# LayerPSD — full text > LayerPSD turns a flat JPG, PNG or WebP into a layered Photoshop file: every element on its own transparent layer, the background filled in behind it, and text you can edit — from $0.071 per layer with credits, no subscription. Short version: https://layerpsd.com/llms.txt --- ## Image to PSD with layers: split any image into editable layers URL: https://layerpsd.com LayerPSD turns a flat JPG, PNG or WebP into a layered Photoshop file: every element on its own transparent layer (Auto splits up to 16 elements per pass and runs extra passes on busy images), the background filled in behind it, and text you can edit — from $0.071 per layer, no subscription, and one free try when you sign in with a Gmail account. A busy poster on Auto usually comes back as about 11 images, so 11 credits: about $0.78–$1.09, depending on the pack. Q: How can I convert a picture to layers? A: Upload the picture here and press Split into layers. The model separates each element onto a transparent layer and fills in the background behind them, then you download a layered PSD. On Seedream 5.0 Flash it costs 1 credit per output image; one run returns the background plus up to 16 layers. Q: Can a JPEG be a layered file? A: No. JPEG and PNG store one flat image, which is why a plain JPG-to-PSD converter gives you a single layer. To get real layers the image has to be taken apart — that is what the AI split does. Q: How do I convert a PNG to a layered PSD? A: Saving or renaming a PNG as .psd gives a PSD with one layer, because a PNG holds one flat image. Upload the PNG here instead: each element (people, products, text, logos) goes onto its own transparent layer, the background behind them is filled in, and you download a PSD that opens in Photoshop, Photopea or Affinity. Social media graphics, banners and thumbnails work the same way. Q: Are these the original layers of the design? A: No. A flat image does not remember how it was built. The layers are rebuilt by the model from what it sees, and areas that were covered are repainted, so fine details can differ from the original file. Q: How many layers do I get? A: Choose Auto to split as finely as possible (up to 16 elements per pass, with extra passes on busy images), or ask for 3, 4, 6 or 8 layers and the model merges the rest into them. Fewer layers cost less and finish sooner. Q: Is it free? A: Sign in with a Gmail account for one free try (a 1K split into background + 4 layers, or one text edit). After that you pay with credits, billed per output image: 1 credit on Flash or 6.7 credits on Pro. Credits come in one-time packs; there is no subscription. Q: Is there a free online tool to separate an image into layers and download a PSD? A: LayerPSD runs in the browser and the first split is free: sign in with a Gmail account for a 1K split into background + 4 layers, with PSD and ZIP export. After that each output image costs 1 credit on Flash. If it has to be free every time, Photopea is a free online editor that saves PSD files, but there you cut each layer out by hand. Q: Is LayerPSD an AI layer splitter or layer extractor? A: Yes, that is the same job under another name: an AI layer splitter, layer extractor or image layer separator takes one flat picture apart into separate transparent layers, and LayerPSD does exactly that, then writes them into a PSD with the background filled in and the text as editable type. Up to 16 elements per pass, 1 credit per output image on Flash. --- ## Edit text in an image and keep the same font URL: https://layerpsd.com/edit-text-in-image Yes, you can change the text in a picture or a screenshot and keep the same font. Upload the image, type what each line says now and what it should say, and LayerPSD removes the old words and draws the new ones in the same place, at the same size, in the same lettering and color, with glow, outline, gradient and shadow carried over. “Same font” means the lettering is matched by eye and redrawn: you do not need the font file or the original design. An edit takes about 25 seconds and uses 1 to 2.5 credits; your first one is free when you sign in with a Gmail account. Q: How can I change text in an image and keep the same font? A: Upload the image, type the current text and the new text, and press Change the text. The model redraws the new words in the same font style, weight, color and effects, in the same place. You do not need to know the font name or have the original design file. Q: Is there a tool that removes text from an image and lets me type new words in the same font and style? A: Yes, this page does both in one step: type what the text says now and what it should say, and the old words are removed and the new ones drawn in their place in the same lettering, size, color and effects. It works on JPG, PNG and WebP, screenshots included, and the first edit is free when you sign in with a Gmail account. Q: Can I replace the text on a poster and match the original font, size, color and placement? A: Yes. Each new line is drawn where the old one was, at the same size and in the same color and lettering, with glow, outline and shadow kept. A much longer replacement is set a little smaller so it fits the same space. Q: How do I edit the text in a flattened poster without rebuilding the design? A: Upload the poster here, type each line as it reads now and what it should say, and the image model redraws the poster with those words changed, in the same lettering, colour and effects, and is told to keep the rest as it was; small details can shift, so compare before you save. You do not need the source file or the font. If you want to edit the poster in Photoshop later instead, split it into layers and export a PSD with editable text. Q: How do I change the date or price on a flyer I only have as a PNG? A: Upload the PNG and add one row per change, for example “AUG 14–16” → “SEP 20–22” and “$3” → “$4”. Both are redrawn in their own lettering in one edit, as in the before/after examples on this page. Q: Does it keep glow, outline and gradient effects? A: Yes, the new words are drawn with the same effects. On the neon festival poster above, the pink glow carried over to the new headline. Very fine textures can come out slightly different, so compare before you save. Q: Does it change the rest of the picture? A: It is asked to change nothing else, and in our tests the rest stays put. Because the model redraws the whole image, tiny details can shift slightly, so compare before and after and keep the original. Q: What if the new text is longer than the old text? A: It keeps the same style but is set smaller so it fits the same place. For the closest match keep replacements about the same length, or split the image into layers and edit the text in a PSD. Q: Is this a photo text editor for JPG and PNG? A: Yes. It edits the words in any JPG, PNG or WebP: photos, screenshots, posters, menus. It is not a text box laid on top; the new words are drawn into the picture in the original lettering, as if they had always been there. To get the text as editable type instead, export an editable text PSD. Q: How do I edit the text in a screenshot? A: The same way as any picture: upload the screenshot (PNG or JPG, from a phone or a desktop), press Read the text for me or type the line as it appears, type the replacement, and press Change the text. The new words are drawn in the same system font, size and colour, so a changed price or button label looks like part of the original screen; see the app checkout example above. Use it on your own screenshots, for a mockup, a demo or a corrected figure, not to make a message or a conversation look like it happened. Q: Can I use it as a screenshot text editor? A: Yes, for text of normal size: a title, button label, menu item or price on an app or web screenshot. Type the old and new words exactly as they appear. Very small interface text is less reliable, and because the whole screenshot is redrawn, fine details elsewhere can shift slightly, so compare before you share it. If you need the screenshot text as editable layers, export an editable text PSD instead. Q: Can it change text in other languages? A: Yes. Latin, Japanese, Korean, Chinese and Arabic text all work; keep each change to a line or two for the best result. Q: Is there a free version? A: Yes. Sign in with Google (a verified Gmail account) and your first text edit, or one layer split, is free. No card is needed. After that each edit uses credits: 1 to 2.5 per image, depending on the model. Q: How much does it cost? A: One edit is one image: 2.5 credits on Nano Banana 2.1, 1.7 on GPT Image 2 or 1 on Seedream Flash. Credit packs start at $9.90 for 100 credits, and your first try is free when you sign in with a Gmail account. --- ## Image to editable PSD with text layers URL: https://layerpsd.com/editable-text-psd Most converters leave text baked into the picture. LayerPSD splits the image first, so every line of text is its own layer — then writes it into the PSD as real type, with its stroke, glow, shadow and gradient as layer styles. Q: How do I make the text in an image editable? A: Split the image so each block of text is its own layer, then export “PSD with editable text”. Every text layer that can be read becomes a Photoshop type layer; the rest stay as pixels. Q: Will it use the exact same font? A: No. It picks a font of the same kind (for example a bold sans or a serif) and keeps the size, color and effects. Swap in the original typeface in Photoshop if you have it. Q: Which text stays as pixels? A: Lettering the model cannot read confidently — heavily stylised logos, text on a curve or in perspective, tiny print — is