See your options side by side
Choose one task and 2–4 tools. Compare pricing, output, test results and trust.
Test scope & comparison notes
These are recorded facts, not a unified ranking. Tested tiers and scope are not consistently recorded. Read pricing with its date and timing with its conditions. Privacy observations reflect visible flows and statements, not a security audit.
Cross-task capability coverage across selected products
Each product is scored independently per task edge; switch tasks below while keeping shared products selected| Task edge | Leonardo.Ai | Inpaint-web | Switch / compare |
|---|---|---|---|
| Background removal | 86/100 2048 × 2048 · PNG | Not tested in topic | Compare in task topic → |
| Image upscaling | 78/100 1024×1024 (2.0x · Unwatermarked high-quality JPG · 271.9 KB) · Historical description; output not attached | Not tested in topic | Compare in task topic → |
| Object removal | 88/100 2048×2048 (Native full-resolution watermark-free PNG) · Historical description; output not attached | 83/100 2048×2048 (Full-Res Unwatermarked PNG) · Historical description; output not attached | Active task |
| Image generation | 88/100 1024×1024 (Unwatermarked JPEG, 147.5KB, 100% pixel retention) · Historical description; output not attached | Not tested in topic | Compare in task topic → |
Select at least 2 tools to compare
You can also filter a task page first, then compare your shortlist.
Browse this task →| Criterion | Leonardo.Ai Canvas (Inpaint / Outpaint) Object removal 88/100 Seven-aspect task score Free credits deliver full size while preserving unmasked areas well; login and stepwise window alignment on large images add effort and waiting. Tested export 2048×2048 (Native full-resolution watermark-free PNG) · Historical description; output not attached Historical test · 2026-10-16 | Inpaint-web (Open-Source Browser WebGPU Inpainting) Object removal 83/100 Seven-aspect task score Free open-source, login-free and ad-free on-device processing exports full size; complex bookshelf repairs show blocky texture and need inspection. Tested export 2048×2048 (Full-Res Unwatermarked PNG) · Historical description; output not attached Historical test · 2026-10-16 |
|---|---|---|
| Pricing & rules | ||
| Pricing plans | Web individual and default three-seat team plans; excludes API · 2026-10-20 Plan · source date Monthly / one-time Annual bill / period Quota & limits Essential2026-10-20 ↗ 2026-10-20 ↗ USD 12/month USD 120/year Monthly 8,500 Fast Tokens/month; token bank 25,500 Annual 8,500 Fast Tokens/month Rules [1] Premium2026-10-20 ↗ 2026-10-20 ↗ USD 30/month USD 288/year Monthly 25,000 Fast Tokens/month; token bank 75,000 Annual 25,000 Fast Tokens/month Rules [1] Ultimate2026-10-20 ↗ 2026-10-20 ↗ USD 60/month USD 576/year Monthly 60,000 Fast Tokens/month; token bank 180,000 Annual 60,000 Fast Tokens/month Rules [1] 10 plan / billing-period records, all listed; — means that period is not recorded. The app payment panel confirms personal annual prices of $120/$288/$576, differing from earlier rounded monthly equivalents on the marketing site. Public plans permit paid users to buy top-ups without quoting pack prices or per-removal token cost. Unlimited relaxed generation does not establish unlimited removal. Verification has gaps; unknown prices and quotas are not inferred. Cancellation, sources & verification recordSelf-serve cancellation in account settings Official plans and usage terms ↗ · 2026-10-20 Capture 1 ↗ · Capture 2 ↗ Official app annual payment selection ↗ · 2026-10-20 Capture 1 ↗ · Capture 2 ↗ · Capture 3 ↗ | Inpaint-web in-browser open-source object removal and upscaling tool · 2026-10-21 Plan · source date Monthly / one-time Annual bill / period Quota & limits Inpaint-web Open-Source Free2026-10-21 ↗ Free Free Free Free open-source tool without login; unlimited local in-browser object removal and 4x upscaling with full-resolution unwatermarked export. 1 plan / billing-period records, all listed; — means that period is not recorded. Open-source free web tool (MIT License) powered by client-side WebGPU and WebAssembly. Models run locally in the browser with no registration, paid plans, or usage limits. Cancellation, sources & verification recordCompletely free open-source project with no accounts or billing. Inpaint-web official web workspace ↗ · 2026-10-21 Capture 1 ↗ |
