DocRes-dewarp
DocRes-dewarp flattens curved or skewed document photos.
A tradução deste card ainda não está disponível; parte do conteúdo aparece em inglês.
DocRes-dewarp is an image model specialized in document rectification. It takes a photograph of a document that is curved, folded, or taken at an angle and produces a straight, flat version with corrected perspective. The model handles common distortions like page curl, fold lines, and mild warping from binding.
Use it when you need to digitize physical documents quickly without a scanner — for example, snapping a page from a book, flattening a receipt, or fixing a crooked photo of a whiteboard. It excels at preserving text legibility and layout structure.
What makes DocRes-dewarp distinctive is its focus on natural-looking results: it avoids stretching text unevenly or introducing artifacts. It works across various paper textures and lighting conditions, making it a reliable tool for document cleanup in creative and productivity workflows.
- Flattening curved book pages for reading
- Straightening skewed receipt or ticket photos
- Correcting fold distortion in scanned documents
- Improving whiteboard or poster captures
- Preparing document images for OCR
Produces a clean, flat document image without unnatural stretching or blurring.
Corrects page curl from thick books or spiral notebooks reliably.
Maintains sharp, readable text even on glossy paper or with shadows.
- Use a flat, well-lit photo of the document for best results.
- Crop the image to focus on the document area before processing.
- Avoid taking photos at very low angles for more reliable corrections.
- Combine with OCR models for end-to-end text extraction from physical pages.
- May struggle with extreme folds that create deep creases.
- Works best on single pages; double-page spreads may show boundary artifacts.
- Cannot correct text that is severely blurred or out of focus in the original photo.
Restore a warped page from an antique book to a flat, scannable image.
Dewarp a crinkled ticket photo so the text and barcode are clear.
Correct the curve in a photo of a spiral notebook page.
Fixes the random starting point for reproducible results.
Controls how strongly the dewarping is applied. Lower values preserve original curves; higher values fully flatten.
Determines the output image dimensions. Larger sizes keep more detail.
Sets the file type of the resulting image (e.g., JPEG or PNG).
Os preços exibidos são o que você paga na Muitto. Sem surpresas.
Does this model preserve the original image resolution?
Yes, the output retains the input resolution while correcting distortion.
Can it handle documents with handwriting?
Yes, it works on any written content as long as the text is legible.
Will it work on photos of screens or monitors?
It is designed for paper documents and may not perform well on screen captures.