kept as the original pixel layer, so nothing disappears. Q: Does it work with Photopea or Affinity? A: The PSD is a standard RGB file. Type layers and layer styles open in Photopea; support for editing PSD type varies in other apps, where the pixel copy is always there. Q: How accurate is the text recognition? A: On our 30-design test set (2026-10-05), 99% of 206 lines of regular-size text came out on their own layer with the words read right, and 99.9% of characters were correct. The free try uses a lighter model and scored 97%. On about 6% of lifted lines a trace of the old lettering was still visible in the background and needs a quick touch-up. Per-design scores: layerpsd.com/data/text-benchmark.json. Q: Does it work on text-heavy designs like menus, slides and app screenshots? A: Yes, they are part of the test set: app and game screens scored 168 of 170 lines, slides, menus and infographics 54 of 54 (two runs each). Every line becomes its own type layer, so a 20-line menu opens as 20 editable lines. --- ## Image to image AI: change a photo with a prompt, keep it editable URL: https://layerpsd.com/image-to-image Image to image AI takes a picture you already have plus a short text prompt and returns a new version of that picture: a different background, season, colors, style or an added object, with the composition kept. On this page you upload a JPG, PNG or WebP, type the change, choose Nano Banana 2.1 (default), Nano Banana Pro, GPT Image 2 or Seedream 5.0 Flash, and get the result back as a PNG. Press Split into layers under the result and the edited picture comes back as a layered PSD: every element on its own transparent layer with the background filled in behind it, so the generated image is editable in Photoshop instead of flat. You pay per image with credits, from 1 credit on Seedream 5.0 Flash to 5 on Nano Banana Pro, and the first try is free when you sign in with a Gmail account. Steps: 1. Upload the image: A photo, product shot, poster or AI render: JPG, PNG or WebP up to 50 MB. Or pick one of the sample images. 2. Describe the change, pick a model: One or two sentences on what should be different. The prompt is sent with an instruction to keep everything you did not mention. 3. Download, or split into layers: Compare before and after, save the PNG, or press Split into layers to get every element on its own Photoshop layer and export a PSD. Facts: - Input: One JPG, PNG or WebP up to 50 MB, plus a prompt of up to 600 characters in any language - Output: One PNG. About 2K on Nano Banana 2.1, Nano Banana Pro and Seedream 5.0 Flash; GPT Image 2 at its default size and medium quality - Models and price: Nano Banana 2.1 2.5 credits (default), Nano Banana Pro 5 credits, GPT Image 2 1.7 credits, Seedream 5.0 Flash 1 credit per output image. Credits from $0.071, no subscription - Composition: The model is asked to change only what the prompt names and keep the framing. The whole picture is regenerated, so small details can shift; compare before and after and keep the original - Then layers: Split into layers sends the result to the layer tool on this page: every element on its own transparent layer, the background filled in, text as editable type. 1 credit per output layer on Seedream 5.0 Flash - Time: In the three sample runs on this page (one run each, the latest on 2026-10-11): 45 s on Nano Banana 2.1, 29 s on GPT Image 2, 21 s on Seedream 5.0 Flash. Nano Banana Pro has not been timed - Free try: One free image edit, or one layer split, after Google sign-in with a Gmail account. The free try runs Nano Banana 2.1, GPT Image 2 or Seedream 5.0 Flash; Nano Banana Pro needs credits - Privacy: Images go through our gateway (which keeps no images) to the LaoZhang API and the model. LayerPSD stores no images. Do not upload confidential work - Content: Prompts are screened before the model runs. Edits to real people’s faces, watermark removal and adult content are not supported What it is not (use instead): - Making a brand-new picture from text alone: Use a text-to-image model. This page starts from a picture you upload. - Face swaps or changes to a real person’s face: Not supported. Prompts and outputs are screened; see the acceptable use page. - Removing a watermark or logo from an image you do not own: Not supported. License the image or ask the owner for a clean copy. - Adult or violent content: Not supported. The gateway screens prompts and outputs and such runs are refused. Q: Is image to image AI free? A: One try is: sign in with Google (a verified Gmail account) and your first image edit, or one layer split, costs nothing and needs no card. After that you pay per image with credits: 1 credit on Seedream 5.0 Flash, 1.7 on GPT Image 2, 2.5 on Nano Banana 2.1 or 5 on Nano Banana Pro. Packs start at $9.90 for 100 credits and there is no subscription. Q: What is the best image to image AI? A: It depends on the picture and the prompt, which is why this page offers four models side by side instead of one. Nano Banana 2.1 is the default: on the product graphic below it put the headphones on a marble counter and left all four labels where they were. In our text-edit test on one poster (2026-09-29) GPT Image 2 kept the lettering close to the original, and Seedream 5.0 Flash changed weight and color but is the cheapest. If the change is to the words themselves, the image text editor page compares the models on 26 designs. Nano Banana Pro is the higher-priced Nano Banana model. Because you pay per image, you can run the same prompt on two models and keep the better one. Q: Is this an AI image to image generator or an image editor? A: Both names describe the same thing: a model that takes an existing image and a prompt and generates a new version of it. People call it an image to image generator, an AI image to image editor, image to image AI online, img2img or an AI image generator from image; on this page it is one tool, used in the browser, with the result downloadable as PNG or split into layers. Q: Can I use it online without installing anything? A: Yes. It runs in the browser on a computer or a phone. Upload or paste the image, type the prompt, and download the PNG. There is no app to install and nothing is saved on our servers. Q: Does it keep the composition of my photo? A: The prompt is sent with an instruction to keep the framing and everything you did not mention, and the models generally do. The whole picture is regenerated, though, so small details can shift and faces or text may come out slightly different. Use the Before / After toggle to check, and keep the original file. Q: Can I edit the result in Photoshop? A: Yes. Press Split into layers under the result and the picture is sent to the layer tool on this page: every element lands on its own transparent layer with the background filled in behind it, and regular-size text comes out as editable Photoshop type. Export a layered PSD and move, hide, recolor or retype any part without generating again. If you already have the AI image, the AI image to PSD page does the split on its own. --- ## Translate text in an image and keep the design URL: https://layerpsd.com/translate-text-in-image Yes, you can translate the text in an image and keep the design. Upload the picture and pick a language: LayerPSD reads every line of text, small print included, translates it in context, and lets you correct any line before the image model draws the translation in the same place, in the same lettering style, color and effects. Names of people, bands, brands and places, numbers, prices and currency symbols are left untranslated, and the redraw is told to copy them exactly; the model redraws the whole picture, so check them in the result. A redraw uses 2.5 credits on Nano Banana 2.1, which took 35–75 seconds in our sample runs, or 1 credit on Seedream 5.0 Flash; reading and translating costs a small fraction of a credit. Your first one is free when you sign in with a Gmail account. Steps: 1. Upload and pick a language: A JPG, PNG or WebP poster, flyer, menu, ad, packaging shot or screenshot. Choose the language to translate into. 2. Read and translate: Every line of text is read and translated in context. Check the translations and change any line you want worded differently. 