| Free & trial allowance on the test date | 150 free daily tokens; each 1024×1024 Canvas pass costs 4 tokens and exports 2048×2048 watermark-free PNGs | 100% free and open-source; unlimited local browser inference and full-resolution exports |
| Cancellation & refunds | Self-serve cancellation in account settings | Completely free open-source project with no accounts or billing. |
| Disclosure of costs & limits | Top-right badge shows live token balance (150 -> 130) and Generate button shows exact cost (4 tokens) | Completely free and open-source with zero hidden charges |
| Task quality | ||
| Measured fixture results | Not verified | Not verified |
| Fixture findings | Delivered full size with surrounding areas retained; erased regions were regenerated by the model. | Processes locally with full-size export; smooth areas worked better than complex bookshelf texture. |
| Original historical observation | Leonardo.Ai Canvas uses layer-based Erase + Inpaint/Outpaint so non-erased pixels remain at their exact 2048×2048 values (unmasked_psnr=31.25dB, #1 overall) while Leonardo Diffusion XL fills the erased holes with detailed books and plants. | Inpaint-web runs the MIGAN inpainting model locally in the browser via ONNX Runtime Web, erasing all 3 signs directly on the 2048×2048 image and exporting a full-size PNG without uploading a single byte to any server. Smooth wood surfaces restore cleanly, while complex bookshelves show lightweight client-model texture patterns. |
| Suitable uses · historical assessment | Designers who want 100% lossless preservation of unmasked pixels (31.25dB) on 2048×2048 images with prompt-guided hole filling | Privacy-conscious users who require 100% local in-browser processing (zero cloud upload) and unlimited free 2048px+ unwatermarked exports |
| Result & output evidence | No native output obtained; the preview does not establish final export quality. | No native output obtained; the preview does not establish final export quality. |
| Delivery capabilities | ||
| Recorded export specifications | 2048×2048 (Native full-resolution watermark-free PNG) · Historical description; output not attached · Check watermark record No native output is attached; check the evidence for specification and watermark claims. | 2048×2048 (Full-Res Unwatermarked PNG) · Historical description; output not attached · Check watermark record No native output is attached; check the evidence for specification and watermark claims. |
| Historical specification note · not an official maximum | 2048×2048 (Native full-resolution watermark-free PNG) | 2048×2048 (Full-Res Unwatermarked PNG) |
| Watermark observation | 100% Watermark-Free | None (100% watermark-free) |
| Batch & API | Supports multi-candidate selection, Inpaint Strength slider, and native 2048×2048 artwork export | Supports multi-step history snapshots, hold-to-compare original, and built-in local 4x upscaling |
| Recorded free-tier dimensions | Not verified | Not verified |
| Attached file dimensions | Not verified | Not verified |
| Speed & reliability | ||
| Timing observation | Network-observed server wait: 9.5s · End-to-end reference: 28.5s | Recorded local processing time: 13.8s · End-to-end reference: 13.8s |
| Date & fixture | 2026-10-16 · anchor_inpaint_v1 · US-West (California, AS2914) | 2026-10-16 · anchor_inpaint_v1 · US-West (California, AS2914) |
| Repeated tests & reliability | No standardized repetition count, success rate or variability range is recorded. | No standardized repetition count, success rate or variability range is recorded. |
| Interaction experience | ||
| Account requirements | Account required in this test | No login in this test |
| Sign-in methods | Free login via Google, Apple, Microsoft, or Email | No registration or login required; works immediately in browser |
| Manual adjustment | Manual erase / adjustment recorded in the historical flow | Not verified |
| Interaction observations | Provides a pro infinite canvas (Canvas Editor) with Erase (hole outpainting) and Draw Mask (masked inpainting) modes; 2048×2048 images require aligning the 1024×1024 generation window across zones (02_workspace.png). | Minimalist open-source workspace with zero popups, ads, or email gates; automatically preserves step-by-step history thumbnails along the top bar (02_workspace.png, 03_download.png). |
