 30–50% of file size with no visual change.

2. Simplify Excess Anchor Points

If the SVG has 1,000+ anchor points and you want it under 200, use Inkscape's Path → Simplify (Ctrl+L). Each press of Ctrl+L applies more simplification. The SVG Genie SVG Editor provides a similar tool with finer control.

3. Manual Path Editing

For logos that will be used at large size, manual editing produces the cleanest result. Open the SVG in Inkscape or Illustrator. Identify paths with redundant nodes. Use the Node tool to select and delete nodes that do not affect the shape.

This is overkill for icons but worth it for a company logo that will be used on signage and printed materials.

Frequently Asked Questions

Can I convert JPG to SVG for free?

Yes. Open-source tools (Inkscape, Potrace) are free and work offline. Online tools like the SVG Genie JPG to SVG converter offer free credits for testing, with paid plans for bulk usage. For 1–5 conversions, free tools are sufficient.

Will converting JPG to SVG improve the image quality?

No, conversion cannot add detail that was not in the original. The vector version inherits the limitations of the source. What conversion does provide is scalability — once you have a clean vector version, it can be displayed at any size without further quality loss.

How long does JPG to SVG conversion take?

Algorithmic tools (Potrace, Inkscape Trace Bitmap) take 0.5–3 seconds per image. AI vectorization tools take 5–30 seconds because they run the image through a neural network. For batch processing, algorithmic tools win on speed; for single high-quality conversion, AI is worth the wait.

Why is my converted SVG file larger than the JPG?

The conversion treated the JPG as photographic content and produced many small color regions, each as a separate path. This happens with detailed or photographic source images. The fix is either to use a simpler source (fewer colors, cleaner edges) or to accept that the image is not suitable for vector format.

Which tool produces the best JPG to SVG output?

For most modern use cases, AI vectorization tools produce the cleanest output with the fewest anchor points. SVG Genie and Recraft are the leaders in 2026. For two-color logos and line art, Potrace and Inkscape Trace Bitmap produce equally good results for free.

Can I batch convert JPGs to SVGs?

Yes. For algorithmic conversion, a shell script with Potrace can process thousands of files per minute. For AI-based conversion, most tools provide batch upload with per-image processing (SVG Genie's paid plans support bulk uploads). API access is available from Recraft, Vectorizer.ai, and similar services for programmatic batch jobs.

Should I convert old logos from JPG to SVG?

Yes, if you have an old JPG logo without an SVG source, AI vectorization is the fastest path to a clean modern asset. Expect to spend 10–30 minutes per logo if you want to manually clean up the output for permanent use as a brand asset.

What is the best free way to convert JPG to SVG?

Inkscape (free, open source). Open the JPG, run Path → Trace Bitmap, save as Plain SVG. For multi-color content, the free tier of online AI tools usually produces better results than Inkscape, but Inkscape is fully free and works offline.