3. Redraw and download: About 35–75 seconds later you get the same picture with the translated text in the same lettering. Compare before and after, then save the PNG. Facts: - Input: One JPG, PNG or WebP up to 50 MB: posters, flyers, menus, ads, packaging, social graphics, screenshots - Translate into: 11 languages: English, Spanish, Portuguese, French, German, Italian, Japanese, Korean, Simplified Chinese, Traditional Chinese, and Arabic - Translate from: Any language with Latin letters, Chinese, Japanese (vertical titles too), Korean or Arabic - Output: The whole picture as a PNG with the translated text, about 2K on Nano Banana 2.1 - Lines: Up to 30 lines per picture, each one editable before the redraw - Price: Reading and translating: a small fraction of a credit. Redraw: 2.5 credits on Nano Banana 2.1 (recommended), 1.7 credits on GPT Image 2, 1 credit on Seedream 5.0 Flash. Credits from $0.071, no subscription - Free try: One free translation, or one layer split, after Google sign-in with a Gmail account. No card - Limits: The whole picture is redrawn, so small details can shift and a vertical title may become horizontal; a word in stylised lettering can come out misspelled and text the reader missed can come out wrong, so check the result before you publish What it is not (use instead): - Reading a sign, menu or label while travelling: Use Google Lens or your phone camera; you do not need a new picture. - Documents with long paragraphs, PDFs or slides: Use a document or PDF translator that keeps the file editable. - A file you will keep translating into many languages: Export an editable text PSD and translate the type layers in Photoshop. - Showing a different product image for each store language automatically: Map the image variants in your shop's translation app or theme; this page makes the translated image files, not the store integration. Q: How do I translate the text in an image and keep the same font? A: Upload the image, choose the language and press Read and translate. Every line is read and translated; correct any line you like, then press Translate the picture. The image model draws each translated line where the original was, in the same lettering style, color and effects. It uses 2.5 credits on Nano Banana 2.1 and your first one is free when you sign in with a Gmail account. Q: Can Google Translate translate the text in an image? A: Yes, for reading: the Images tab in Google Translate and Google Lens show the translation over the picture in a plain font. When you need a poster, flyer or product image to publish, with the translation in the original lettering and layout, use this page. Q: Can ChatGPT translate the text in a picture? A: It can read and translate the words, and its image model can redraw a picture. Here the reading, the translation and the redraw are set up to keep the layout, names and prices, and you see and can correct every line before anything is drawn. Q: Will names, prices and numbers be translated too? A: No. Names of people, bands, brands, products, venues and streets, and every number, date digit, price and currency symbol are left in the original in the translation you review, and the image model is told to draw them unchanged. In the festival sample the artist names stay in English and in the café sample the $3 price stays. Because the model redraws the whole picture, check brand names, model numbers, sizes and prices in the result before you publish. Q: Can I translate a supplier's product images and keep the brand, model number and sizes? A: Yes, and you review every line before anything is drawn. In the product sample on this page the brand, model and filter numbers, the sizes and the wattage stayed as they were, and we corrected the decimal points to German commas in the review boxes before the redraw. Two things to check in the result: very small print that was not read (on our sample, the tiny text on the device display came out garbled), and every number and unit against the original. Q: Which languages can it translate into? A: 11: English, Spanish, Portuguese, French, German, Italian, Japanese, Korean, Simplified Chinese, Traditional Chinese, and Arabic. The text in the picture can be in any language with Latin letters, Chinese, Japanese, Korean or Arabic. Q: Does it work with vertical Japanese or Chinese text? A: Yes. Characters stacked in a vertical title are read as one line and translated as a whole. The translation is drawn in the same spot: stacked letter by letter as in the summer festival sample, turned on its side, or horizontal. Q: Can I change the translation before it is drawn? A: Yes. After Read and translate every line is shown with its translation in a box you can edit, so you can fix a word, shorten a line or keep a phrase in the original language. Q: What if the translation is longer than the original? A: It is set a little smaller so it fits the same space, in the same style. For the closest match, shorten long lines in the translation box before you redraw. Q: Can I translate a menu, flyer, product image or screenshot? A: Yes. Anything with readable text works: posters, flyers, menus, ads, packaging, product images, social graphics and app or web screenshots. Very small print is less reliable. Q: Is it free? A: Sign in with Google (a verified Gmail account) and your first translation, or one layer split, is free, with no card. After that a redraw uses 2.5 credits on Nano Banana 2.1 or 1 credit on Seedream 5.0 Flash, from a one-time credit pack (credits from $0.071). Q: How accurate is the translation? A: It is a machine translation made with the whole picture as context, and you see every line before the redraw. Check names, slogans and anything legal or medical with a native speaker before you print; the picture model can also slip on text it was not given, so compare before and after. Q: Can I get the translated text as editable layers instead? A: Yes. Split the image into layers and export an editable text PSD: every line of regular-size text becomes real Photoshop type that you can translate and restyle there. --- ## AI manga cleaner: remove text from manga speech bubbles online URL: https://layerpsd.com/manga-cleaner LayerPSD's manga cleaner is an online manga text remover for scanlation and translation teams: it reads each page once, finds every line of lettering in the speech bubbles and narration boxes, lifts it off and fills the spot with the bubble's own colour. The art is never redrawn, so line work and screentones outside the text stay pixel for pixel. It handles English and vertical Japanese lettering, takes up to 20 pages at a time and returns PNGs at the original size, for about 0.2 credits a page (credits from $0.071). Sound effects drawn over the art stay in place. The free credit you get when you sign in with a Gmail account covers about 20 pages. Steps: 1. Upload the pages: Drop in one page or a whole chapter: JPG, PNG or WebP, raw scans or digital pages. 2. Clean: Each page is read, and every line of lettering in a bubble or box is lifted off and filled with the bubble colour. About ten seconds a page. 3. Download and typeset: Compare before and after, then download each page as a PNG or the whole batch as a ZIP and letter your translation on top. Facts: - Removes: Lettering inside speech bubbles, thought bubbles, narration boxes and captions on a flat fill - Keeps: All artwork, bubble outlines and tails, screentones outside the text, and lettering drawn over the art (SFX, signs) - Lettering: English and other Latin scripts, Japanese horizontal or vertical (tategaki); Korean and Chinese are read the same way - Batch: Up to 20 pages per run, 25 MB each; download one PNG or all as a ZIP - Output: PNG at the size you uploaded - Price: About 0.2 credits a page on average; busy pages cost a little more. Free try: about 20 pages What it is not (use instead): - Removing sound effects drawn over the art: They need the art behind them redrawn. Use a manual redraw, or split the page into layers on LayerPSD and repaint that part with Fix this layer. - Bubbles filled with screentone or a gradient: The fill is a single colour, so a patterned bubble comes back flat there. Check those pages and touch them up. - Long vertical webtoon strips: Cut the strip into pages first; a very tall image is read at too low a resolution to find small lettering. Q: What is a manga cleaner? A: The person, or the tool, that removes the original text from manga pages so a translation can be lettered in. It is the step between the raw scan and typesetting. This page is the tool: it clears the bubbles on up to 20 pages at once. Q: How do I remove text from manga speech bubbles? A: Upload the pages here and press Clean. Every line of lettering in the bubbles and boxes is lifted off and filled with the bubble colour, and you download the clean pages as PNGs or a ZIP. It costs about 0.2 credits a page. Q: Does it work on Japanese raws with vertical text? A: Yes. Vertical (tategaki) columns are found and cleared the same way as horizontal lines; our Japanese test page came back with all five columns gone and the art untouched. Q: Does it remove sound effects (SFX)? A: No. Sound effects drawn across the artwork stay, because removing them means redrawing the art behind them. Clean them by hand, or split the page into layers here and repaint that area. Q: Is the manga cleaner free? A: The free credit you get when you sign in with a Gmail account covers about 20 pages. After that it is pay as you go: about 0.2 credits a page, with credits from $0.071 and no subscription. Q: Will it change my artwork? A: Only the pixels of the lettering change. No image model repaints the page, so line art, faces and tones outside the text stay exactly as uploaded. Q: Can I clean a whole chapter? A: Yes, up to 20 pages per run, each up to 25 MB. Pages are cleaned one after another and the whole batch downloads as one ZIP. Q: Does it work on manhwa, manhua and colour comics? A: Yes, as long as the bubbles have a flat fill. Cut long webtoon strips into pages first so small lettering is still readable. Q: How do I remove text from manga online? A: Open this page, drop in the pages (JPG, PNG or WebP), press Clean and download the result. Nothing to install: up to 20 pages a run, each coming back as a PNG at the size you uploaded. Q: Does the tool remove the text or repaint the bubble? A: It removes the text. Each line of lettering is lifted off with its own pixels and the spot it covered is filled with the bubble colour; no image model repaints the bubble or the page, so the outline, tail and everything outside the lettering are the original pixels. Q: What is the best manga text remover for a scanlation team? A: One that clears whole chapters without touching the art and costs little per page: here that is up to 20 pages a run, about 0.2 credits a page, with vertical Japanese and Korean webtoon pages handled the same way as English. Sound effects drawn over the art still need a redraw. --- ## What LayerPSD does, in one table - Output: Layered PSD (RGB), ZIP of transparent PNGs, single PNGs, or a PNG of the visible layers - Layers: Every element on its own layer: Auto (up to 16 per pass, extra passes for busy images) or a fixed 3, 4, 6 or 8 - Background behind objects: Filled in (amodal inpainting), so moving or hiding a layer leaves no hole - Text: Every line of regular-size text is read before the split and lands on its own layer as native Photoshop type you can double-click and retype; stroke, glow, shadow and gradient become layer styles. 