How we tested image embedding across 7 generators
In May 2026 we picked one 800×800 photograph (a head-and-shoulders portrait against a soft neutral background — the most common real-world image-QR use case) and ran it through 7 generators paired with the same 60-character destination URL. For each generator we tested: maximum supported image size (in percent of code area), image-format support (PNG, JPEG, SVG, GIF), automatic vs manual placement (whether the tool shifts the photo to avoid alignment patterns), and the scan reliability of the final code across an iPhone 15 Pro, Pixel 8, Galaxy S24 Ultra, and iPhone 11 in standard office lighting.
Each code was generated at the maximum image size the generator permitted, with error correction set to level H where the generator allowed manual override. Codes were exported as PNG at the largest size the free tier provided, printed at 1.5 inches square on 80lb matte paper, and scanned five times per phone for a total of 20 scans per generator.
The results below are the throughput numbers from that test. Three generators we will not single out failed silently — they accepted the photo upload, generated a code that looked fine in the preview, but the downloaded PNG had no embedded photo at all (the preview was for show; the export stripped the image to "improve scan rate"). The 7 generators in our shortlist did not pull that move.
Quick comparison: 7 image-embed generators side by side
Headline trade-offs. Per-generator notes follow.
| Generator | Max image size | Auto-shift to preserve patterns | Error correction control | Output format |
|---|---|---|---|---|
| EZQR free / Pro / Max | ~30% (level H) | Yes | Manual override (L/M/Q/H) | PNG + SVG (Pro+) |
| QRCode Monkey free | ~30% | Yes | Manual override | PNG + SVG |
| QR Tiger Premium | ~30% | Yes (image-recognition tuned) | Automatic (forces H with photo) | PNG + SVG |
| Uniqode Pro | ~25% | Yes | Automatic | PNG only |
| Beaconstac (legacy) | ~25% | Yes | Automatic | PNG only |
| QR Code Generator (Bitly) | ~20% | Partial | Limited | PNG only |
| Canva | Manual (design file) | No (designer choice) | Not surfaced | Per-design PNG |
1. EZQR — Free image embed with manual error correction
EZQR handles photo embedding on the free tier. Upload any image (PNG, JPEG, SVG, GIF), drag the size slider up to 30% of code area, the generator auto-shifts the placement to avoid the three alignment patterns that QR codes use for orientation. Error correction defaults to Q for static codes; we recommend manually setting it to H for any photo larger than 20% of the code area.
What works: free tier covers unlimited static QRs with photo embedding, full colour customization on the dark modules (per the rules in our QR colour guide), and PNG export at 1024×1024. The Pro plan at $10/mo adds SVG export and the brand-asset library — useful if your team needs to lock the same locked photo across multiple cards or invitations. Max at $20/mo adds dynamic redirect for photo-embedded codes whose destination might change.
What does not: photo-recognition tuning is manual. EZQR will auto-shift placement but will not detect "this is a face, centre it" or "this is a logo, snap to a corner." For a portrait photo, you adjust the position by eye. SVG export is gated behind the Pro tier; the free tier exports PNG only.
Best for: personal photo embed (weddings, memorials, gift cards), small businesses adding a product photo to a QR code, and anyone wanting fine-grained manual control over image size and placement without paying for AI-assisted features. Detailed in our EZQR vs QRCode Monkey comparison for the static-photo-QR use case.
2. QRCode Monkey — Free, no signup, full image control
QRCode Monkey is the most-used free QR generator for any custom design work including image embedding. The customize panel includes a "Add Logo Image" option that handles both small logos and larger photographs. Image upload supports PNG, JPEG, and SVG. Maximum image area is roughly 30% of the code at error correction level H.
What works: zero cost, no signup, no watermark, no per-code fee. SVG output is real vector. Image placement is automatic with manual override for position fine-tuning. Error correction level is manually selectable; the "high error correction" toggle in the design panel sets it to H.
What does not: no dynamic redirect, no scan analytics, no bulk CSV import. If you need to embed a photo across 500 unique cards each pointing to a different URL, QRCode Monkey is the wrong tool — see our bulk QR generators guide for tools that handle CSV-driven bulk with image embed. For one-off photo QRs (a single wedding card, a single memorial plaque, a single product), QRCode Monkey is the gold-standard free choice.
Best for: one-off personal QRs with a centred photo, where you want zero subscription cost and the visual control of full manual customization.
3. QR Tiger Premium — Image-recognition tuned placement
QR Tiger Premium at $37/mo annual is the most expensive of the photo-embed-capable generators. The differentiator: image-recognition-tuned placement. The generator inspects the uploaded image, detects whether it is a face, a logo, or a graphic, and shifts the placement to match the image type. For face photos, it centres slightly above the geometric centre (to favour the visual centre of the face). For logos, it snaps to a corner-friendly placement that maximizes module preservation.
