PDG is a digital book format used by Chaoxing. In normal use, it is opened with the official Chaoxing reader, commonly known as SSReader. If the reader is installed, opening a PDG file is usually as simple as double-clicking it.
The difficulty is that “PDG” does not refer to one single, uniform kind of file. Over time, several types appeared, and they differ greatly in clarity, structure, file naming, searchability, and what happens after decryption or conversion.
A rough classification looks like this:
- Image-based PDG
- File names usually have six characters and no underscore.
- It is essentially made from scanned page images.
- It includes several common forms:
- Clear version: generally the preferred image version; often associated with formats such as
02Hand04H; after decryption it is commonly CCITT or JPG. - Fast version: mainly created to reduce load and speed up delivery; now generally regarded as something to avoid; often associated with
05H,AxH, or6xH; after decryption it is commonly DjVu. - Large-image version: usually better than the fast version; commonly not encrypted; often consists of renamed JPG or PNG files, though the original SSReader does not recognize PNG.
- Clear version: generally the preferred image version; often associated with formats such as
- Text-based PDG
- File names contain an underscore and are not fixed in length.
- Internally it is closer to a compressed and packaged PDF made from text, or from mixed text and images.
- It can usually be searched, selected, and copied in SSReader.
- CX PDF
- One book is stored as one file.
- Different historical periods used different encryption methods and different levels of clarity.
- After decryption, it is generally a PDF with mixed text and images.
For anyone new to the format, these distinctions can be confusing at first. The most important line to draw is between image-based PDG and text-based PDG.
Image-based PDG: faithful to the original book, but not convenient for text work
In the early days of PDG, books were produced from scanned page images. Although the files used the PDG extension and were opened in SSReader, their nature was still image-based. When reading this kind of PDG in the official reader, the text cannot be searched, selected, or copied. If text is needed, OCR must be performed separately.
This type is usually called image-based PDG.
Its main advantage is fidelity. Since each page comes from a scan of the original book, the layout, typeface, illustrations, page structure, and other visual details are preserved. For citation, proofreading, or checking the original appearance of a book, this is often more reliable.
The disadvantage is also obvious: image files are large. They increase pressure on servers and network bandwidth, and they are inconvenient when readers need searchable or copyable text. This practical problem led to the later appearance of text-based PDG.
Text-based PDG: smaller and searchable, but not always trustworthy
To reduce file size and improve usability, scanned images were later processed with OCR. The recognized text was then compressed, packaged, encrypted, and stored in PDG form. Internally, this kind of PDG is closer to a pure-text PDF or a PDF with mixed text and images.
When opened in SSReader, text-based PDG can usually be searched directly. Text can also be selected and copied. When zoomed in, the characters do not show the jagged edges typical of enlarged page images. This is why it is commonly called text-based PDG, in contrast to the earlier image-based PDG.
The file name is often the quickest clue:
- An original text-based PDG file name contains an underscore.
- An image-based PDG file name has no underscore and is usually six characters long.
However, OCR is never perfect. No OCR engine can realistically guarantee 100% accuracy, and it is hard to imagine any serious software company making such a promise. Because of this, text-based PDG has traditionally been considered less dependable than image-based PDG. OCR errors may introduce wrong characters, which can affect reading, quotation, and research use.
This is also why some readers specifically avoid text-based PDG when looking for books.
In practice, the quality varies. Early pure-text PDG files did have more obvious problems. Later mixed text-and-image versions improved the situation somewhat: characters that OCR could recognize confidently were stored as text, while uncertain parts were left as images. This does not eliminate errors, but it is generally better than forcing the whole page into text.
There is one annoying side effect. In some text-based PDG files, the image parts were not set as transparent during production. If a custom background is enabled in the reader, the page may look messy: parts of the page appear as little white-background image blocks mixed into the text. One solution is to make those images transparent manually. Another is to read the file with UnicornViewer and force the background to be transparent. The limitation is that the free version of UnicornViewer does not support text-based PDG or PDF.
There are also rare high-quality text-based PDG files generated directly from original publishing layout files rather than from OCR. These do not suffer from OCR mistakes, except for errors already present in the book itself, and they can preserve the original layout very well. Such files are worth keeping, but whether one encounters them is largely a matter of luck.
Common signs of a text-based PDG
Text-based PDG files often show several recognizable traits.
First, the file name contains an underscore. This is the most basic external sign. Image-based PDG file names do not contain underscores.
Second, preface pages often appear at the end after processing. A likely reason is the PDG naming convention: preface pages begin with fow, so during OCR processing they are sorted after body pages whose names begin with numbers. This small detail says a lot about the care, or lack of care, sometimes involved in official production.
Third, after decrypting and decompressing a text-based PDG into PDF, the main text may appear in a boldface-style font rather than the expected Song-style font when opened in Acrobat. A practical fix is to replace the font parameters throughout the PDF with a tool such as PdfToy.
Fourth, the reading experience may feel strange even when there are not many obvious wrong characters. Some characters may look rough or slightly inconsistent. If the reader background is changed, those rough-looking parts often reveal themselves as small images with white backgrounds embedded among the text.
Fifth, the total size of the original text-based PDG files can be quite large, but after conversion and merging into a single PDF, the resulting PDF may be much smaller—provided the merging software can remove redundant objects. Acrobat, for example, can do this.
The reason is simple. To ensure each individual text-based PDG file can be opened independently, CX had to embed fonts inside the PDF contained in every PDG page file. But within a single book, the body pages usually use the same fonts. This means the original set of text-based PDG files contains a large amount of repeated font data. When Acrobat merges them, identical fonts are removed automatically, reducing the final file size. If a merged PDF remains almost the same size as the total original text-based PDG files, it is worth checking whether the wrong merging tool was used.
Clear version, fast version, and why this distinction belongs to image-based PDG
Text-based PDG is smaller and more convenient, but OCR takes time and effort. It was never realistic to convert every book into text-based PDG. To reduce server and network pressure while still serving scanned books, image-based PDG developed several variants, most notably the clear version and the fast version.
This distinction applies to image-based PDG only. A text-based PDG is not normally described as a clear version or fast version.
The clear version is generally the one to prefer among image-based PDG files. It is closer to the goal of preserving readable page images and is often associated with 02H or 04H. After decryption, it commonly becomes CCITT or JPG image data.
The fast version was mainly designed for speed and reduced load. It is often associated with 05H, AxH, or 6xH, and after decryption it is commonly DjVu. In terms of preservation and reading quality, it is usually not the preferred target.
The large-image version is usually stronger than the fast version. It may be unencrypted and may consist of JPG or PNG files with changed names. One practical caveat is that the original SSReader does not recognize PNG in this context.
CX PDF is a separate branch: one book, one file. Depending on the period in which it was produced, it may use different encryption methods and may vary in clarity. After decryption, it is typically a PDF combining images and text.
Once these categories are clear, many PDG quirks become easier to understand: why some files can be searched while others cannot, why some file names contain underscores, why converted PDFs sometimes shrink dramatically, and why some “text” pages still contain small image fragments.