99% of 206 lines in our 30-design test, with 99.9% of characters right - Change words, keep the font: The model redraws the new words in the same lettering, including custom and hand-drawn type - Every object separately: Multi-object segmentation: people, products, logos, text and decorations each on a layer, named by content - Price: 1 credit per output image on Flash, about 6.7 on Pro; background + 4 layers = 5 credits. Credits from $0.071, packs from $9.90. One free try with a Gmail sign-in. No subscription - Speed and size: 56–138 s per split; input up to 50 MB; output 1K, 1.5K or 2K - Privacy: The PSD is built in your browser. Images go through our gateway (which keeps no images) to the LaoZhang API and the model; LayerPSD stores nothing. Do not upload confidential work --- ## How the split works Layer decomposition: The image goes to Seedream 5.0 with layer decomposition on. The model detects each element (instance segmentation), returns it as its own RGBA image with a bounding box and a name, and returns a background with the covered areas repainted (amodal inpainting). Assembly in your browser: Your browser places each layer at its box on the background and writes the PSD, ZIP or PNG locally. Nothing is assembled on a server. Editable text: Before the split, a vision model reads every line of regular-size text: the words, size, color, stroke, glow, shadow and gradient. Each line is cut out onto its own layer and written into the PSD as native Photoshop type with those effects as layer styles. Headlines taller than 8% of the image stay in the picture so the model can repaint behind them; “PSD with editable text” reads them at export. Scores: layerpsd.com/data/text-benchmark.json. Retyping text: To change words, the model redraws only that text in the same lettering and leaves the rest of the image alone, so it also works for custom or hand-drawn type that no installed font matches. --- ## Doing it by hand in Photoshop, and doing it here - Select each element: by hand in Photoshop: Object Selection Tool, then Select and Mask for hair and soft edges, for every object. In LayerPSD: All elements found in one pass. - Put it on its own layer: by hand in Photoshop: Layer via Copy or a layer mask, one by one. In LayerPSD: Every element already on a transparent layer, named. - Fill what was behind it: by hand in Photoshop: Generative Fill or Content-Aware Fill on each hole, then clean up. In LayerPSD: Background repainted automatically. - Make text editable: by hand in Photoshop: Match Font, then retype each line and rebuild stroke and shadow. In LayerPSD: Export native type layers with effects as layer styles. - Time for a 10-element poster: by hand in Photoshop: Typically 30–60 minutes. In LayerPSD: About 1–2 minutes, 11 credits on Flash. --- ## Is there a free version? Yes, one free try per Google account. Sign in with a verified Gmail account and you get one split (Seedream 5.0 Flash, 1K, background + 4 layers, with PSD and ZIP export) or one text edit for free. No card is needed. After that you pay with credits, from $0.071 per layer, with no subscription. --- ## Remove text from an image, keep the background URL: https://layerpsd.com/remove-text-from-image Upload the picture and press Split into layers. LayerPSD finds every line of text automatically (no brushing), cuts each onto its own layer and fills in what was behind it, so you switch the text layers off and download a clean image instead of a smudged one. It takes about 56–138 seconds at 1 credit per output layer; the flyer on this page (11 layers + background) used 12 credits. Steps: 1. Upload the image: Drop a JPG, PNG or WebP (up to 50 MB), or paste it with Ctrl + V. Nothing to install. 2. Split into layers: Seedream 5.0 finds every element, cuts each onto a transparent layer and fills in the background behind it. About 56–138 seconds. 3. Hide the text, download: Click the eye on each text layer you want gone, then Download composite for a clean PNG, or export a PSD with the text kept as hidden layers. What you get: - A clean image with the text gone and the background filled in behind it - Every line of text on its own layer: remove one line, keep another - The same text layers to move, resize or restyle instead of deleting - PNG, ZIP of layers, or a layered PSD What it is not (use instead): - Watermarks or logos on images you do not own: License the image or ask the owner for a clean copy. LayerPSD is not for this. - Text printed on a product, like a label on a bottle or a mug: It belongs to that object's layer and stays. Change it with Edit text in image. - Changing the words rather than deleting them: Use Edit text in image: it redraws the new words in the same font. - Copying the text out as plain text: Use an OCR tool. LayerPSD returns layers, not a transcript. Q: How do I remove text from an image without messing up the background? A: Split the image into layers first. The text is cut onto its own layers and the area behind it is repainted, so hiding those layers leaves a clean background instead of a blurred patch. Q: Can I delete the text and fill in what was behind it automatically? A: Yes. You do not paint over anything: every line of text is detected, separated and the background under it is filled in. You only choose which text layers to hide. Q: Can I remove just one line and keep the rest? A: Yes. Each line or block of text is its own layer; hide only the ones you want gone. Q: What images does it work best on? A: Designed graphics: posters, flyers, banners, thumbnails, social covers and product images. Very small, dense text can come back grouped as one block. Q: Can it remove a date stamp or a logo from my own photo? A: Yes, on images you own or have the right to edit. In our test (2026-10-05) a camera date stamp and a corner logo on our own product photo each came back as a layer of their own; hiding it left the table and the sunglasses filled in. The objects are redrawn by the model, so fine detail and contact shadows can change slightly. Removing watermarks from images you do not own is not a use LayerPSD supports. --- ## Make an AI-generated image editable in Photoshop URL: https://layerpsd.com/ai-image-to-psd Images from ChatGPT, Midjourney, Gemini or Seedream come out as one flat picture, so fixing one part usually means regenerating the whole thing. LayerPSD splits the finished image into Photoshop layers instead: every element on its own transparent layer (Auto splits up to 16 elements per pass and runs extra passes on busy images), the background filled in behind it, every line of regular-size text as editable Photoshop type (99% of 206 lines in our 30-design test). The illustration on this page was made with GPT Image 2 and split in 138 seconds for 13 credits. Steps: 1. Upload the image: Drop a JPG, PNG or WebP (up to 50 MB), or paste it with Ctrl + V. Nothing to install. 2. Split into layers: Seedream 5.0 finds every element, cuts each onto a transparent layer and fills in the background behind it. About 56–138 seconds. 3. Edit in Photoshop: Export a layered PSD and move, resize, recolor or swap any element without regenerating the whole image. What you get: - Your exact image, in layers, instead of a new generation that changes everything - The background filled in behind each element - Text layers you can retype in the same lettering, or export as editable type - A PSD for Photoshop, Photopea, GIMP or Affinity What it is not (use instead): - Generating the image in the first place: Use a text-to-image model, or change a picture you already have with a prompt on our image to image page. The split starts from a finished image. - Upscaling or restoring the image: Use an upscaler. Splitting keeps the resolution you choose (1K–2K). - Removing AI watermarks or provenance marks: Not a use LayerPSD supports. Leave provenance marks in place. Q: How do I split a Midjourney or ChatGPT image into layers? A: The models only return one flat image. Upload that image here and press Split into layers to get a layer per element with the background filled in, then export a PSD. Q: Can I do it in Photoshop by hand? A: Yes: select each object, mask it onto its own layer, fill the hole with Generative Fill, repeat. For a 10-element image that is typically 30–60 minutes; here it is one pass of about 1–2 minutes. Q: Will it change my image? A: The elements are cut out as they are. The only new pixels are the filled-in areas behind them, which stay hidden until you move something. Q: Can I fix the text the AI got wrong? A: Yes. Retype a text layer and the model redraws the new words in the same lettering, or use Edit text in image on the flat picture. --- ## Turn a flat poster into an editable PSD URL: https://layerpsd.com/poster-to-psd When all you have is the exported JPG or PNG of a poster, flyer, banner or thumbnail, LayerPSD rebuilds it as a layered PSD: headline, subtitle, product, badges and decorations each on their own layer (Auto splits up to 16 elements per pass and runs extra passes on busy images), the background filled in behind them, and text as native Photoshop type layers you can double-click and retype, with stroke, shadow and glow as layer styles. 