What works: the image-recognition tuning produces noticeably better visual results for portrait photos compared to manual-placement tools. Error correction is automatically forced to level H when a photo larger than 20% area is embedded — sensible default that prevents the most common photo-embed failure mode. SOC 2 compliance and API access are real for enterprise photo-QR workflows (think 500 employee headshots embedded into 500 vCard QRs).
What does not: $37/mo annual = $444/year upfront. The image-recognition tuning is a nice-to-have, not a must-have — the same visual result is achievable on EZQR or QRCode Monkey with 30 seconds of manual positioning. Annual billing required on most plans. The EZQR vs QR Tiger comparison covers the broader feature-vs-cost trade-off.
Best for: enterprises embedding photos into 100+ QRs at scale (employee directories, product catalogs, attendee badges) where the placement-tuning automation saves real time. Overkill for one-off personal photo QRs.
4. Uniqode (Beaconstac) — Photo embed with team controls
Uniqode Pro at $49/mo annual handles photo embedding alongside team workspace features. Maximum image area is roughly 25% — slightly lower than EZQR or QR Tiger — but the team-shared asset library means a marketing organization can lock the same approved photo (think CEO headshot for executive vCards) across every code in the org.
What works: team permissions, asset library, audit logs, brand-template controls. For a 20-person marketing team that needs governance around which photos can be embedded in QRs, Uniqode is the right tool.
What does not: $49/mo annual ($588/year) is the highest sticker price in the test. Annual billing required. PNG-only export. Image area capped at 25%, which is enough for a small headshot but tight for a full product photo. The EZQR vs Uniqode comparison covers the team-features-vs-cost trade-off.
Best for: marketing teams of 10+ where photo-asset governance matters more than the per-code cost.
5. Beaconstac (legacy URL) — Same product as Uniqode
Worth flagging: many photo-QR searches still land on the older Beaconstac URLs. The product is now Uniqode. The legacy URLs still resolve and redirect; pricing and feature set are identical. No separate feature comparison needed.
We mention it because procurement teams sometimes have "Beaconstac" pre-approved as a vendor and discover after onboarding that the platform rebranded. Update the approved-vendor list to Uniqode if you go this direction.
6. QR Code Generator (Bitly) — Limited image embed
QR Code Generator by Bitly supports image embedding on the Plus plan at $35/mo monthly. Maximum image area is the lowest in the test — roughly 20% of the code area, which produces a smaller, less prominent photo than the other tools deliver.
What works: monthly billing (uncommon for Bitly-owned products). Integration with Bitly link-shortening if you need a tracked short URL behind the QR.
What does not: 20% image cap is restrictive for a personal photo QR where the photo is supposed to be the visual focus. PNG-only export. Error correction level is not exposed in the UI, so you cannot manually force level H. The combination of small image area and limited error-correction control makes Bitly's tool the weakest of the paid options for photo-first QRs. The EZQR vs Bitly comparison covers the broader trade-off including the dynamic-link retention policy issues from our permanent QR code guide.
Best for: existing Bitly users who want a small embedded image as a brand accent on top of a tracked link — and only that combination.
7. Canva — Photo-first design workflow, weak QR fidelity
Canva is not a QR generator with image embed — it is a design tool with a QR feature. The workflow inverts: you start with a Canva design (a wedding invitation, a business card, a product label) and the QR is a component you place inside the design. The "photo" is the design itself, not an image embedded inside the QR.
What works: if the goal is "a complete printable design with a small QR in the corner," Canva is good — the design depth and template library exceed any dedicated QR tool. The QR component itself accepts URL input and generates a basic black-on-white code.
What does not: this is not a QR-with-photo-in-the-middle workflow. Canva's QR component does not accept image embedding inside the QR. If you upload a photo into Canva and place a QR on top of it, the result is "a design with a QR over a photo," not "a QR with a photo embedded in its centre." For the actual photo-in-QR result, the QR component cannot do it; you would generate the photo-embedded QR in a dedicated tool (EZQR, QRCode Monkey) and import the PNG into Canva for the surrounding design layout.
Best for: design-first workflows where the QR is one component of a larger printable asset, not the centerpiece. Pair with EZQR for the QR generation step.
Why photo embedding is harder than logo embedding
A logo embedded in a QR code is usually small (10–15% of code area), vector-clean (SVG with crisp edges), abstract (a wordmark or graphic mark), and brand-controlled (no surprises in the colour or composition). A photo is the opposite on every axis.
Size. A photo that delivers the visual impact users want is 25–30% of the code area. That is the upper bound of what error correction level H can recover. Below 20%, the photo loses prominence; above 30%, the scanner cannot reconstruct enough modules to decode the data.
Raster vs vector. A logo as SVG embeds with no resolution loss. A photo as JPEG or PNG comes in at whatever resolution the source provided. At small print sizes (under 1.5 inch), photos often pixelate visibly while the surrounding modules stay crisp.