| Background & reputation | ||
| Service lifecycle notice | Not verified | Not verified |
| Operator | Leonardo Interactive Pty Ltd (Canva Group) | Inpaint-web Open Source Project (lxfater) |
| Recorded origin | Sydney, Australia | Open-Source Developer Community (GitHub: lxfater/inpaint-web) |
| Domain history record | 2022–Present | 2023–Present (Flagship WebGPU client-side AI vision project) |
| Background & reputation assessment | Developed by Leonardo Interactive Pty Ltd (Sydney, Australia; acquired by Canva in 2024). | Popular GitHub open-source project lxfater/inpaint-web powered by ONNX Runtime Web / WebGPU with fully auditable MIT-licensed source code. |
| External reputation record | Not verified | Not verified |
| Engineering & privacy | ||
| Processing location | Uploaded for cloud processing | On-device processing |
| Third-party tracking record | Not verified | Not verified |
| Engineering & privacy assessment | Multi-layer canvas compositing replaces only the erased holes while leaving non-erased pixels 100% untouched at 2048×2048 (unmasked_psnr=31.25dB, #1 in category). | 100% local browser-side WebGPU/WASM execution: once the ONNX model is cached in IndexedDB, all pixel processing happens inside local browser memory with zero server uploads. |
| Retention statement | Explicit retention period unconfirmed; inspect policy record | This flow runs locally; post-close deletion not independently verified |
| Evaluation basis | ||
| Pricing transparency | 29 / 35 Requires a free login; free accounts receive 150 daily resetting tokens (01_landing.png), each 1024×1024 Canvas inpaint pass costs just 4 tokens, and clicking Download Artwork on the left toolbar exports the full 2048×2048 watermark-free PNG for free (03_download.png). | 34 / 35 100% free open-source tool (01_landing.png): zero registration, zero login, unlimited runs, and no subscription paywalls. Clicking Download exports a native 2048×2048 unwatermarked PNG (03_download.png). |
| Interaction restraint | 26.5 / 30 Provides a pro infinite canvas (Canvas Editor) with Erase (hole outpainting) and Draw Mask (masked inpainting) modes; 2048×2048 images require aligning the 1024×1024 generation window across zones (02_workspace.png). | 29 / 30 Minimalist open-source workspace with zero popups, ads, or email gates; automatically preserves step-by-step history thumbnails along the top bar (02_workspace.png, 03_download.png). |
| Background & reputation | 18 / 20 Developed by Leonardo Interactive Pty Ltd (Sydney, Australia; acquired by Canva in 2024). | 14 / 20 Popular GitHub open-source project lxfater/inpaint-web powered by ONNX Runtime Web / WebGPU with fully auditable MIT-licensed source code. |
| Engineering & privacy | 14 / 15 Multi-layer canvas compositing replaces only the erased holes while leaving non-erased pixels 100% untouched at 2048×2048 (unmasked_psnr=31.25dB, #1 in category). | 14 / 15 100% local browser-side WebGPU/WASM execution: once the ONNX model is cached in IndexedDB, all pixel processing happens inside local browser memory with zero server uploads. |
| Task quality | 57.3 / 65 Occluded Region GT Restoration (25.3/30: structural PSNR=15.54dB · SSIM=0.5344) · Unmasked Preservation (20.0/20: unmasked_psnr=31.25dB #1 in category) · Texture Coherence (12.0/15: rich bookstore shelves, hanging vines, and counter plants) | 49.5 / 65 Masked GT Reconstruction (21.5/30: structural PSNR=14.63dB · SSIM=0.4166) · Unmasked Fidelity (17.5/20: zero pixel alteration outside brush strokes) · Texture Continuity (10.5/15: smooth wood table restoration, though complex bookshelf spines show stylized MIGAN block patterns) |
| Deliverable specifications | 20 / 20 Free direct 2048×2048 native full-resolution watermark-free PNG via Download Artwork (20/20) | 20 / 20 Exports native 2048×2048 full-resolution unwatermarked PNG without login (20/20) |
| Processing speed | 10.5 / 15 ~9.5s per 1024×1024 Leonardo Diffusion XL canvas pass (~28.5s across 3 zones) | 9.5 / 15 Client-side ONNX/WASM 2048×2048 local inference takes ~13.8s per stroke with 0.0s network upload overhead |