1 credit per output layer, no subscription. Steps: 1. Upload the image: Drop a JPG, PNG or WebP (up to 50 MB), or paste it with Ctrl + V. Nothing to install. 2. Split into layers: Seedream 5.0 finds every element, cuts each onto a transparent layer and fills in the background behind it. About 56–138 seconds. 3. Update and export: Retype text, swap the product, move the badge, then export a PSD with editable text or a ZIP of every layer. What you get: - Headline, body text, badges, products and decorations on separate layers - Text as native Photoshop type with stroke, glow, shadow and gradient as layer styles - A clean background to reuse for the next version - A real sample PSD to download and check before you pay What it is not (use instead): - The exact original fonts: Fonts are matched by style (serif, sans, script). Pick the exact typeface after opening, or use a font identifier. - Print-ready CMYK files: The PSD is RGB; convert it in Photoshop for print. - Multi-page documents or PDFs: Export one page as an image first. LayerPSD works on single images. Q: I lost the PSD of my flyer. Can I get it back? A: Not the original file, but LayerPSD can rebuild a layered PSD from the exported image, with text as editable layers and the background filled in. Q: Will the text be editable? A: Choose "PSD with editable text" when exporting. Each text layer becomes native type you can double-click and retype, with its effects as layer styles; lettering the model cannot read stays as pixels. Q: Can I change "SALE" to "SOLD" in the same font without Photoshop? A: Yes, with Edit text in image: type the current and new words and the model redraws them in the same lettering. Q: How many layers will I get? A: Auto finds every element, up to 16 per pass, or ask for 3, 4, 6 or 8. --- ## Extract every object from an image as transparent PNGs URL: https://layerpsd.com/extract-objects-from-image A background remover cuts out one subject. LayerPSD separates everything: each object, text block and decoration becomes its own transparent PNG, named by what it is (Auto splits up to 16 per pass and runs extra passes on busy images), plus a background with those areas filled in. Download them as a ZIP, one by one, or as a layered PSD. 1 credit per output image; the desk photo on this page gave 10 cut-outs and a background for 11 credits. Steps: 1. Upload the image: Drop a JPG, PNG or WebP (up to 50 MB), or paste it with Ctrl + V. Nothing to install. 2. Split into layers: Seedream 5.0 finds every element, cuts each onto a transparent layer and fills in the background behind it. About 56–138 seconds. 3. Download the cut-outs: Download a ZIP of numbered PNGs in layer order, a single layer, or the whole set as a PSD. What you get: - Every element as its own transparent PNG, not just the main subject - Files named by content ("perfume bottle", "lilac floral group") - A background plate with the cut-out areas filled in - Split any cut-out again if it still holds several things What it is not (use instead): - Cutting out one person or product: A background remover (remove.bg, Photoshop Remove Background) is faster and often free. - Hair-perfect portrait masks: Use a dedicated matting tool; fine hair and glass edges can be softer here. - Batch processing hundreds of images: The site handles one image at a time. For a large batch, write to hi@layerpsd.com. Q: How do I extract every object from a picture at once? A: Upload it and press Split into layers. Every element is cut onto its own transparent layer; download them as a ZIP of PNGs. Q: How do I put text behind a person or object? A: Split the image here: the person or object comes out on its own transparent layer and the background is repainted behind it. Open the PSD in Photoshop or Photopea, add your text layer between the background and that layer, and the words sit behind the subject. You can still move and restyle the text afterwards. Q: How is this different from a background remover? A: A background remover keeps one subject and deletes the rest. LayerPSD keeps everything, each object separately (multi-object segmentation). Q: Are the PNGs transparent? A: Yes. Each cut-out is an RGBA PNG with a transparent background, cropped to its own bounds. Q: Can I split a cut-out further? A: Yes. Any layer can be split again; the pieces nest under it. --- ## Split a product photo into layers URL: https://layerpsd.com/product-photo-to-psd Upload a product photo (JPG, PNG or WebP) and LayerPSD returns a layered PSD: the product on its own transparent layer, each prop (plant, cup, sunglasses, packaging) on another, and a background plate with every object removed and the surface behind them repainted. Auto splits up to 16 elements per pass. Shadows are not kept as a layer of their own: the product is cut clean and the plate is repainted without them, so add a new shadow in Photoshop once things are in place. 1 credit per output layer on Flash, no subscription. Steps: 1. Upload the image: Drop a JPG, PNG or WebP (up to 50 MB), or paste it with Ctrl + V. Nothing to install. 2. Split into layers: Seedream 5.0 finds every element, cuts each onto a transparent layer and fills in the background behind it. About 56–138 seconds. 3. Restage and export: Move or swap the product, drop props, reuse the empty background plate, then export the PSD or a ZIP of transparent PNGs. What you get: - The product on its own transparent layer, cut without its shadow - Each prop as a separate layer, named by what it is - An empty background plate, repainted where the objects stood - A real product-photo PSD to download and check before you pay What it is not (use instead): - A separate shadow layer: Shadows go with the cut. Add one in Photoshop with a Drop Shadow style or a soft brush on a new layer. - One product on white, nothing else: A background remover is faster for a single cut-out. Use LayerPSD when you also want the props and the background plate. - Glass, liquids and reflections kept exactly: The model repaints what was behind them, so check transparent and mirrored areas and touch them up. Q: Can I get the product, the props and the background on separate layers? A: Yes. Every object in the photo becomes its own transparent layer, named by content, and the background comes back as a separate plate with the objects removed and the area behind them filled in. Q: Is the shadow kept on its own layer? A: No. The product layer is cut without its shadow and the background plate is repainted without it. Add a new shadow in Photoshop once the product sits where you want it. Q: Can I reuse the background for another product? A: Yes. The base layer is the scene with every object removed, so you can place a different product on it and export again. Q: How many layers will I get? A: Auto finds every object, up to 16 per pass, or ask for 3, 4, 6 or 8. The sneaker photo on this page came back as the background plus 6 layers in 79 seconds. --- ## An imagetolayers.com alternative without a subscription URL: https://layerpsd.com/alternatives/imagetolayers imagetolayers.com is a subscription layer splitter: a free layered preview, then Creator at $19 a month ($10 billed yearly) or Studio at $49 ($30 billed yearly), with "Lossless 4K" output and an API on Studio. LayerPSD does the same core job — every element on its own transparent layer (Auto splits up to 16 per pass and runs extra passes on busy images), the background repainted behind it, export as a layered PSD, a ZIP of PNGs or single PNGs — billed per output image with credits: 1 credit per layer on Seedream 5.0 Flash (from $0.071), so background + 4 layers is 5 credits and a busy poster about 11 ($0.78–$1.09). Credit packs are one-time purchases from $9.90. It can also export text as native Photoshop type layers. Pick imagetolayers.com for 4K output, an API or steady daily volume; pick LayerPSD for occasional jobs and editable text. Pick imagetolayers.com if: - You want to look at the layers before paying anything: it offers a free layered preview. - You need output above 2K: it states "Lossless 4K"; LayerPSD outputs 1K to 2K. - You need an API: it comes with the Studio plan. - You split images every day, where a flat monthly plan can cost less than paying per image. Pick LayerPSD if: - You split a few images a month and do not want another subscription: a one-time $9.90 pack (100 credits) covers about 9 busy poster splits or 20 four-layer splits on Flash, and the credits last 12 months. - You need the text editable: LayerPSD can export text as native Photoshop type with stroke, glow, shadow and gradient as layer styles. imagetolayers.com does not say whether its text layers are editable type. - You want to split a layer again: any layer can be split once more and the pieces nest under it. - You want to see real results first: every sample on the examples page comes with its PSD, model, time and cost. Q: Is LayerPSD free like the imagetolayers.com preview? A: Not unlimited. You get one free split when you sign in with a Gmail account (Flash, 1K, background + 4 layers). After that you pay per output image with credits: 1 credit on Flash, about 6.7 on Pro, from $0.071 a credit. There is no plan to cancel. Q: Do both fill in the background behind objects? A: Yes. imagetolayers.com says it reconstructs "plausible hidden areas behind overlaps"; LayerPSD repaints the background behind every element. In both, the filled area is generated by AI, not recovered from an original file. Q: Which gives the higher resolution? A: imagetolayers.com advertises "Lossless 4K". LayerPSD outputs 1K, 1.5K or 2K from inputs up to 50 MB. --- ## A CopySlides alternative for image to PSD URL: https://layerpsd.com/alternatives/copyslides