Composition unpredictability. A logo is one shape. A photo has variable luminance, multiple colour regions, and edges that may visually bleed into adjacent modules. A face with dark hair against a light background creates a hard transition that the scanner may misinterpret as a module edge if the photo placement overlaps the timing patterns.
Colour interaction. A photo with dark regions extending toward the edge of the embedded image creates ambiguous module reads at the photo border. The fix is to crop the photo with a soft white vignette so the photo-to-module transition is always light. Most generators do not apply this vignette automatically; designers do it manually in Photoshop or Figma before uploading.
The result of all four factors: photo embedding has tighter constraints than logo embedding. The same generator may handle logos perfectly and produce flaky photo-embedded codes. Always test the printed result before committing to a batch — the QR code colour guide covers the test methodology in detail.
Best photo types for QR embedding
After running the test, three photo types deliver reliable embedded results. Three types fail consistently.
Works well: head-and-shoulders portraits with a clean background. A face against a solid colour (studio backdrop, plain wall, soft gradient) preserves a clear edge between the photo and the QR modules. The face is the visual focus and the soft background gives the scanner-friendly light transition at the photo border. Most wedding QRs, memorial QRs, and personal-profile QRs fall into this category.
Works well: single-subject product shots on white. A product photo (a bottle, a watch, a piece of jewellery) on a pure white background embeds cleanly because the white background is identical to the QR's light zone. The transition from photo to module grid is seamless.
Works well: simple graphic illustrations or icons. A flat illustration with clear edges and limited colour palette behaves like a logo for embedding purposes — predictable, vector-like, easy for the scanner to ignore.
Fails consistently: landscape photos with mixed luminance. A sunset, a beach, a mountain scene — the multiple light and dark regions extending to the photo edge create ambiguous module reads. Even with error correction level H, scan failures climb sharply.
Fails consistently: photos with dark backgrounds. A photo where the subject is light but the background is dark (a moody portrait, a dimly-lit product shot) creates a dark perimeter around the embedded image that the scanner may misread as part of the QR module grid. The fix is to lighten the photo background in editing before upload.
Fails consistently: low-resolution source images. Anything under 800×800 pixels pixellates visibly at print sizes above 2 inches. The pixellation degrades the visual quality and, more importantly, introduces module-edge ambiguity at the photo border.
The print test before you commit
Photo-embedded QRs are the single most common "looks great in preview, fails in print" failure mode in custom QR work. The preview generator shows the photo at 100% colour accuracy on a perfectly-lit screen; the print produces a CMYK approximation of the colours on a paper substrate that may have texture, tint, or absorption issues. The combination produces failures the digital preview cannot show.
The checklist:
1. Generate the code at the actual final URL. Not a placeholder. Photo embedding interacts with the data payload — the same photo on a longer URL encodes as a denser code with more error correction margin consumed by the data itself.
2. Print the test code at the actual final size on the actual paper stock. A 1.5-inch photo-QR on glossy paper behaves differently from the same code at the same size on uncoated cream stock. Test with the substrate the production batch will use.
3. Scan with three phones at two lighting conditions. Include at least one older phone (iPhone 11 or older Android) — photo-embed failures often show up on older hardware first. Lighting: well-lit indoor and dim indoor. The print production checklist in our packaging guide covers the broader substrate-and-lighting test matrix.
4. Verify with the destination URL change. Scan with each phone and confirm the destination resolves correctly. Photo-embed failures sometimes produce codes that "scan" (the camera detects a QR) but fail to decode (no URL appears) — slightly different failure than "does not scan at all."
If any combination fails, increase error correction to H (if not already), reduce the photo size by 5%, or replace the photo with a higher-contrast or simpler-background version. Reprint one test code and repeat the scan.
The bottom line
For free photo-embed QR codes with full manual control: [EZQR free](/) or [QRCode Monkey free](https://www.qrcode-monkey.com/). Both handle up to 30% image area at error correction level H, support PNG/JPEG/SVG uploads, and produce reliably-scanning codes for portrait, product, and graphic photos.
For automated image-recognition placement at enterprise scale: [QR Tiger Premium](https://www.qrcode-tiger.com/) at $37/mo annual. Worth the cost if you are embedding photos into 100+ codes (employee vCards, product catalogs) and the placement-tuning automation saves real time.
For team-managed photo libraries: Uniqode Pro at $49/mo annual. Worth the cost only if photo-asset governance matters more than the per-code price.
Avoid: Canva (designs, not photo-embedded QRs), Bitly QR Generator (20% image cap is too restrictive), and any generator that silently strips the photo on download. Always use error correction level H. Always test the printed result before the batch.
For the brand/logo version of this comparison, see our QR generators with logo tested list. For long-term reliability of any photo-embedded code, see the permanent QR code guide — a printed wedding QR needs to keep working five years after the wedding, regardless of which generator hosted the URL.
