EditTextImage

What is EditTextImage?

EditTextImage is an online AI tool for changing text on an image that already exists. Instead of reopening a design file, identifying the typeface and rebuilding a layer, a user uploads the finished picture, types the words currently in it and the words that should replace them, and the tool redraws only those glyphs — matching the surrounding font, color, size, perspective and background.

The site is operated by SparkTech and went live on 2026-04-16. The team describes it as an independent SaaS built by design-tool and applied-AI engineers, positioned deliberately narrowly: it edits the type on a final image and leaves the rest of the picture alone. Adjacent jobs — background removal, object deletion, photo restoration, colorization, style changes and upscaling — are handled by the team's sister product, SparkPix. EditTextImage claims 133,000+ users across 200+ countries.

How EditTextImage works

The pipeline runs server-side and is described in three stages:

1. Glyph segmentation. A vision model identifies which pixels belong to the target text, including drop shadows, strokes, foil effects and gradients, without touching surrounding content.
2. Background reconstruction. Generative inpainting rebuilds the texture, photography or illustration underneath the old text so the area looks as though it never carried type.
3. Typography-matched re-render. The new text is drawn in a font matched to the sampled typography, including paid fonts the user does not own, hand-lettering and stylized display type. No font installation is required.

The editor uses a Gemini-family vision model with a custom inpainting pipeline trained on typographic preservation rather than free-form generation, with inference routed through apimodels.app. Uploads accept PNG, JPEG or WebP up to 10 MB; the browser only sends the source image and downloads the result. The stated flow is roughly 30–60 seconds. A sign-in is required to generate.

Main features

• Text replacement with typography preservation across posters, screenshots, packaging, signs, menus, product photos and memes.
• Fix garbled text, handling up to six text boxes per run.
• Translate text in image, charged per language.
• Remove text from an image, and blur or pixelate, the latter free in the browser.
• Output tiers: free renders at 1K, standard edits delivered at 2K, and a Pro edit rendered natively up to 4K.
• Dedicated modes for screenshots, font matching and game fonts.
• Commercial use included, credits that never expire, and automatic refunds for failed generations.
• Mobile apps for iOS and Android, plus a multilingual interface (English, Spanish, Portuguese, Japanese, Korean, Traditional Chinese, Russian, German and French).

Pricing

Pricing is credit-based and sold as one-time packs in USD — there is no subscription. Credits do not expire and unused credits are refundable within 14 days of purchase. Billing runs through PayPal or Stripe.

Plan	Price	Credits	Effective rate
Starter	$14.90	40	$0.37 per edit
Pro	$26.90	100	$0.27 per edit
Studio	$39.90	180	$0.22 per edit

All three plans include 2K HD output (free renders at 1K), commercial use, non-expiring credits and auto-refunds on failed edits. On the consumption side, the first edit on a new account is free at 1K without a watermark; a standard edit costs 1 credit, a Pro edit 1.5 credits, fixing garbled text 1 credit, translation 1 credit per language, and text removal 3 free per day before reverting to 1 credit. Custom pricing is available for volumes above 1,000 credits.

Common use cases

The documented examples cluster around finished assets whose source file is unavailable or inconvenient to reopen: event programs where a speaker changed or a name was misspelled; poster and flyer series that share one design; app screenshots whose copy was rewritten after export; e-commerce product photos with a changed discount; social ad headlines being A/B tested; and event artwork where only the date moved. Menu edits, banner ads, YouTube/TikTok/Instagram thumbnails, App Store screenshots and correcting typos in AI-generated images are also listed.

Limitations

EditTextImage is explicitly out of scope for documents used to deceive. IDs, passports, driver's licences, visa support letters, official seals, bank statements, currency, tampered receipts, fabricated chat or email screenshots, and deceptive transcripts or certificates are not processed; such uploads are flagged and blocked, and violations lead to account termination without refund. Request logs are preserved and may be shared with law enforcement. Users must confirm they own or are licensed to edit the image and that the edit is not intended to deceive a third party.

Practical constraints: edits require an account, uploads are capped at 10 MB in PNG, JPEG or WebP, free renders are limited to 1K, and higher resolution or volume consumes credits. The tool only handles words on an image — anything beyond that belongs to SparkPix. Beyond the stated 14-day refund window and the listed credit allowances, no further guarantees are published on the pages reviewed here.

Summary

EditTextImage targets a narrow, common problem: one line of text on an already-exported image needs to change. It answers that with typography-matched AI inpainting, paid for through non-expiring credit packs rather than a subscription, and draws a hard line at fraudulent document editing.