CopySlides is mainly an image- and PDF-to-PowerPoint tool. Its image-to-PSD feature writes text that passes confidence, orientation and layout checks as native Photoshop text layers and keeps the rest as pixels with a warning; its site does not say whether the background behind objects is filled in. Plans are Starter $9 a month, Plus $19 a month or Lifetime $149 once, with PSD conversions not priced separately. LayerPSD puts every element — text, products, people, logos — on its own transparent layer (up to 16 per pass in Auto, with extra passes on busy images), repaints the background behind them and can export the text as native type with its effects as layer styles, at 1 credit per output image on Flash (from $0.071) with no subscription. Pick CopySlides if: - You also need slides: turning images and PDFs into editable PPTX is its main product. - You prefer to pay once: it has a $149 lifetime plan. - You only need the lines of text as type, and the rest of the picture can stay as it is. Pick LayerPSD if: - You need the objects separated too, not only the text: every element lands on its own layer, named by what it contains. - You need to move things: the background is repainted behind every element, so nothing leaves a hole. - Your text has effects: stroke, glow, shadow and gradient come across as Photoshop layer styles. - Your file is large: LayerPSD takes images up to 50 MB; CopySlides states 20 MB and 16 million pixels. - You only need a few files: pay per output image instead of a monthly plan. Q: Is the text editable in both? A: Yes, within limits. CopySlides writes text that passes its checks as native text layers and keeps the rest as pixels. LayerPSD reads each text layer and writes it as native type, with its effects as layer styles, over the original pixels; lettering it cannot read stays as pixels. Fonts are matched by style, so check the result before you send it on. Q: Can I change the words without retyping them in Photoshop? A: In LayerPSD, yes: retype a text layer and the model redraws the new words in the same lettering, including custom or hand-drawn type no installed font matches. A text edit uses 1 to 2.5 credits per image depending on the model. Q: Does LayerPSD make PowerPoint files? A: No. LayerPSD exports a layered PSD, a ZIP of PNGs or single PNGs. For PPTX, CopySlides is the better fit. --- ## An imagetopsd.com alternative with editable text URL: https://layerpsd.com/alternatives/imagetopsd imagetopsd.com sells prepaid credit packs — Starter $4.9 for 200 credits up to Studio $49.9 for 3,000, valid 365 days — at about 10 credits for 2 layers, with 20 free credits (up to 4 layers per run) to start. It puts text regions on separate layers and lists 16 to 50 layers by plan, with background inpainting on paid plans; it does not say whether text becomes editable type. LayerPSD also sells one-time credit packs, from $9.90 for 100 credits and valid 12 months: 1 credit per output image on Flash, so background + 4 layers is 5 credits. Auto splits up to 16 elements per pass and runs extra passes on busy images. It fills in the background on every run, the free try included, and can export text as native Photoshop type with its effects as layer styles. Pick imagetopsd.com if: - You want more layers per run: it lists 16 to 50 by plan; LayerPSD returns up to 16 elements per pass (busy images get extra passes). - You like prepaid credits: its packs stay valid for 365 days. - You want several small splits free: it gives 20 credits at sign-up. Pick LayerPSD if: - You need text you can retype: native Photoshop type layers with stroke, glow, shadow and gradient as layer styles. - You want the background filled in on the free try too: imagetopsd.com lists occlusion fill on paid plans. - You need to change words but keep the lettering: the retype tool redraws the new words in the same style. - Your images are large: up to 50 MB in (imagetopsd.com: up to 30 MB). Q: Are both pay-as-you-go? A: Yes. Both sell prepaid credit packs valid for a year: imagetopsd.com from $4.9 for 200 credits, LayerPSD from $9.90 for 100 credits at 1 credit per output layer on Flash. Neither needs a monthly subscription. Q: Which gives the higher resolution? A: imagetopsd.com states "Lossless 2K" on paid plans; LayerPSD outputs 1K, 1.5K or 2K. Q: Is the text editable? A: imagetopsd.com says text regions go on separate layers but does not say they are editable type. LayerPSD can export them as native Photoshop type layers; lettering it cannot read stays as pixels. --- ## A Canva Magic Layers alternative that exports PSD URL: https://layerpsd.com/alternatives/canva-magic-layers Canva Magic Layers (beta) turns a flat PNG or JPG into editable layers inside a Canva design, text included. It is a Premium AI tool on Canva's shared monthly AI allowance: up to 20 Premium-AI uses on Canva Free and up to 200 on Pro, by Canva's estimate. Canva's download types (JPG, PNG, SVG, PDF and others) do not include PSD, so the layers stay in Canva. LayerPSD does the split outside Canva and gives you the file: every element on its own transparent layer (Auto splits up to 16 per pass and runs extra passes on busy images), the background repainted behind it, exported as a layered PSD, a ZIP of PNGs or single PNGs, with text as native Photoshop type. It costs 1 credit per output image on Flash (from $0.071) with no subscription. Stay with Magic Layers when you finish the design in Canva; use LayerPSD when the layers have to leave Canva. Pick Canva Magic Layers if: - You finish and publish the design in Canva: the layers land straight in the editor you already use. - You already pay for Canva Pro: Magic Layers runs on the AI allowance you have, up to 200 Premium-AI uses a month by Canva's estimate. - You want to retype a line and change its font in the same place you edit everything else. Pick LayerPSD if: - You need a PSD: for Photoshop, Photopea, Affinity or GIMP, or for a designer who works in Photoshop. - You need the background filled in behind every element: LayerPSD repaints it on every split; Canva's pages do not say whether Magic Layers does. - Your image is large or photo-like: LayerPSD takes up to 50 MB, and the sneaker photo on the examples page split into 6 layers and a background in 79 s. Magic Layers takes up to 5,000 px on the longest edge and says photo-realistic images may not separate as cleanly. - You do not otherwise use Canva: pay per output image with credits (about 11 for a busy poster, $0.78–$1.09) instead of a yearly plan (Canva Pro is US$144 a year in the US). - You want to change words but keep the exact lettering: the retype tool redraws the new words in the same style, including custom lettering that no installed font matches. Q: Can Canva Magic Layers export a PSD? A: Canva's download types are JPG, PNG, SVG, PDF, PPTX, DOCX, GIF and MP4; PSD is not one of them (checked 2026-10-05). The layers stay editable inside Canva. To take them into Photoshop, split the image with a tool that writes a layered PSD, such as LayerPSD. Q: How do I convert an image to layers in Canva? A: Select the image in the Canva editor and open Edit, then Magic Layers. It takes a single-page PNG or JPG up to 5,000 px on the longest edge, uses Canva's monthly AI allowance, and the layers stay inside the Canva design. For a layered PSD instead, upload the image to LayerPSD. Q: How do I export a whole Canva design to PSD? A: Canva has no PSD download. Download the design as a PDF with Flatten PDF off and open it in Photopea, which turns each element into a layer and saves a PSD; or download a PNG and split it with LayerPSD. The Canva to PSD guide at layerpsd.com/canva-to-psd walks through both. Q: Is Canva Magic Layers free? A: It runs on Canva's shared monthly AI allowance. Canva estimates up to 20 Premium-AI uses a month on Canva Free and up to 200 on Pro, fewer for complex tasks. LayerPSD gives one free split when you sign in with a Gmail account, then charges 1 credit per output image on Flash, from $0.071 a credit. Q: Can I use LayerPSD layers in Canva? A: Yes. Download the split as a ZIP of transparent PNGs, one per element, and place them in any editor that takes PNG, Canva included. Use the PSD when the design continues in Photoshop. Q: Why is Canva Magic Layers not working or not showing? A: Canva's help page lists the usual causes (checked 2026-10-10): the file is not a single-page PNG or JPG (PDFs, vector graphics and multi-page documents are not supported), the image is over 5000 px on its longest edge, the monthly AI allowance is used up, or the image is photo-realistic and does not separate cleanly. To find it, select the image in the editor and open Edit, then Magic Layers, or start from Magic Layers on the Canva homepage. If it still fails, LayerPSD takes JPG, PNG or WebP up to 50 MB, photos included, and gives you the layers as a PSD or PNG files. --- ## Fix text in AI generated images URL: https://layerpsd.com/fix-text-in-ai-image Yes, you can fix the text in an AI-generated image without regenerating it. Upload the picture, press Read text to see what the model wrote (or type it as it reads), type what each line should say, and the image model removes the wrong letters and draws the correct words in the same place, size, font style and colour, with glow, outline and shadow carried over. The model redraws the whole picture from yours and is told to keep everything else as it is; in our tests the median change outside the edited lines was under one percent, but small details can shift, so compare before and after. An edit takes about 20 seconds on the default model and uses 1 to 2.5 credits; your first one is free when you sign in with a Gmail account. Facts: - Input: One AI-generated JPG, PNG or WebP up to 50 MB, from any generator - Output: The same picture as a PNG with the corrected words, about 2K on Nano Banana 2.1 - What is kept: The composition, colours, faces and everything outside the edited words, as closely as the model redraws them (small details can shift); the lettering style, size, colour and effects of the words themselves, matched by eye rather than with the original font - Changes per edit: Several lines at once, for example a headline and a date - Models and price: Seedream 5.0 Flash 1 credit (fixed every misspelling in our runs), Nano Banana 2.1 2.5 credits (closest lettering, but can read a near-miss as already correct), GPT Image 2 1.7 credits. No subscription - Time: About 20 s on Seedream 5.0 Flash, about 30 s on GPT Image 2, from 20 s to over a minute on Nano Banana 2.1 - Free try: One free edit, or one layer split, after Google sign-in with a Gmail account. No card - Languages: Latin scripts, Chinese, Japanese, Korean and Arabic - Limits: Complete gibberish has no "current text" to match, so type it roughly as it reads and say what it should be; tiny print is less reliable; the whole picture is redrawn, so compare before you save Sample runs (Seedream 5.0 Flash; the misspellings were asked for in the generation prompt): - "GRAND OPENNING" → "GRAND OPENING"; "FRESH BREAD EVRY MORNING" → "FRESH BREAD EVERY MORNING": 20 s, 2026-10-11 - "CAPUCCINO" → "CAPPUCCINO"; "BLUBERRY MUFFIN" → "BLUEBERRY MUFFIN": 22 s, 2026-10-11 Q: How do I fix the text in an AI-generated image? A: Upload the image here, press Read text so the model lists each line as it reads, type the correct wording next to any line that is wrong, and press Fix the text. The image model removes the wrong letters and draws the right words in the same place, in the same lettering, colour and effects. You get your picture back with the text corrected; it is redrawn from your image, not generated again from a prompt, so check small details before you save. Q: Why do AI-generated images have weird or misspelled text? A: Image generators draw letters as shapes rather than spelling words, so a plausible-looking shape in the right spot satisfies them. Short, common words usually come out right; long words, invented names, numbers and small print are where the mistakes land. Newer models make fewer of them, but a wrong date or a doubled letter still slips through. Q: How can I correct the spelling in an AI image without regenerating it? A: Use a text edit instead of a new generation. On this page you type the misspelled line and the correct line, and the model is told to change only those letters; in our tests the composition, colours and the rest of the picture stayed put, with small details sometimes shifting. Compare before and after. Regenerating, even with the text in quotes, gives you a different picture each time. Q: What if the text is complete gibberish? A: Type it roughly as it reads (Read text will give you the model's best guess) and put the words you want in the second box. The edit replaces whatever is in that spot with the right words in the same style. If the gibberish is long and the replacement is short, the new line is set at the same size and centred where the old one was. Q: Does it keep the same font and effects? A: The new letters are drawn to match the existing lettering by eye: style, weight, colour, size and position, plus glow, outline, gradient and shadow. In our 28-design text-edit test the lettering carried over in 27 of 28 before/after crops on Nano Banana 2.1 and in 25 of 28 on Seedream 5.0 Flash. You do not need to know or own the font. Q: Will the rest of the picture change? A: It is asked to change nothing else, and in our tests the rest stays put; the median share of the picture that changed outside the edited lines was under one percent. Because the model redraws the whole image, tiny details can shift slightly, so compare before and after and keep the original. Q: Does it work with Midjourney, DALL·E, Gemini, GPT Image and Flux pictures? A: Yes. The fix is done on the exported picture, so the generator does not matter. Upload the JPG, PNG or WebP you downloaded from Midjourney, ChatGPT, Gemini, Nano Banana, GPT Image, Seedream, Flux, Ideogram, Stable Diffusion, Canva, Firefly or Leonardo. Q: Can I fix AI image text on my phone or iPhone? A: Yes. The page works in the phone browser: upload from your camera roll, type the correction and download the PNG. No app is needed. Q: Which model should I pick for spelling fixes? A: Start with Seedream 5.0 Flash (1 credit). On the two posters above it corrected every misspelled line in one pass. Nano Banana 2.1 (2.5 credits) draws the closest lettering on ordinary edits, but it can read a near-miss as if it were already spelled right and leave it alone: in six runs on the bakery poster it fixed “EVRY” every time and “OPENNING” never, and on the chalkboard it fixed “BLUBERRY” in three of six. Asked for the misspelled word on its own it did better: “BLUBERRY” in two runs of two, “OPENNING” in one of two. If Flash changes the lettering more than you like, try Nano Banana 2.1 with only the wrong word, or GPT Image 2. Q: Is it free? A: Sign in with Google (a verified Gmail account) and your first text edit, or one layer split, is free, with no card. After that each edit uses credits: 1 to 2.5 per image depending on the model, from a one-time pack starting at $9.90 for 100 credits. Q: Can I get the corrected text as an editable layer in Photoshop instead? A: Yes. Split the image into layers and export an editable text PSD: every line of regular-size text becomes real Photoshop type, with its outline, glow and shadow as layer styles, so you can retype it in Photoshop as often as you like. --- ## Canva to PSD with layers URL: https://layerpsd.com/canva-to-psd Canva cannot export a PSD: its download types are JPG, PNG, SVG, PDF, PPTX, DOCX, GIF and MP4. There are two ways to a layered PSD. If you can still open the design in Canva, download it as a PDF with Flatten PDF switched off, open the PDF in Photopea (free, in the browser), and save it as a PSD: each element becomes its own layer and the text stays live type. If all you have is the exported PNG or JPG, upload it to LayerPSD: every element is cut onto its own layer, the background is repainted behind them, and regular-size text comes back as editable Photoshop type. The first split is free when you sign in with a Gmail account. You still have the Canva design: PDF → Photopea: 1. In Canva, open Share → Download and choose PDF as the file type (Digital or Print). 2. Under the advanced settings, leave Flatten PDF off. Flattening merges the design into one image. 3. Open the PDF in Photopea (photopea.com, File → Open). Each page becomes an artboard, each element a layer. 4. Check the text layers, then File → Save as PSD. Text arrives as live type layers with its font name, size and colour; photos arrive as Smart Objects and vector shapes as editable shapes. Install the fonts the design uses, or the text shows in a substitute font. Photos stay whole: a person on a background is still one picture. You only have the PNG or JPG: LayerPSD: 1. Download the design from Canva as PNG (or use the image you were sent). 2. Upload it above and press Split into layers. 3. Each element lands on its own transparent layer and the background is repainted behind them. 4. Export "PSD with editable text" to get regular-size text as Photoshop type, or a ZIP of PNGs. This also works when the Canva file is gone, belongs to someone else, or was flattened, and it can split the photos inside a design into parts. Text is rebuilt rather than recovered: lines the model reads become type matched by style, the rest stays as pixels. Opening the Canva PDF straight in Photoshop does not give you layers: its Import PDF dialog turns each page into pixels, and File → Place brings the page in as one Smart Object. Q: Does Canva let you download a PSD? A: No. Canva's download types are JPG, PNG, SVG, PDF, PPTX, DOCX, GIF and MP4 (checked 2026-10-05). To get a PSD, download a PDF and convert it in Photopea, or download a PNG and split it into layers with LayerPSD. Q: How can I convert a Canva design to a PSD for free? A: Download it from Canva as a PDF with Flatten PDF off, open the PDF in Photopea, and choose File → Save as PSD. Photopea turns each page into an artboard, each element into a layer and the text into type layers. It is free and runs in the browser. Q: Why does my Canva PDF open as one flat layer? A: Either Flatten PDF was on when you downloaded it, which merges the design into a single image, or you opened it in Photoshop, whose Import PDF dialog rasterizes each page. Download again with Flatten PDF off and open the PDF in Photopea instead. Q: Can I open a Canva design in Photoshop with layers? A: Not directly: Photoshop rasterizes a PDF page when it opens it, and File → Place brings the page in as one Smart Object. Convert the PDF to a PSD in Photopea first and open that PSD in Photoshop, or split a PNG of the design with LayerPSD and export a PSD. Q: Will my Canva fonts stay editable? A: Through the PDF route the text stays live type and keeps its font name, so it is editable as long as the font is installed where you open it; otherwise it shows in a substitute. Through LayerPSD the text is read from the picture and rebuilt as Photoshop type in a matching style, so you do not need the font, but it is not the original typeface. Q: What if I only have the PNG, not the Canva file? A: Upload the PNG or JPG to LayerPSD. Every element is cut onto its own transparent layer, the background is repainted behind them, and regular-size text becomes editable Photoshop type when you export "PSD with editable text". Your first split is free when you sign in with a Gmail account. Q: Can Canva Magic Layers export a PSD? A: No. Magic Layers splits a flat image into layers inside a Canva design, and those layers stay in Canva because PSD is not a Canva download type. Our Canva Magic Layers alternative page compares the two. Sources: - Canva Help: download file types: https://www.canva.com/help/download-file-types/ (checked 2026-10-05) - Canva Help: download designs as flattened PDF: https://www.canva.com/help/download-flattened-pdf-variantb/ (checked 2026-10-14) - Photopea: convert PDF to layered PSD: https://www.photopea.com/tuts/convert-pdf-to-layered-psd/ (checked 2026-10-14) - Adobe Help: open PDF files in Photoshop: https://helpx.adobe.com/photoshop/desktop/create-open-import-images/import-files/open-pdf-files-in-photoshop.html (checked 2026-10-14) --- ## Image to editable PPT URL: https://layerpsd.com/image-to-ppt Upload an image above, press Split into layers, then choose Download PPTX. The text is read into editable text boxes in a common Office font of the same style; icons, photos and chart parts become movable pictures at their original positions; the background is repainted behind them. One image makes one slide, and charts stay pictures. The first split is free when you sign in with a Gmail account (1K, background + 4 layers). Steps: 1. Take a screenshot of the slide or save it as PNG or JPG (for a PDF, export the page as an image first). 2. Upload it above, or paste it with Ctrl+V / ⌘V, and press Split into layers. The text is read first, then the rest of the image is split. 3. When the layers are ready, press Download PPTX. 4. Open the .pptx in PowerPoint, Keynote or Google Slides and edit the text boxes, or move and delete the pictures. What comes out editable: - Titles, bullets, labels, footers: Text boxes you can retype, in a common Office font of the same style, with the colour, size and alignment read from the image - Very large or decorative lettering: Usually a picture, like a logo; regular-size text is what gets read - Icons, logos, shapes, photos: Separate pictures with transparent backgrounds, at their original size and position - Charts and tables: Pictures (in the sample each bar is its own picture); labels may be read as text, but there is no chart data behind them - Background: One full-slide picture with the area behind every element repainted - Several slides, animations, speaker notes: Not included: one image makes one slide Limits to know before you present: - The text is read by a model: a word, a punctuation mark, a size or a colour can come out wrong, and a line can stay inside a picture. Read the slide over before you present it. - The original typeface cannot be recovered from pixels; the text is set in a common Office font of the same kind. - Charts and tables come through as pictures, with no chart data behind them. - Elements that touch can stay in one picture, and the repainted background can show gaps or changed details. - In right-to-left languages, lines that mix words with digits and percent signs can change order. Check them in the app you present with. - One image makes one slide: no PDFs, several slides, animations or speaker notes. Inspect real image-to-PPT results Three AI-generated test slides with made-up business data, converted on 2026-10-14 (the Japanese and Arabic files were exported again on 2026-10-18 with the current exporter). Each comes with its input image, the PPTX and what went wrong in it. - English slide with a chart: 13 text boxes, 9 foreground pictures, 1 background. Export checked 2026-10-18. The text, icons, chart parts and photo show how a flat slide is represented as PowerPoint objects. The chart bars are pictures, with no editable chart data. Four axis numbers stayed in the chart picture, the reader added a full stop after “Acme Corp”, and some grid lines have gaps. PPTX: https://layerpsd.com/image-to-ppt/q3-growth-slide.pptx Input: https://layerpsd.com/image-to-ppt/slide-source.png - Japanese text and graphics: 13 text boxes, 7 foreground pictures, 1 background. Export checked 2026-10-18. Japanese titles and labels as text boxes; the donut chart and the photo as pictures. Preview rendered in macOS Quick Look. The text is set in a substitute font, so line breaks and spacing differ from the source; on a device without Yu Gothic another Japanese font is used. PPTX: https://layerpsd.com/image-to-ppt/japanese/sample.pptx Input: https://layerpsd.com/image-to-ppt/japanese/source.png - Arabic text, right to left: 10 text boxes, 6 foreground pictures, 1 background. Export checked 2026-10-18. Six Arabic text boxes set right to left; the numbers-only boxes stay left to right. Preview rendered in macOS Quick Look. In the preview, lines that mix Arabic with digits and percent signs change order; the chart lost or repeated some axis labels and parts of its grid; small labels stayed inside the chart picture. The original font is not recovered. PPTX: https://layerpsd.com/image-to-ppt/arabic/sample.pptx Input: https://layerpsd.com/image-to-ppt/arabic/source.png Q: How do I convert an image to an editable PPT? A: Upload the image to LayerPSD, press Split into layers, then Download PPTX. The text becomes text boxes, the other elements separate movable pictures, and the background is repainted behind them. Open the file in PowerPoint, Keynote or Google Slides to edit it. Q: Can I turn a screenshot of a slide into an editable PowerPoint? A: Yes. A screenshot is the case this was built for: titles, bullets and labels are read into text boxes, and icons, photos and chart parts are cut out as pictures in the same places. Use the sharpest screenshot you can; small, blurry text is harder to read. Q: Will the text be editable, and in which font? A: Yes, each line that is read becomes a text box you can retype. The original typeface cannot be recovered from pixels, so the text is set in a common Office font of the same kind (Arial, Times New Roman; Microsoft YaHei, Yu Gothic and Malgun Gothic for Chinese, Japanese and Korean), with the colour, size and alignment read from the image. You can change the font afterwards. Q: Are charts and tables editable after the conversion? A: No. Charts and tables come through as pictures, often one per part (in our sample, each bar was its own picture), and their labels may be read as text, but there is no chart data behind them. To change the numbers, rebuild the chart in PowerPoint. Q: Can I convert a PDF or several slides at once? A: Not yet: one image makes one slide. Export the PDF page you need as a PNG (in Preview on a Mac, File → Export), split it, and repeat for each slide. Q: How do I make NotebookLM slides editable? A: NotebookLM's help says a Slide Deck can be downloaded as a PDF or as a PowerPoint (.pptx) file from its three-dot menu. Try the PowerPoint download first. If a slide still arrives as one picture, or all you have is the PDF or a screenshot, save that slide as an image and convert it here. Q: Does the PPTX work in Google Slides and Keynote? A: It is a standard .pptx file, the format PowerPoint, Google Slides (File → Import slides) and Keynote open. We opened the English sample in Keynote, retyped two of its text boxes and moved the photo (see the edited slide on this page), and rendered the Japanese and Arabic samples in macOS Quick Look; we have not checked them in PowerPoint itself. Q: How much does it cost? A: Your first split is free when you sign in with a Gmail account (other Google accounts cannot get the free try). It runs at 1K with background + 4 layers and a lighter text reader, so on a busy slide fewer elements come out as separate pictures and more lines can stay inside them. After that a split on Seedream 5.0 Flash costs 1 credit per output layer. Downloading the PowerPoint file adds nothing, or about a tenth of a credit when some layers still need their text read. Sources: - NotebookLM Help: generate a Slide Deck (download as PDF or PowerPoint): https://support.google.com/notebooklm/answer/16757456 (checked 2026-10-14) --- ## Pricing URL: https://layerpsd.com/pricing Buy a credit pack once and spend it as you go: one layer on Seedream 5.0 Flash is 1 credit, from $0.071 a credit. A text edit is one image. Sign in with a Gmail account for one free try first. Q: What counts as one output? A: Every image the model returns: the filled background is one, and each layer is one. A split into background + 4 layers is 5 outputs. Q: How do credits work? A: One credit is one output layer on Seedream 5.0 Flash. Other models use more per image: about 6.7 credits on Seedream 5.0 Pro and 2.5 on Nano Banana 2.1. Your balance is in the header, and a run uses credits only for the images it returns. Q: Do credits expire? A: Credits stay valid for 12 months after your latest purchase. Buying another pack extends your whole balance. Q: Can I get a refund? A: If you have not used any of a pack, email hi@layerpsd.com within 3 days (72 hours) of buying it and we refund it in full; an accidental duplicate purchase is refunded too. Credits already spent are not refundable. A chargeback suspends the account and its remaining credits. Q: What does the free try include? A: One split on Seedream 5.0 Flash at 1K into background + 4 layers (with PSD and ZIP export), or one text edit. It needs a verified @gmail.com account, one per Google account. Q: Does exporting cost anything? A: PSD, ZIP and PNG export run in your browser and are free. “PSD with editable text” reads the text with a small vision model, about a tenth of a credit per export.